PowerFlex 755T Flux Vector Tuning

Transcription

1 Application Technique Original Instructions PowerFlex 755T Flux Vector Tuning Catalog Number 20G; 20J

2 Important User Information Read this document and the documents listed in the additional resources section about installation, configuration, and operation of this equipment before you install, configure, operate, or maintain this product. Users are required to familiarize themselves with installation and wiring instructions in addition to requirements of all applicable codes, laws, and standards. Activities including installation, adjustments, putting into service, use, assembly, disassembly, and maintenance are required to be carried out by suitably trained personnel in accordance with applicable code of practice. If this equipment is used in a manner not specified by the manufacturer, the protection provided by the equipment may be impaired. In no event will Rockwell Automation, Inc. be responsible or liable for indirect or consequential damages resulting from the use or application of this equipment. The examples and diagrams in this manual are included solely for illustrative purposes. Because of the many variables and requirements associated with any particular installation, Rockwell Automation, Inc. cannot assume responsibility or liability for actual use based on the examples and diagrams. No patent liability is assumed by Rockwell Automation, Inc. with respect to use of information, circuits, equipment, or software described in this manual. Reproduction of the contents of this manual, in whole or in part, without written permission of Rockwell Automation, Inc., is prohibited Throughout this manual, when necessary, we use notes to make you aware of safety considerations. WARNING: Identifies information about practices or circumstances that can cause an explosion in a hazardous environment, which may lead to personal injury or death, property damage, or economic loss. ATTENTION: Identifies information about practices or circumstances that can lead to personal injury or death, property damage, or economic loss. Attentions help you identify a hazard, avoid a hazard, and recognize the consequence. IMPORTANT Identifies information that is critical for successful application and understanding of the product. Labels may also be on or inside the equipment to provide specific precautions. SHOCK HAZARD: Labels may be on or inside the equipment, for example, a drive or motor, to alert people that dangerous voltage may be present. BURN HAZARD: Labels may be on or inside the equipment, for example, a drive or motor, to alert people that surfaces may reach dangerous temperatures. ARC FLASH HAZARD: Labels may be on or inside the equipment, for example, a motor control center, to alert people to potential Arc Flash. Arc Flash will cause severe injury or death. Wear proper Personal Protective Equipment (PPE). Follow ALL Regulatory requirements for safe work practices and for Personal Protective Equipment (PPE).

3 Table of Contents Preface Chapter Background Mechanical Loading Inertia Compliance Backlash Resonances Performance Bandwidth Damping Factor System Bandwidth System C/U Select Control Structure Overview Position Loop Velocity Loop Torque Loop Chapter 2 Product Features Torque Scaler Load Observer Benefits How It Works Configuration Adaptive Tuning Benefits How It Works Configuration Notch Filters Overview Configuration Low Pass Filters Lead Lag Filters Chapter 3 Out-of-Box Tuning Gain Calculation Recommended Default Settings Single Knob Tuning Is Further Tuning Required? Rockwell Automation Publication 750-AT006A-EN-P - June 207 3

4 Table of Contents Chapter 4 Auto Tuning General Modes Motor Electrical Parameters Configuration Modes Inertia Tests Gain Calculation Modes Configuration Is Further Tuning Required? Chapter 5 Manual Tuning Initial Configuration Tune the Current Loop (Optional) Tune the Velocity Loop Tune the Position Loop (Optional) Chapter 6 Applications Friction Compensation Types of Friction Configuration Torque Sharing Droop Vertical Loads Process PID Controller Modes Features Diagrams Rockwell Automation Publication 750-AT006A-EN-P - June 207

5 Preface This Publication This document is intended to assist PowerFlex 755T drive users in tuning Flux Vector position and velocity loops, filters, and other features to achieve high performance. The information given here is intended for drive users with skill levels ranging from novice to advanced. Each component of the control structure is described in detail and out-of-box tuning, autotuning, and manual tuning techniques are presented. Quick Start Guide See Recommended Default Settings on page 69. Additional Resources These documents contain additional information concerning related products from Rockwell Automation. Resource PowerFlex Drives with TotalFORCE Control Programming Manual, publication 750-PM00 PowerFlex Family Selection Guide, publication PFLEX-SG002 PowerFlex 750 Installation Manual, publication 750- IN00 PowerFlex 750 Reference Manual, publication 750- RM002 PowerFlex 700 to 750 Migration Guide, publication PFLEX-AP005 Wiring and Grounding Guide, publication DRIVES- IN00O PowerFlex for Crane and Hoist Applications, publication PFLEX-BR009 PowerFlex 700S and PowerFlex 755 Drives Tuning Manual, publication DRIVES-AT004 Drives Engineering Handbook, publication DEH Motion System Tuning Guide, publication MOTION- AT005 Kinetix 5700 Drive Systems Design Guide, publication GMC-RM00 Kinetix 5700 Servo Drives User Manual, publication 298-UM002 Description Provides detailed information on: I/O, control, and feedback options Parameters and programming Faults, alarms, and troubleshooting Determine which drive is right for your application Install, configure, and troubleshoot PowerFlex 750 series applications Determine what you need for PowerFlex 750 applications Migrate from PowerFlex 700 to PowerFlex 750 series Wiring guidelines for PWM drives Crane and hoist application notes Tune PowerFlex 700S/755 drives Formulas for drive systems and application examples Tune Kinetix 5x00/6x00 drives Determine what you need for Kinetix 5700 applications Install, configure, and troubleshoot Kinetix 5700 applications You can view or download publications at To order paper copies of technical documentation, contact your local Allen-Bradley distributor or Rockwell Automation sales representative. Rockwell Automation Publication 750-AT006A-EN-P - June 207 5

6 Preface Notes: 6 Rockwell Automation Publication 750-AT006A-EN-P - June 207

7 Chapter Background Topic Page Background 9 Mechanical Loading 9 Performance 9 Control Structure Overview 25 This document is intended to assist PowerFlex 755T drive users in tuning Flux Vector position and velocity loops, filters, and other features to achieve high performance. Here, the primary control mode is Induction FV. The information is intended for users who are familiar with the following: PowerFlex 755T drive connectivity and configuration Drive communication through a personal computer or Human Interface Module (HIM) Use of Connected Components Workbench (CCW) software Understanding how control loops work in drive applications Parameters are referenced as Port Number:Parameter Number [Parameter Name]. For example: 0:900 [Motor Inertia] denotes port 0:parameter number 900 named Motor Inertia. 0:65 [Pri MtrCtrl Mode] denotes port 0:parameter number 65 named Pri MtrCtrl Mode. Quick Start Guide See Recommended Default Settings on page 69. Mechanical Loading To effectively control the motion of a mechanical system and achieve high performance, first understand the mechanical properties for the system that dictate performance. This section describes these mechanical properties, how they limit performance, and what tuning features apply to compensate for them. Obviously, the best approach is to build a mechanical system that is as rigid as possible and with a constant inertia, but that is not always possible or cost-effective. Rockwell Automation Publication 750-AT006A-EN-P - June 207 7

8 Chapter Background Inertia With rotary motors and rotating machinery, inertia is rotating mass. An object with mass requires a certain amount of torque to accelerate and decelerate it at a specified rate. The larger an object is, the more torque is required. As a result, inertia affects how large of a drive and motor is required which ultimately affects drive tuning. A few things happen as inertia increases: More torque is required to accelerate the inertia to a given speed. Or it takes longer to accelerate the inertia to a given speed with a fixed torque. Adverse effects of non-rigid components (compliance and back lash) are amplified, making it more difficult to tune. A motor with a connected load is shown in the figure as a simple two mass system. Figure - Two Mass System Motor Coupling Load J M J L The following definitions are given: J M = 0:900 [Motor Inertia] J L = Load Inertia reflected back to the motor shaft R = 0:90 [Load Ratio] = J L /J M J T = Total Inertia = J M + J L = J M (R+) 0:900 [Motor Inertia] is used to calculate the Torque Scaler, an internal parameter that compensates for the effects of inertia. This parameter affects overall tuning. See Torque Scaler on page 35 for more information. Four ways to determine motor inertia are given in order from most to least effective. Data Sheet: Enter a value from the motor nameplate or data sheet in units of [kg*m 2 ]. Divide [lb.*ft 2 ] or [WK 2 ] by to convert to [kg*m 2 ]. Measured: Run Autotune with 0:90 [Autotune] = InertiaMotor (4) or JMtr BW Calc (7) with the load disconnected to dynamically measure motor inertia. See Auto Tuning on page 75 for more information. This method is only an option if the load can be disconnected to run the test. 8 Rockwell Automation Publication 750-AT006A-EN-P - June 207

9 Background Chapter Estimated: If data is not available or a motor inertia test is not possible, use the following equation to approximate 0:900 [Motor Inertia] based on motor nameplate horsepower (HP): J M = HP/250*(HP/500 +) Divide [kw] by.34 to convert to [HP]. Default Value: This value is based on a motor power rating equal to the drive power rating. Load Inertia that is reflected back to, and present at, the motor shaft includes coupling inertia and accounts for any gear ratio present in the coupling mechanics. It is measured in the same units as motor inertia. Coupling mechanics between the motor and load can consist of rotary couplings, gearboxes, belts and pulleys, shafts and gears, linkages, or any combination of them. Changing or unknown load inertias are best compensated for by enabling the Load Observer on page 39. 0:90 [Load Ratio] is the ratio of load inertia divided by motor inertia. It tells you how large the load is with respect to the motor driving it. Load ratio is used to calculate the Torque Scaler, which compensates for the effects of inertia and is described in Torque Scaler on page 35. This parameter affects overall tuning. Load ratio is also used in calculating controller gains. Three ways to determine load ratio are given in order from most to least effective. Default Value: A value of zero is used when the load is unknown or compliant. A load ratio R = 0 initiates out-of-box tuning gain calculations that work well for most applications. See Out-of-Box Tuning on page 65 for more information. Entered: Enter a value in [kg*m 2 ] from your own calculations that are based on knowledge of the mechanical design. For a changing inertia, enter the lowest value. A load ratio R > 0 initiates autotune gain calculations that work well on rigid loads. See Auto Tuning on page 75 for more information. Measured: Run Autotune with 0:90 [Autotune] = InertiaTotal (5) or JtotalBWCalc (8) with the load connected to dynamically measure total inertia and load ratio. This method is only an option if motion can be initiated to rotate the load during the test. Furthermore, this calculation assumes 0:900 [Motor Inertia] is accurate. If you get an error resulting from a negative load ratio, the motor inertia is incorrect and must be reduced. Where possible, perform the Autotune at the mechanical point of lowest inertia. A load ratio R > 0 initiates autotune gain calculations that work well on rigid loads. See Auto Tuning on page 75 for more information. Total Inertia is the sum of load inertia and motor inertia. It is the total inertia of the moving mechanical system that is driven by the drive. Total inertia determines the maximum acceleration and deceleration of the mechanical system. Rockwell Automation Publication 750-AT006A-EN-P - June 207 9

10 Chapter Background Compliance Compliance is the elastic deformation of mechanical components. In rotating machinery, it is a twisting of these components that stores and releases energy in the form of torque. More flexing or twisting means more energy storage. Compliance is common in couplings, shafts, linkages, and belts. The effect is modeled as a two mass system that is coupled by a spring. Figure 2 - Compliant Two Mass System T M V M J M s T R b + k/s V L J L s T D The following definitions are given. T M = Motor torque V M = Motor velocity, typically produced by encoder feedback T R = Reaction torque on the motor shaft from the load k = Spring constant b = Viscous friction V L = Load velocity T D = Load torque disturbance When motor torque is applied, the coupling can twist when transferring torque to the load. As the coupling twists, it stores energy and does not immediately rotate the load. The energy is then released when the motor torque is removed, which causes continued acceleration of the load. A dynamic model is given where PI is a proportional-integral controller and LPF is a low pass filter. Figure 3 - Compliant Load Model VCMD Power T + PI M + _ Converter _ s J M VM LPF Velocity Loop TR b k/s + _ T D + J L s VL 0 Rockwell Automation Publication 750-AT006A-EN-P - June 207

11 Background Chapter Here, the spring constant k is in units of [Nm/rad], which can be typically found in rotary coupling specifications. For example, a shaft with a spring constant of.0 deflects or twists by radian when a steady Nm torque is applied to one end while the other end is anchored to a stationary rigid body. The effect of compliance is frequency dependent, which creates tuning problems in the form of resonances. A compliant coupling creates a resonance and an anti-resonance in signals that are measured by the motor side encoder. Furthermore, another resonance is present at the end of the load or end effector. This load side resonance has a resonant frequency equal to the motor side anti-resonant frequency. Figure 4 - Compliant Load Resonances Motor Side Resonances Load Side Resonances V CMD PI Power Converter T M T E V M V L LPF Velocity Loop See Resonances on page 4 for information on the types of resonances, how they affect tuning, and how to control them. The resonant frequency F R and anti-resonant frequency F A can be calculated in units of [Hz], with an equivalent inertia J E that is used in the following calculations. F R = k J E 4 2 F A = k = J L 4 2 F R R + J L J M J E = J = L + J J R M R + M The severity of this tuning problem increases under the following conditions: Rockwell Automation Publication 750-AT006A-EN-P - June 207

12 Chapter Background As the spring constant decreases, the coupling becomes less rigid and twists more. As the load ratio increases, the load on the end of the spring becomes larger, which causes it to twist more. Larger load ratios amplify tuning problems. Compliance is also referred to as non-rigid. There are only two absolutes: When k = 0, the load is disconnected When k =, the load is rigid When 0 < k <, everything else is compliant and/or has backlash. This condition constitutes most industrial applications over a wide range of dynamic behaviors. 0:902 [Load Coupling] This parameter reflects the type of mechanical coupling between the motor and load. It adjusts calculated control loop gains during an autotune bandwidth calculation test. 0:902 [Load Coupling] = Rigid (0) is where the load consists of few mechanical components with direct connection to the motor shaft. The components are high performance, have no flex or twist, and misalignment is not likely to occur. This setting is used for high performance machines. 0:902 [Load Coupling] = Compliant () is a non-rigid load where position misalignment, backlash, and flexing of couplings, gearboxes, belts, and shafts can occur which creates an indirect connection of the load to the motor shaft. This setting is common for most machines. Backlash Backlash is the result of a small space between two mechanical components in a drive train, where they are not always in direct contact. Backlash is caused by small geometric variations, manufacturing tolerances, and misalignments that are common in couplings, gearboxes, linkages, and other mechanical components. In gearboxes, it is a small space between two inter-meshing gear teeth. Figure 5 - Backlash in Gears 2 Rockwell Automation Publication 750-AT006A-EN-P - June 207

13 Background Chapter When the gears are not in contact and motor torque is applied, the input gear momentarily moves without transferring torque to the load. Here, the total inertia driven by the motor is J M. When the gears come in contact, torque is then transferred to the load. Here, the total inertia driven by the motor is J T. As a result, the effect of backlash is an instantaneous position-dependent step in total inertia. Figure 6 - Backlash Effect Inertia J T J M Position Backlash Distance A similar problem exists in ballscrews where rotary motion is converted to linear motion. Ballscrews are often internally preloaded to reduce backlash at the expense of added friction. Another method of minimizing the effect of backlash in positioning applications is to always approach the target position from a forward direction. When moving in reverse direction, the target position is surpassed and then a direction reversal is done. The target position is then reached from the forward direction. This method cancels the effect of backlash distance and confirms that the final load side position is consistent, regardless of where the starting load side position was. This nonlinearity creates tuning problems in the form of resonances. A dynamic model is given. Figure 7 - Backlash Model V CMD Power T M + PI + _ Converter _ s J M V M LPF Velocity Loop T R s _ + T D + J L s V L Rockwell Automation Publication 750-AT006A-EN-P - June 207 3

14 Chapter Background The severity of this problem increases with the following: As the backlash distance increases, the resonant frequencies decrease. As the load ratio increases, the step in inertia becomes larger. Since the load is intermittently disconnected from the motor, this step is called the load disconnect ratio: J T = R + J M If the velocity regulator is tuned for peak performance, the velocity regulator response is under-damped at best and unstable at worst. Typically, the motor shaft oscillates wildly between the limits that are imposed by the backlash distance. The net effect is an audible buzzing sound when the motor is at rest, commonly referred to as chatter. Resonances occur at the fundamental frequency of this chatter and its harmonics. If this situation persists, the coupling or gearbox wears out prematurely. The following section describes the effects of compliance and backlash and how to control them. Resonances Resonance is the tendency for a system output to oscillate with greater magnitude at some frequencies than at the others. Resonant frequencies are frequencies where the magnitude is at a relative maximum. Here, the system is sensitive to input and the gain is very high. Resonances are difficult to control because even small input forces can produce large magnitude oscillations in the output. Figure 8 - Tuning Fork Resonance Anti-resonant frequencies are frequencies where the magnitude is at a relative minimum. Here, the system does not respond well to input and the gain is very low. Anti-resonances are difficult to control because even large input forces can produce little or no response in the output. Compliance and backlash create resonances. Increasing the load inertia makes these problems worse. As a result, mechanical loads exhibit resonances that limit performance, damage hardware, consume energy, and generate noise. It is often left to the user to suppress these resonances through manual tuning, a 4 Rockwell Automation Publication 750-AT006A-EN-P - June 207

15 Background Chapter challenging and time consuming task. Resonances typically increase in number and severity as controller gains are increased. They range in frequency from a few Hz to a few thousand Hz. A typical approach is to reduce regulator gains to a low enough level so that the system bandwidth falls below the lowest resonant frequency. However, this approach results in limited performance. There are two main types of resonances: motor side resonances and load side resonances. Motor side resonances are further broken down into three subtypes based on their frequency range: Low Frequency (LF) resonances, Mid Frequency (MF) resonances, and High Frequency (HF) resonances. Figure 9 - Types of Resonances LF Resonances are in-band HF Resonances are out-of-band Load Resonances from end effector vibration Motor Side Resonancess Load Side Resonances MF Resonances are near CL BW V CMD PI Power Converter T M T E V M V L LPF Velocity Loop Each type is handled uniquely during tuning and is described in the following sections. Motor Side Resonances Motor side resonances are resonances that are observable in the feedback signal and are detected by the feedback device on the motor. As a result, they are suppressed by tuning control loop gains, load observer gains, and torque loop filter parameters. Most mechanical resonances are motor side and most drive applications exhibit them. Furthermore, all applications ultimately exhibit resonance if gains are increased enough. Motor side resonances are categorized into three types, based on their frequency range. Rockwell Automation Publication 750-AT006A-EN-P - June 207 5

16 Chapter Background Low Frequency (LF) Resonances LF resonances have resonant frequencies below the Torque Loop Bandwidth. Since they are within the closed loop bandwidth (in-band), they are automatically suppressed when the load observer is applied with the recommended out-of-box settings. See Load Observer on page 39 for more information. Otherwise, they can cause classical instability that generates an audible lowpitch growling noise. This instability requires detuning the control loop gains. High Frequency (HF) Resonances HF resonances have resonant frequencies above the Torque Loop Bandwidth. They typically generate an audible high-pitch squealing noise. Since HF resonances are outside the closed loop bandwidth (out-of-band), they are suppressed using torque loop filters. Adaptive tuning addresses these resonances by automatically configuring these filters. See Adaptive Tuning on page 44 for more information. Otherwise, it is left to the user to identify the audible frequencies and manually configure torque notch filters. If there are more HF resonances than there are notch filters available, a torque low pass filter is applied to suppress resonances with the highest resonant frequencies. A last resort is detuning control loop gains and load observer gains until the resonances vanish. Mid Frequency (MF) Resonances MF resonances are resonant frequencies near the Torque Loop Bandwidth. Here, torque loop filters are applied at frequencies close to the closed loop bandwidth. This close proximity allows phase lag generated by these filters to interfere with the closed loop dynamics and cause instability. Adaptive tuning addresses the issue by automatically narrowing torque notch filter widths as resonances approach the closed loop bandwidth. If instability occurs, adaptive tuning Gain Stabilization automatically decreases control loop gains and load observer gains to restore stability. See Adaptive Tuning on page 44 for more information. Otherwise, it is left to the user to manually reduce notch filter widths, decrease control loop gains, and decrease load observer gains. Similarly, torque loop low pass filters impa ct stability when they are applied at frequencies as low as times the closed loop bandwidth because they generate more phase lag than notch filters. As a result, only apply low pass filters if you run out of notch filters, and only reserve them for resonances with the highest resonant frequencies. Load Side Resonances Even when all motor side resonances are suppressed and the motor shaft is tightly controlled using closed loop feedback, the load end effector can still oscillate at a few Hertz through a compliant mechanical connection or linkage to the motor. These oscillations are load side resonances that are unobservable in the feedback signal and are not measurable by the feedback device on the motor. Load side resonances result in end effector vibration that is common in robots, cranes, cantilevered loads, anti-sway, liquid sloshing, laser cutting, and material handling applications. 6 Rockwell Automation Publication 750-AT006A-EN-P - June 207

17 Background Chapter Load side resonance suppression requires one of the following techniques, or a combination of them: Determine the load oscillation frequency with a stopwatch and apply a reference notch filter at that frequency. Then select a smooth reference move profile. This technique is recommended, and is the simplest to implement. Determine the load oscillation frequency with a stopwatch and generate a smooth reference CAM profile that does not contain any load oscillation frequency content. Place a feedback device on the load and set the drive to load side feedback mode See Position Loop on page 24 or Velocity Loop on page 26 for more information on reference notch filters. The following figure shows which features are used to control various resonances. Figure 0 - Controlling Resonance Position Command s Feed Forwards Kvff s Kaff Adaptive Tuning Torque Loop Filters System Under Control P REF RN PI PI K J LL LP N K T 2 J T s P REG Velocity Feedback Filter Fs V REG Velocity Estimate Torque Estimate Load Observer Adaptive Tuning compensates MF Resonances HF Resonances Load Torque Position Feedback Reference Notch Filters compensate Load Resonances Load Observer compensates Inertia LF Resonances Performance The following sections describe the metrics that are used to adjust performance during tuning. Bandwidth Bandwidth (BW) is a widely used term that indicates performance. It is defined as the usable range of frequencies where the gain through a system is above -3 db. Rockwell Automation Publication 750-AT006A-EN-P - June 207 7

18 Chapter Background Figure - System Gain Versus Frequency Bandwidth directly equates to transient response and how fast a system physically responds under a load to an input. There must be a way to quantify and compare drive performance across various drive instances and the metric that makes the most sense is bandwidth, which is measured in units of [Hz]. Position, velocity, and torque loop bandwidth indicate the respective performance of each regulator in the drive. Higher bandwidth improves transient response, decreases error, and makes the motor shaft performance stiffer. The following figure shows how bandwidth affects actual response (solid) compared to its reference move profile (dashed). Here, feed forward gains are disabled. Figure 2 - How Bandwidth Affects Transient Response Higher Bandwidth Lower Bandwidth The drive, motor, and feedback device have a significant impact on the bandwidth that can be achieved on a system through tuning. Closed loop bandwidth is affected by these factors: Load Ratio (lower is better) Compliance and backlash (rigid coupling is better) Feedback resolution (higher is better) Drive loop update rate (higher is better) Drive Model Time Constant (smaller is better) See System Bandwidth on page 9 for more information on drive model time constant. 8 Rockwell Automation Publication 750-AT006A-EN-P - June 207

19 Background Chapter Damping Factor The damping factor is commonly referred to as zeta (Z). It affects the rise time for a given bandwidth. The following figure shows how the damping factor affects actual response (solid) compared to its reference move profile (dashed). Here, feed forward gains are disabled. Figure 3 - How Damping Affects Transient Response Over Damped Under Damped High: Z < Critically Damped Medium: Z = Low: Z > The examples shown in the previous figure are discussed: A damping factor of Z < produces high responsiveness, which is characterized by a faster rise time with overshoot. A damping factor of Z = produces medium responsiveness, which is characterized by the fastest possible rise time without overshoot. This value is the default and recommended setting. A damping factor of Z > produces low responsiveness, which is characterized by a slower response without overshoot, similar to decreasing the bandwidth. 0:907 [System Damping] adjusts the position, velocity, and torque loop spacing of the calculated control loop gains and load observer gains. It also adjusts the integrator spacing of calculated gain parameters to generate the required responsiveness. A lower damping factor decreases the spacing between the position, velocity, and torque loop bandwidths. It also increases integrator gains. This value generates under-damped responses in the position and velocity response. A higher damping factor generates over-damped responses. System Bandwidth System bandwidth is calculated from an internal quantity that is called the drive model time constant (DMTC). The DMTC is the sum of all delays around the torque loop for a given drive and motor combination. Rockwell Automation Publication 750-AT006A-EN-P - June 207 9

20 Chapter Background Figure 4 - Delays Associated with DMTC Command PI Regulator Computational Delay Current Loop Time Constant Motor Electrical Time Constant Actual Feedback Filter Time Constant Feedback Sample Delay The following parameters are used to calculate the DMTC. 0:445 [VCL CReg BW] This parameter is the current regulator bandwidth 0:407 [Motor Poles] This parameter is the number of motor poles (p). It is calculated as follows p = round (20 x [Motor NP Hertz] / [Motor NP RPM]) Primary Encoder Resolution This parameter is the total resolution in edge counts per revolution (EPR) specified by parameters on the primary feedback option card. Low-Resolution Example: Resolution = 024 pulses per revolution * 4 quadrature edge counts per pulse = 4096 EPR (2-bit). The lowresolution PPR comes directly from a parameter on the option card. When both A and B channels are selected for an incremental encoder, the edge count multiplier is 4. This value is the typical and default setting. When only channel A is selected, then the edge count multiplier is 2. High-Resolution Example: Resolution = 024 pulses per revolution * 024 edge counts per pulse =,048,576 EPR (20-bit). For highresolution devices, the overall resolution choices are 20-bit default or an optional 24-bit when the corresponding configuration bit is selected. Typical values for current regulator bandwidth are given for various drive carrier frequencies. Table - Current Regulator Bandwidths 0:445 [VCL CReg BW] 0:425 [PWM Frequency] 375 Hz 4 khz 250 Hz 2 khz 25 Hz.33 khz The bandwidth that is associated with the DMTC is the torque loop bandwidth. TorqueLoopBandwith Tbw = H z 2 DMTC 20 Rockwell Automation Publication 750-AT006A-EN-P - June 207

21 Background Chapter 0:906 [System BW] System bandwidth is calculated using the torque loop bandwidth and the following parameters. 0:90 [Load Ratio] 0:902 [Load Coupling] 0:907 [System Damping] 0:000 [Pri Vel Fb Sel] Enter the 2-digit port location followed by the 4-digit parameter number of the primary feedback device. System bandwidth is used in the calculated control loop parameters. It is also used as a tuning knob. See Single Knob Tuning on page 72 for more information. 0:00 [Alt Fb GnScale] Similar to how system bandwidth adjusts multiple calculated control loop parameters, this parameter internally scales control loop gains and filter bandwidths during alternate feedback switchover. The source of feedback is automatically switched from the primary channel to the alternate channel when an Automatic Tach Switchover has occurred. This parameter adjusts tuning to account for different encoders on the primary and alternate feedback channels. The the following controller gains are internally scaled. Load Observer Bandwidth (Kop) Load Observer Integrator Bandwidth (Koi) Velocity Loop Bandwidth (Kvp) Velocity Loop Integrator Bandwidth (Kvi) The following diagram shows how system bandwidth and alternate feedback gain scaler are calculated out of the box and during an autotune bandwidth calculation. Figure 5 - System Bandwidth Calculation Rockwell Automation Publication 750-AT006A-EN-P - June 207 2

22 Chapter Background System C/U Select 0:905 [System C/U Select] allows either calculated or user entered parameters to be applied to controller gains and filter bandwidths. It also allows you to transfer calculated parameter values to user entered parameter values so you can use them as a starting point for manual tuning if desired. The following parameters are applied to the controller when 0:905[System C/U Select] = Calculated. Kpp = 0:754 [c PReg Kp] = Position Loop Bandwidth [Hz] Kpi = 0:756 [c PReg Ki] = Position Integrator Bandwidth [Hz] Kvp = 0:955 [c VReg Kp] = Velocity Loop Bandwidth [Hz] Kvi = 0:957 [c VReg Ki] = Velocity Integrator Bandwidth [Hz] Kop = 0:202 [c LdObs Kp] = Load Observer Bandwidth [Hz] Koi = 0:2023 [c LdObs Ki] = Load Observer Integrator Bandwidth [Hz] Kaff = 0:973 [c Accel FF Gain] = Acceleration Feed Forward gain [%] 0:254 [c Trq LPF BW] = Torque Low Pass Filter Bandwidth[Hz] 0:002 [c VelFb LPF BW] = Velocity Feedback Low Pass Filter Bandwidth[Hz] 0:008 [c AltVelFbLPF BW] = Alternate Velocity Feedback Low Pass Filter Bandwidth[Hz] 0:928 [c Vel Comp Gain] = Velocity Compensation Gain These parameters are auto-calculated as a function of 0:906 [System BW], 0:907 [System Damping], 0:90 [Load Ratio], and 0:902 [Load Coupling] during out-of-box and autotune calculations. For single parameter tuning, you can also manually increase or decrease system bandwidth to automatically adjust these parameters. The following parameters are applied to the controller when 0:905 [System C/U Select] = User Entered. Kpp = 0:755 [u PReg Kp] = Position Loop Bandwidth [Hz] Kpi = 0:757 [u PReg Ki] = Position Integrator Bandwidth [Hz] Kvp = 0:956 [u VReg Kp] = Velocity Loop Bandwidth [Hz] Kvi = 0:958 [u VReg Ki] = Velocity Integrator Bandwidth [Hz] Kop = 0:2022 [u LdObs Kp] = Load Observer Bandwidth [Hz] Koi = 0:2024 [u LdObs Ki] = Load Observer Integrator Bandwidth [Hz] Kaff = 0:974 [u Accel FF Gain] = Acceleration Feed Forward gain [%] 0:255 [u Trq LPF BW] = Torque Low Pass Filter Bandwidth[Hz] 22 Rockwell Automation Publication 750-AT006A-EN-P - June 207

23 Background Chapter 0:003 [u VelFb LPF BW] = Velocity Feedback Low Pass Filter Bandwidth[Hz] 0:009 [u AltVelFbLPF BW] = Alternate Velocity Feedback Low Pass Filter Bandwidth[Hz] 0:929 [u Vel Comp Gain] = Velocity Compensation Gain This selection allows you to enter values for these parameters directly. The following diagram shows how various parameters impact the System C/U Selection. Autotune updates the parameters in red, which then triggers the automatic calculations upon any change of their input parameters. Figure 6 - Control Loop Gain Selection Control Structure Overview This section provides an overview of the control structure and the location of various product features relative to one another. The following chapter goes into detail about each feature. Flux Vector mode is selected by 0:65 [Pri MtrCtrl Mode] = Induction FV (3). It implements a position loop, that is wrapped around a velocity loop, that is wrapped around a torque loop. Rockwell Automation Publication 750-AT006A-EN-P - June

24 Chapter Background Figure 7 - Flux Vector Control Structure Position Command s Feed Forwards s Adaptive Tuning Kvff Kaff Torque Loop Filters System Under Control P REF RN PI PI K J LL LP N K T J T s 2 P REG V REG Load Torque Torque Estimate Velocity Feedback Filter Fs Velocity Estimate Load Observer Position Feedback This type of control structure has the following advantages: Precise control of position, velocity, and torque Flexibility to switch between position, velocity, and torque modes without changing tuning gains Simple Inside-Out tuning Flux Vector mode can operate in these configurations: Torque control: Only the torque loop is enabled Velocity control: Only velocity and torque loops are enabled Position control: Position, velocity, and torque loops are all enabled Each loop and their internal functions are described in the following sections. Position Loop The position loop consists of position reference, reference notch filters, and position regulator sections. Figure 8 - Position Loop Position Command s Feed Forwards s Adaptive Tuning Kvff Kaff Torque Loop Filters System Under Control P REF RN PI PI K J LL LP N K T J T s 2 P REG V REG Load Torque Torque Estimate Velocity Feedback Filter Fs Velocity Estimate Load Observer Position Feedback 24 Rockwell Automation Publication 750-AT006A-EN-P - June 207

25 Background Chapter The position, velocity, and torque loops are all enabled in position mode. Elements of the position loop are summarized: Position Reference (PRef ) The position reference can come from various sources. Some are selected through digital inputs or via bit manipulation of the Network Logic Command Word: Network Communication from Logix Controller position direct mode Internal PTP planner point-to-point position mode Analog Input Encoder Position Reference Process PID Controller See the PowerFlex 750 Reference Manual, publication 750-RM002, and PowerFlex Drives with TotalFORCE Control Programming Manual, publication 750-PM00, for more information. Reference Notch Filters (RN) Two reference notch filters are placed on command signals to remove load-side resonances and vibration that is common in, robots, cranes, cantilevered loads, liquid sloshing, laser cutting, and material handling applications. See Load Side Resonances on page 6 for more information. These filters are capable of adjustable width and depth. See Notch Filters on page 54 for more information. Position Regulator (PReg) The position regulator consists of a proportional-integral (PI) controller. In legacy PowerFlex 755 drives, Kpp is in parallel with Kpi. Here, legacy Kpp 0:839 [Psn Reg Kp] is the position loop bandwidth in units of [rad/sec] and legacy Kpi 0:838 [Psn Reg Ki] has a squared relationship to it in units of [rad 2 /sec 2 ]. In this drive however, Kpp is in series with Kpi and a factor of 2π is applied to each gain. Figure 9 - Position Regulator from Parallel to Series Form K PP [rad/sec] 2 K PP s K PI [rad 2 /sec 2 ] s 2 K PI [Hz] [Hz] PowerFlex 755 PowerFlex 755T As a result, all tuning parameters represent bandwidths of physically measurable signals that are easy to understand. More importantly, removing the squared relationships simplifies math when tuning a drive, because all gains and filter bandwidths are now related to each other by simple ratios. This approach makes the tuning experience more transparent and intuitive. Rockwell Automation Publication 750-AT006A-EN-P - June

26 Chapter Background Use the following equation to convert legacy position loop gains to PowerFlex 755T gains. K PP = K PP LegacyP839 K PI LegacyP , K = PI 2 K PP LegacyP839 Velocity Feedfor ward Gain (Kvff ) Velocity feedforward is set to 00% in the drive, resulting in Kvff =. Velocity Loop The velocity loop is active in position and velocity control modes. In position mode, the position, velocity, and torque loops are all enabled. Here, the velocity loop consists of the highlighted elements that are shown in the following figure. Figure 20 - Velocity Loop Position Command s Feed Forwards Kvff s Kaff Adaptive Tuning Torque Loop Filters System Under Control P REF RN PI PI K J LL LP N K T J T s 2 P REG V REG Load Torque Torque Estimate Velocity Feedback Filter Fs Velocity Estimate Load Observer Position Feedback In velocity mode, the velocity and torque loops are enabled and the position loop is replaced by a velocity reference command. Here, the velocity loop consists of the highlighted elements that are shown in the following figure. Figure 2 - Velocity Reference Feed Forward s Kaff Adaptive Tuning Torque Loop Filters System Under Control V REF RN PI K J LL LP N K T J T s 2 V REG Load Torque Torque Estimate Velocity Feedback Filter Fs Velocity Estimate Load Observer Position Feedback 26 Rockwell Automation Publication 750-AT006A-EN-P - June 207

27 Background Chapter Elements of the velocity loop are summarized: Velocity Reference (VRef ) The velocity reference can come from various sources. Some are selected through digital inputs or via bit manipulation of the Network Logic Command Word: HIM (local or remote) Analog Input Preset Speed Parameters Jog Speed Parameters Encoder Velocity Reference Network Communication Process PID Controller See the PowerFlex 750 Reference Manual, publication 750-RM002, and PowerFlex Drives with TotalFORCE Control Programming Manual, publication 750-PM00, for more information. Reference Notch Filters (RN) Two reference notch filters are placed on command signals to remove load-side resonances and vibration that is common in robots, cranes, cantilevered loads, liquid sloshing, laser cutting, and material handling applications. See Load Side Resonances on page 6 for more information. These filters are capable of adjustable width and depth. See Notch Filters on page 54 for more information. Velocity Feedback Filter (Fs) The velocity feedback filter actually consists of two filters that work in combination to remove quantization noise that is generated by low-resolution feedback devices. These filters are typically disabled when high-resolution feedback devices are used. There is an FIR filter followed by a velocity feedback low pass filter on the primary feedback channel. There is another set of these filters on the alternate feedback channel. The primary feedback channel is used when the Automatic Tach Switchover feature is disabled or before a switchover has occurred. Feedback is then switched from the primary to the alternate feedback channel when the Automatic Tach Switchover feature is enabled and a switchover has occurred. Relevant parameters are given: 0:000 [Pri Vel Fb Sel] Enter the 2-digit port location followed by the 4-digit parameter number of the primary feedback device. 0:00 [Vel Fb Taps] This parameter sets the number of primary FIR taps used to calculate velocity as the derivative of position over a sample time corresponding to 2 n number of taps. Rockwell Automation Publication 750-AT006A-EN-P - June

28 Chapter Background 0:002 [c Vel Fb LPF BW] This parameter is automatically calculated for the primary velocity feedback low pass filter bandwidth. It is applied to the drive when 0:905 [System C/U Select] = Calculated (0). It is calculated as a function of the following parameters and it affects the DMTC and 0:906 [System BW] values. 0:407 [Motor Poles] This parameter is the number of motor poles (p). It is calculated as follows: p = round (20 x [Motor NP Hertz] / [Motor NP RPM]) Primary Encoder Resolution This parameter is the total resolution in edge counts per revolution (EPR) specified by parameters on the primary feedback option card. Low-Resolution Example: Resolution = 024 pulses per revolution * 4 quadrature edge counts per pulse = 4096 EPR (2-bit). The lowresolution PPR comes directly from a parameter on the option card. When both A and B channels are selected for an incremental encoder, the edge count multiplier is 4. This value is the typical and default setting. When only channel A is selected, then the edge count multiplier is 2. High-Resolution Example: Resolution = 024 pulses per revolution * 024 edge counts per pulse =,048,576 EPR (20-bit). For high-resolution devices, the overall resolution choices are 20-bit default or an optional 24-bit when the corresponding configuration bit is selected. 0:003 [u Vel Fb LPF BW] This parameter sets the primary velocity feedback low pass filter bandwidth. This value is applied to the drive when 0:905 [System C/U Select] = User Entered (). 0:006 [Alt Vel Fb Sel] Enter the 2-digit port location followed by the 4- digit parameter number of the alternate feedback device. 0:007 [Alt Vel Fb Taps] This parameter sets the number of alternate FIR taps used to calculate the derivative of position. 0:008 [c AltVelFbLPF BW] This parameter is automatically calculated for the alternate velocity feedback low pass filter bandwidth. It is applied to the drive when 0:905 [System C/U Select] = Calculated (0). It is calculated as a function of the following parameters and it affects the DMTC and 0:00 [Alt Fb GnScale] values. 0:407 [Motor Poles] This parameter is the number of motor poles (p). It is calculated as follows: p = round (20 x [Motor NP Hertz] / [Motor NP RPM]) Alternate Encoder PPR and Configuration This parameter is the total resolution in edge counts per revolution (EPR) specified by parameters on the alternate feedback option card. 28 Rockwell Automation Publication 750-AT006A-EN-P - June 207

29 Background Chapter Low-Resolution Example: Resolution = 024 pulses per revolution * 4 quadrature edge counts per pulse = 4096 EPR (2-bit). The lowresolution PPR comes directly from a parameter on the option card. When both A and B channels are selected for an incremental encoder, the edge count multiplier is 4. This value is the typical and default setting. When only channel A is selected, then the edge count multiplier is 2. High-Resolution Example: Resolution = 024 pulses per revolution * 024 edge counts per pulse =,048,576 EPR (20-bit). For highresolution devices, the overall resolution choices are 20-bit default or an optional 24-bit when the corresponding configuration bit is selected. 0:009 [u AltVelFbLPF BW] This parameter sets the alternate velocity feedback low pass filter bandwidth. This value is applied when 0:905 [System C/U Select] = User Entered (). More FIR taps filter out more noise, but reduces the overall bandwidth attainable in the velocity and position loops. Similarly, a lower velocity feedback LPF bandwidth filters out more noise, but also reduces the overall bandwidth attainable in the velocity and position loops. Typical values are given in the following table. Table 2 - Typical Velocity Feedback Filter Values 0:407 [Motor Poles] Encoder PPR Number of Taps Velocity Fb LPF BW [Hz] 8 024* * (Filter Off) (Filter Off) Velocity Regulator (VReg) The velocity regulator consists of a proportional-integral (PI) controller. In legacy PowerFlex 755 drives, Kvp is in parallel with Kvi. Here, legacy Kvp 0:645 [Speed Reg Kp] equals 0:636 [Speed Reg BW] in units of [rad/sec] times 0:76 [Total Inertia] in units of [sec] and legacy Kpi 0:647 [Speed Reg Ki] has a squared relationship to Kvp in units of [rad 2 /sec]. In this drive however, Kpp is in series with Kpi and a factor of 2π is also applied to each gain. Figure 22 - Velocity Regulator from Parallel to Series Form K BW * VP J T [rad] 2 K VP s K VI BW * K 4z VP 2 [rad 2 /sec] s 2 K VI [Hz] [Hz] PowerFlex 755 PowerFlex 755T Rockwell Automation Publication 750-AT006A-EN-P - June

30 Chapter Background As a result, all tuning parameters represent bandwidths of physically measurable signals that are easy to understand. More importantly, removing the squared relationships simplifies math when tuning a drive, because all gains and filter bandwidths are now related to each other by simple ratios. This approach makes the tuning experience more transparent and intuitive. Use the following equation to convert legacy position loop gains to PowerFlex 755T gains. K VP K VP LegacyP645 K = , VI LegacyP647 K 2 J VI = T 2 K LegacyP76 VP LegacyP645 0:952 [Servo Lock Gain] This parameter sets the gain of an additional integrator in the velocity regulator. The effect of Servo Lock is to increase stiffness of the velocity response to a load disturbance. It behaves like a position regulator with velocity feed forward, but without the pulse accuracy of a true position regulator. The gain is normally set to less than 0. times system bandwidth, or for the desired response. A value of zero disables this feature. Load Observer Load observer forces mechanical loads to behave consistently by compensating for unknown inertia, changing inertia, and unknown amounts of friction, backlash, and non-rigidity present in the mechanics. This function allows out-of-box tuning to yield high performance more often. See Load Observer on page 39 for more information. Acceleration Feed Forward (Kaff ) During velocity changes, a certain level of torque is required to overcome load inertia over and above the level of torque used to run at constant speed. The acceleration feed forward signal attempts to predict the motor torque required to accelerate and decelerate the load inertia. The 0:2070 [Accel FF Output] signal can be fed forward into the torque reference, becoming an available input to the 0:34 [PsnVelTrq Actv] selector to be summed with 0:969 [VReg Output] making for smoother accelerations and decelerations, especially in high dynamic applications. These signals have acceleration units of [rev/sec 2 ]. They are applied before the Torque Scaler conversion to percent [% motor torque]. 0:972 [Accel FF Mode] enables this feature and selects possible velocity reference sources. 0:972 [Accel FF Mode] = Disabled (0) Disables the function. 0:972 [Accel FF Mode] = Int Ramp Ref () Selects the internal ramp for the source. The internal ramp is generated from the rate of change of 0:925 [VRef Filtered], the typical setting that is used for inertia compensation on a standalone drive. 0:972 [Accel FF Mode] = Ext Ramp Ref (2) Selects the external ramp reference for the source. This reference comes from 0:978 [Ext Ramped Ref ]. This setting is available for applications that supply a ramped speed reference external to the drive. For example, when the velocity ramp is generated in a Logix Controller. 30 Rockwell Automation Publication 750-AT006A-EN-P - June 207

31 Background Chapter 0:972 [Accel FF Mode] = Spd Rate Ref (3) Selects the Feed Forward Velocity Rate for the source. This rate comes from 0:93 [FF Vel Rate Ref ]. This selection is also used with an externally supplied ramp rate. 0:973 [c Accel FF Gain] This parameter is a calculated gain that is applied as Kaff in the drive during acceleration when 0:905 [System C/U Select] = Calculated (0). 0:974 [u Accel FF Gain] This parameter is applied as Kaff in the drive during acceleration when 0:905 [System C/U Select] = User Entered (). 0:975 [Accel FF GainNeg] This parameter is applied as Kaff in the drive during deceleration. 0:2070 [Accel FF Output] This parameter displays the inertia compensation output in [rev/sec 2 ]. 0:978 [Ext Ramped Ref ] This parameter is used when 0:972 [Accel FF Mode] = Ext Ramp Ref (2). Send a value to this parameter from an external controller to generate the Acceleration Feedforward signal. It is meant for an external motor speed ramp input signal and is entered in units of Hz or RPM, depending on the value of 0:89 [Velocity Units]. For additional illustration of the control, refer to PF755 Control Block Diagrams in the PowerFlex Drives with TotalFORCE Control Programming Manual, publication 750-PM00. Torque Loop The torque loop is active in position, velocity, and torque control modes. In position mode, the position, velocity, and torque loops are enabled. In velocity mode, the velocity and torque loops are enabled. Here, the torque loop consists of the highlighted elements that are shown in the following figure. Figure 23 - Torque Loop Position Command s Feed Forwards s Adaptive Tuning Kvff Kaff Torque Loop Filters System Under Control P REF RN PI PI K J LL LP N K T J T s 2 P REG V REG Load Torque Torque Estimate Velocity Feedback Filter Fs Velocity Estimate Load Observer Position Feedback Rockwell Automation Publication 750-AT006A-EN-P - June 207 3

32 Chapter Background In torque mode, the torque loop is enabled and the position and velocity loops are replaced by a torque reference command. Here, the torque loop consists of the highlighted elements that are shown in the following figure. Figure 24 - Torque Reference Torque Loop Filters System Under Control T REF LL LP N K T J T s 2 Load Torque Elements of the torque loop are summarized: Torque Reference Selection The torque reference can come from various sources. Some can be selected through digital inputs or via bit manipulation of the Network Logic Command Word: Velocity Regulator output HIM (local or remote) Preset Torque Parameters Analog Input Network Communication Process PID Controller See the PowerFlex Drives with TotalFORCE Control Programming Manual, publication 750-PM00, for more information. Torque Scaler (K J ) The torque scaler is an overall gain in the system that affects tuning and overall position and velocity mode performance. See Torque Scaler on page 35 for more information. Friction Compensation Friction compensation is used in some applications to compensate for known amounts of friction. It is not shown in the previous figures. See Friction Compensation on page 93 for more information. However, load observer is often used instead because it requires no knowledge of these mechanical quantities. See Load Observer on page 39 for more information. Torque Notch Filters (N) The torque notch filters consist of 4 second order filters that each have adjustable width, depth, and gain. See Notch Filters on page 54 for more information. The adaptive tuning feature can be configured to automatically adjust these filters. See Adaptive Tuning on page 44 for more information. These filters are typically used to remove mid and high frequency motor-side resonances. See Resonances on page 4 for more information. Torque Low-Pass Filter (LP) The torque low-pass filter is a first order filter. See Low Pass Filters on page 60 for more information. The adaptive tuning Gain Stabilization feature can be configured to automatically adjust this filter. See Adaptive Tuning on page 44 for more information. 32 Rockwell Automation Publication 750-AT006A-EN-P - June 207

33 Background Chapter Torque Lead-Lag Filter (LL) The torque lead-lag filter is a first order filter with adjustable gain. See Lead Lag Filters on page 63 for more information. Adaptive Tuning The adaptive tuning feature can be configured to automatically adjust torque notch and low pass filters. See Adaptive Tuning on page 44 for more information. Rockwell Automation Publication 750-AT006A-EN-P - June

34 Chapter Background Notes: 34 Rockwell Automation Publication 750-AT006A-EN-P - June 207

35 Chapter 2 Product Features Topic Page Product Features 35 Torque Scaler 35 Load Observer 39 Adaptive Tuning 44 Notch Filters 54 Low Pass Filters 60 Lead Lag Filters 63 Torque Scaler Torque Scaler (K J ) is a torque loop gain that accounts for load inertia by multiplying by its inverse. Since inertia converts torque to acceleration, then the torque scaler converts acceleration to torque. Scaling torque does the following: Makes the velocity loop response match the velocity loop bandwidth Allows the velocity loop response to not be affected by motor gain or load inertia Calibrates the control loops so that all gains represent physically measurable bandwidths Scales the system under control to unity gain Acts as an overall system gain Torque scaler is calculated from the following parameters. It is a function of rated motor torque and total inertia. 0:403 [Motor NP RPM] 0:405 [Mtr NP Pwr Units] 0:406 [Motor NP Power] 0:900 [Motor Inertia] 0:90 [Load Ratio] Rockwell Automation Publication 750-AT006A-EN-P - June

36 Chapter 2 Product Features Furthermore, it is recalculated anytime one of these parameters change. Equations for torque scaler are given: When 0:405 [Mtr NP Pwr Units] = 0 and 0:406 [Motor NP Power] are in units of HP Motor Rated Torque [Nm] = ([Motor NP Power] / [Motor NP RPM]) x System Acceleration [rev/sec 2 ] = [Motor NP Power] /([Motor NP RPM] x [Motor Inertia] x ([Load Ratio]+)) x When 0:405 [Mtr NP Pwr Units] = and 0:406 [Motor NP Power] are in units of kw Motor Rated Torque [Nm] = ([Motor NP Power]/[Motor NP RPM]) x System Acceleration [rev/sec 2 ] = [Motor NP Power]/([Motor NP RPM] x [Motor Inertia] x ([Load Ratio] + )) x Torque Scaler [%/(rev/sec 2 )] = 00 / System Acceleration Torque Scaler [sec] = [Motor NP RPM]/ (System Acceleration * 60) Let s review a few cases to understand how it is calculated. Rigid Load Rigid loads with constant inertia are typically found in high performance applications. There is a direct coupling to the motor with few mechanical components. The mechanical components have no flex or twist and there is no misalignment. The following figure shows a rigid mechanical load with a load inertia that does not change over time. K T is the motor rated torque constant. Figure 25 - System with a Rigid Load System Under Control Motor Electrical K T Motor and Load Mechanics J T s 2 Motor Position Load Torque The motor and load mechanics are represented with J T, the total of all inertia in the system. In this case, the torque scaler K J = J T /K T brings the system under control to unity gain. This places K J at its maximum value because J T is the largest inertia value the system can be. As a gain in the signal path, K J generates a high system gain that produces great performance when the system is in fact rigid. However, if the system is not rigid, and most systems are not, then this high gain could produce instability. As a result, the control loop and observer gains must be reduced. Also with high gain, the velocity loop and load observer only have enough stability margin to compensate for small disturbances in the load torque. For rigid systems, the load ratio R > 0 is 36 Rockwell Automation Publication 750-AT006A-EN-P - June 207

37 Product Features Chapter 2 calculated by Autotune or it is a known positive nonzero value and J T = J M *(R+). Non-Rigid Load Most mechanical systems have a non-rigid load. They exhibit some combination of compliance, backlash, unknown dynamics, and unknown or changing inertia. Here, only the effect of motor inertia ( J M ) is immediate and a direct multiplier on rigid body mechanics. As a result, the motor and load mechanics are represented with J M and the effect of load inertia is dynamic and reflected in the load torque signal being applied to the motor shaft. Figure 26 - System with an Unknown, Non-rigid, or Changing Load System Under Control Motor Electrical K T Motor and Load Mechanics J M s 2 Motor Position Load Torque In this case, the torque scaler K J = J M /K T brings the system under control to unity gain. This places K J at its minimum value because J M is the smallest inertia value the system can be; a no-load condition. As a gain in the signal path, K J generates a lower system gain, which allows the control loop and observer gains room to be increased. Here, the velocity loop and load observer have enough stability margin to compensate for the entire load torque. For non-rigid systems or when the load is unknown, the load ratio R = 0. Calculation It has been shown that K J = J T /K T for rigid loads and K J = J M /K T for non-rigid loads. In the drive, torque scaler is calculated as K J = J M (R+)/K T to allow for all cases. Figure 27 - Torque Scaler Calculation System Under Control Torque Scalar (K J ) J M ( R ) K T Motor Electrical K T J M Motor and Load Mechanics ( R ) s 2 Motor Position Load Torque For rigid systems, the load ratio R > 0 is calculated by autotune or it is a known positive nonzero value and J T = J M *(R+). For non-rigid systems or when the load is unknown, the load ratio R = 0 and J T = J M. Rockwell Automation Publication 750-AT006A-EN-P - June

38 Chapter 2 Product Features 0:902 [Load Coupling] This parameter reflects the type of mechanical coupling between the motor and load. It adjusts calculated control loop gains during an autotune bandwidth calculation test. 0:902 [Load Coupling] = Rigid (0) is where the load consists of few mechanical components with direct connection to the motor shaft. The components are high performance, have no flex, or twist, and misalignment is not likely to occur. This setting is used for high performance machines. 0:902 [Load Coupling] = Compliant () is a non-rigid load where position misalignment, backlash, and flexing of couplings, gearboxes, belts, and shafts can occur which creates an indirect connection of the load to the motor shaft. This setting is common for most machines. The term out-of-box refers to default control loop gain settings that are preconfigured in a new drive. Since the load is unknown at this point, the motor is assumed to be unloaded, the load ratio R = 0, and the load coupling is compliant. This configuration also applies when the load is known to be compliant. See Out-of-Box Tuning on page 65 for more information. However, when the load is known or an autotune total inertia test has been performed to determine the load ratio, the control loop gains are configured for a load ratio R > 0 and the load coupling is rigid. See Auto Tuning on page 75 for more information. This value is the primary difference affecting out-of-box and autotune rules. Thus, the term out-of-box implies a compliant load with R = 0 and the term autotune implies a rigid load with R > 0. Recommended settings are shown in the following table. Table 3 - Recommended Load Ratio Settings Control Mode Load Coupling Tuning Case Load Ratio Torque Scaler Load Observer Mode Load Observer Bandwidth Velocity Compliant Out of Box R = 0 K J = J M /K T Load Observer Only Kop = Kvp Rigid Autotune R > 0 K J = J M *(R+)/K T Kop = Kvp Position Compliant Out of Box R = 0 K J = J M /K T Load Observer with Velocity Kop = 4* Kvp Estimate Rigid Autotune R > 0 K J = J M *(R+)/K T Kop = Kvp Other considerations are given: The motor torque constant K T actual value in the motor changes over time. It decreases as the motor heats up. 38 Rockwell Automation Publication 750-AT006A-EN-P - June 207

39 Product Features Chapter 2 Load Observer The load observer feature is a control loop inside the drive that estimates the mechanical load on the motor and compensates for it while the drive is running. This feature allows for high performance and control loop tuning simplicity similar to that of a mechanically disconnected motor. Its primary function is to: Automatically compensate for unknown inertia, compliance, and low frequency resonance Automatically compensate for disturbances and changing inertia Force consistent dynamic behavior, which makes the drive easy to tune Benefits You can use the load observer with out-of-box control loop gains, where the load is unknown or compliant and thus 0:90 [Load Ratio] = 0, or with autotuned control loop gains, where 0:90 [Load Ratio] > 0 is a known positive value or calculated by performing an autotune procedure. When the load observer is enabled with the recommended out-of-box control loop gains, load observer does the following: Automatically compensates for changing inertia, compliance, and low frequency resonances Provides relatively high performance control without tuning There is no longer the need for a tuning expert No periodic retuning to account for machine wear over time Reduces commissioning time, especially for high drive count When the load observer is enabled with autotuned control loop gains, load observer does the following: Automatically compensates for load disturbances Minimizes tracking errors, machine vibration, and power consumption Allows bandwidth and line speeds to be increased Provides tighter control of moving parts, which reduces wear and saves material costs How It Works Load observer acts on the acceleration signal within the control loops and monitors the Acceleration Reference and Position Feedback signals. Load observer uses an internal plant model (P) of an ideal motor and generates a load Torque Estimate signal that represents any deviation in response between the actual loaded motor and the ideal model. This deviation represents the reaction torque that is placed on the motor shaft by a mechanical load. The load torque is estimated in real-time and compensated by closed loop operation. This technique provides system response that is similar to the response of a mechanically disconnected motor. Rockwell Automation Publication 750-AT006A-EN-P - June

40 Chapter 2 Product Features In the drive, the load Torque Estimate is applied through an Acceleration Estimate that is subtracted from the Acceleration Reference signal, which is then converted to a torque signal by multiplying by the Torque Scaler. Figure 28 - Load Observer Block Diagram Unloaded Motor Response System Under Control P REF Control Loops Power Converter Unloaded Motor Velocity Feedback Filter Fs Torque Estimate Velocity Estimate Load Observer Acceleration Reference Torque Load Position Feedback As a result, the control loops can be tuned as if controlling a disconnected motor. The load observer also generates a Velocity Estimate signal that can be applied to the velocity loop. The Velocity Estimate has less delay than the Velocity Feedback signal that comes from the velocity feedback FIR and low pass filters. It also helps to reduce high frequency output noise caused by the load observer aggressively acting on the Acceleration Reference signal. Together, load observer with velocity estimate provides the best overall performance for position control applications. Configuration A block diagram of the load observer is given and it shows the internal signals and gains. Figure 29 - Load Observer Configuration 0:2027 [LdObs Acc Ref In] 0:2037 [LdObs AccRef Out] 0:2036 [LdObs Torque Est] K J 0:2035 [LdObs Accel Est] K OF K OU Position Feedback FIR 0:2028 [LdObs VelFb In] 0:203 [LdObs Vel Error] 2 K OP P 0:2034 [LdObs Vel Est] s 2 K OI Load Observer Plant 40 Rockwell Automation Publication 750-AT006A-EN-P - June 207

41 Product Features Chapter 2 There are seven signals that are mapped to the following parameters: 0:2027 [LdObs Acc Ref In] Load observer acceleration reference input signal 0:2028 [LdObs VelFb In] Load observer velocity feedback input signal 0:203 [LdObs Vel Error] Load observer velocity error signal 0:2034 [LdObs Vel Est] Load observer velocity estimate signal 0:2035 [LdObs Accel Est] Load observer acceleration estimate signal 0:2036 [LdObs Torque Est] Load observer torque estimate signal 0:2037 [LdObs AccRef Out] Load observer acceleration reference output signal There are five gains, however only two require user interaction: load observer bandwidth Kop and load observer integrator bandwidth Koi. They are mapped to the following parameters: 0:202 [c LdObs Kp] This value is applied to Kop when 0:905 [System C/U Select] = Calculated (0). This sets load observer bandwidth in units of [Hz]. 0:2022 [u LdObs Kp] This value is applied to Kop when 0:905 [System C/U Select] = User Entered (). This sets load observer bandwidth in units of [Hz]. 0:2023 [c LdObs Ki] This value is applied to Koi when 0:905 [System C/U Select] = Calculated (0). This sets load observer integrator bandwidth in units of [Hz]. 0:2024 [u LdObs Ki] This value is applied to Koi when 0:905 [System C/U Select] = User Entered (). This sets load observer integrator bandwidth in units of [Hz]. Guidelines for setting these gains are provided in the following sections. In general, Kop operates like a good velocity integrator without windup and Koi operates like a good position integrator without windup. Koi = 0 by default. Load observer gains that do not require user interaction are the load observer feedback gain Kof and the load observer input gain Kou. They are set internally based on the load observer configuration. The torque scaler K J is also set internally. See Torque Scaler on page 35 for more information on which parameters affect K J. Recommended settings are shown in the following table. Table 4 - Recommended Load Observer Settings Control Mode Load Coupling Tuning Case Load Ratio Torque Scaler Load Observer Mode Load Observer Bandwidth Velocity Compliant Out of Box R = 0 K J = J M /K T Load Observer Only Kop = Kvp Rigid Autotune R > 0 K J = J M *(R+)/K T Kop = Kvp Position Compliant Out of Box R = 0 K J = J M /K T Load Observer with Velocity Kop = 4* Kvp Estimate Rigid Autotune R > 0 K J = J M *(R+)/K T Kop = Kvp Rockwell Automation Publication 750-AT006A-EN-P - June 207 4

42 Chapter 2 Product Features Other considerations are given: When the load observer and the torque low-pass filter are both enabled and the low-pass filter bandwidth is less than five times the load observer bandwidth, their interaction can interfere with each other and cause instability. 0:2020 [LdObs Mode] This parameter sets the load observer mode. Each mode is described in the following sections. Disabled 0:2020 [LdObs Mode] = Disabled (0) This mode disables the load observer function. The Torque Estimate and Velocity Estimate signals of the load observer are not applied to the control loops. The velocity regulator uses the default filtered Velocity Feedback instead of the Velocity Estimate signal. Figure 30 - Load Observer Disabled Configuration System Under Control P REF Control Loops Power Converter Unloaded Motor Velocity Feedback Filter Fs Torque Estimate Velocity Estimate Load Observer Acceleration Reference Torque Load Position Feedback Load Observer Only 0:2020 [LdObs Mode] = LdObs Only () This mode is recommended for velocity control applications and it functions equivalent to the legacy PowerFlex load observer. It compensates for dynamic loads that are connected to the motor, providing high disturbance rejection and dynamic stiffness. However, the aggressive behavior in position control applications often requires the observer bandwidth to be decreased for stable operation. This mode applies the Torque Estimate signal to the control loops but not the Velocity Estimate signal. The velocity regulator uses the default filtered Velocity Feedback instead of the Velocity Estimate signal. 42 Rockwell Automation Publication 750-AT006A-EN-P - June 207

43 Product Features Chapter 2 Figure 3 - Load Observer Only Configuration System Under Control P REF Control Loops Power Converter Unloaded Motor Velocity Feedback Filter Fs Torque Estimate Velocity Estimate Load Observer Acceleration Reference Torque Load Position Feedback Load Observer with Velocity Estimate 0:2020 [LdObs Mode] = LdObs VelEst (2) This mode is recommended for position control applications. It performs well in situations with changing inertia and unknown levels of compliance and backlash. It compensates for most loads that are connected to the motor, providing high disturbance rejection and dynamic stiffness, and allows gains to be increased for quicker system response. It is not desirable for velocity mode applications because a steady state velocity error could be present that is hard to detect. This mode applies both Torque and Velocity Estimate signals to the control loops. It replaces the default filtered Velocity Feedback signal with the Velocity Estimate signal. Figure 32 - Load Observer with Velocity Estimate Configuration System Under Control P REF Control Loops Power Converter Unloaded Motor Velocity Feedback Filter Fs Torque Estimate Velocity Estimate Load Observer Acceleration Reference Torque Load Position Feedback Velocity Estimate Only 0:2020 [LdObs Mode] = Vel Est Only (3) This mode can be used in some position control applications. It removes phase lag associated with velocity feedback filtering and allows gains to be increased for quicker system response. It is not desirable for velocity mode applications because a steady state velocity error is present that is hard to detect. This mode applies the Velocity Estimate signal to the control loops but not the Torque Estimate signal. It replaces the default filtered Velocity Feedback signal with the Velocity Estimate signal. Rockwell Automation Publication 750-AT006A-EN-P - June

44 Chapter 2 Product Features Figure 33 - Velocity Estimate Only Configuration System Under Control P REF Control Loops Power Converter Unloaded Motor Velocity Feedback Filter Fs Torque Estimate Velocity Estimate Load Observer Acceleration Reference Torque Load Position Feedback Acceleration Feedback 0:2020 [LdObs Mode] = Accel Fdbk (4) This mode functions equivalent to the legacy Inertia Adaption. It compensates for some loads that are connected to the motor, providing disturbance rejection and dynamic stiffness. However, the aggressive behavior often requires the observer bandwidth to be decreased for stable operation. This mode applies the Torque Estimate signal to the control loops but not the Velocity Estimate signal. The velocity regulator uses the default filtered Velocity Feedback instead of the Velocity Estimate. The Torque Estimate signal consists of filtered acceleration feedback. Figure 34 - Acceleration Feedback Configuration System Under Control P REF Control Loops Power Converter Unloaded Motor Velocity Feedback Filter Fs Torque Estimate Velocity Estimate Load Observer Acceleration Reference Torque Load Position Feedback Adaptive Tuning The adaptive tuning feature is an algorithm inside the drive that continuously monitors and, if necessary, adjusts or adapts the various filter parameters and control loop gains to compensate for the unknown and changing load conditions while the drive is running. Its primary function is to: Monitor motor side resonances Automatically adjust torque loop notch and low pass filter parameters to suppress resonances Automatically de-tune control loop gains to avoid instability when it is detected 44 Rockwell Automation Publication 750-AT006A-EN-P - June 207

45 Product Features Chapter 2 Benefits When adaptive tuning is enabled with recommended out-of-box control loop settings, adaptive tuning does the following: Automatically suppresses continuously changing resonances There is no need for periodic identification of resonances and retuning There is no longer the need for a tuning expert There is no need for periodic retuning of filters over time Reduces commissioning time, especially for high drive count Minimizes tracking error, machine vibration, and power consumption Allows bandwidth and line speeds to be increased Provides tighter control of moving parts, which reduces wear and saves on material costs How It Works Adaptive tuning is always running in the background to detect motor side resonances. Every second, it analyzes the frequency response of torque loop signals to identify, track, and measure resonances. It also analyzes the frequency response of command signals to make sure that dominant command frequencies are not mistaken for resonances. This technique is known as command rejection. The action taken to adaptively change tuning parameters largely depends on the adaptive tuning mode of operation. This approach is covered in the following sections. Rockwell Automation Publication 750-AT006A-EN-P - June

46 Chapter 2 Product Features Configuration A block diagram of adaptive tuning is given and it shows relevant parameters. Figure 35 - Adaptive Tuning Configuration Table 5 - Adaptive Tuning Settings All torque filter parameters go through adaptive tuning so that adaptive tuning can take control of these filters when required. Parameter settings are summarized in the following table, along with descriptions of how they work in various modes of operation. Parameter Name Description 0:20 [AdptTune Config] 0:2 [Trq NF Threshold] 0:22 [Trq NF Freq LLim] 0:23 [Trq NF Freq HLim] 0:24 [Trq NF WidthMin] 0:25 [Trq NF WidthMax] 0:26 [Trq LPF BW LLim] 0:29 [AdptTuneMinScale] Adaptive Tuning Configuration Torque Notch Filter Tuning Threshold Torque Notch Filter Low Frequency Limit Torque Notch Filter High Frequency Limit Torque Notch Filter Minimum Width Torque Notch Filter Maximum Width Torque Low Pass Filter Bandwidth Low Limit Adaptive Tuning Gain Scaling Factor Minimum Selects the adaptive tuning mode of operation. See sections Disabled Tracking Notch on page 50 and Gain Stabilization on page 5 for a detailed description of each mode. Adaptive tuning identifies resonances that are not associated with command signals between these low and high frequency limits with magnitudes above this tuning threshold. The adaptive tuning Tracking Notch function adjusts torque notch filter widths proportional to the frequency estimate in relation to the high and low frequency limits. It sets torque notch filter widths equal to this minimum width when the frequency estimate is equal to the low frequency limit. It sets torque notch filter widths equal to this maximum width when the frequency estimate is equal to the high frequency limit. The default setting is for minimum and maximum width. The adaptive tuning Gain Stabilization function incrementally decreases the actual torque low pass filter bandwidth to stabilize the system if necessary or until the actual torque low pass filter bandwidth equals this low limit. The adaptive tuning Gain Stabilization function incrementally decreases the gain scaling factor to stabilize the system if necessary or until the gain scaling factor equals this low limit. 46 Rockwell Automation Publication 750-AT006A-EN-P - June 207

47 Product Features Chapter 2 Internal signals and their descriptions are summarized in the following table. These signals are operational even when adaptive tuning is disabled. Table 6 - Adaptive Tuning Internal Signals Parameter Name Description 0:223 [Trq NF Freq Est] 0:224 [Trq NF Mag Est] Torque Notch Filter Frequency Estimate Torque Notch Filter Magnitude Estimate Displays the center frequency estimate of the first mechanical resonance that is identified by adaptive tuning. If multiple resonances exist with magnitudes above the threshold, then the one with the largest magnitude is selected. If two resonances with equal magnitude exist, then the one with the lowest center frequency is selected. Displays the magnitude estimate of the first resonance that is identified by adaptive tuning. Adaptive tuning identifies resonances and estimates their magnitudes. Table 7 - Adaptive Tuning Outputs Output parameters and their descriptions are summarized in the following table. Parameter Name Description 0:220 Adaptive Tuning Status Displays the adaptive tuning status bits. Descriptions of each bit are given in the next table. [AdptTune Status] 0:22 [AdptTune GnScale] 0:256 [Trq LPF BW Act] 0:260 [Trq NF Freq Act] 0:262 [Trq NF Wdth Act] 0:264 [Trq NF Dpth Act] 0:266 [Trq NF Gain Act] 0:270 [Trq NF2 Freq Act] 0:272 [Trq NF2 Wdth Act] 0:274 [Trq NF2 Dpth Act] 0:276 [Trq NF2 Gain Act] Adaptive Tuning Gain Scaling Factor Torque Low Pass Filter Bandwidth Active Torque Notch Filter Frequency Active Torque Notch Filter Width Active Torque Notch Filter Depth Active Torque Notch Filter Gain Active Torque Notch Filter 2 Frequency Active Torque Notch Filter 2 Width Active Torque Notch Filter 2 Depth Active Torque Notch Filter 2 Gain Active Displays the adaptive tuning Gain Scale Factor. The adaptive tuning Gain Stabilization function incrementally decreases this value to scale position loop, velocity loop, and load observer gains to stabilize the system if necessary. The instability is caused from resonances that are not already suppressed by filters or the instability is caused by filter bandwidths that are too close to the closed loop bandwidth. Displays the active value of the torque low pass filter bandwidth. When the adaptive tuning Gain Stabilization function is not active, the calculated value 0:254 [c Trq LPF BW] is applied when 0:905 [System C/U Sel] = Calculated and the user value 0:255 [u Trq LPF BW] is applied when 0:905 [System C/U Sel] = User Entered. However, when the adaptive tuning Gain Stabilization function is active, adaptive tuning can control the torque low pass filter and incrementally decrease this value to suppress additional resonances that are unsuppressed by notch filters. See 0:20 [AdptTune Config] for adaptive tuning modes. Displays the active center frequency of the first torque notch filter. The user entered value 0:259 [Trq NF Freq] is applied when the adaptive tuning Tracking Notch function is not active. However, the adaptive tuning estimated value 0:223 [Trq NF Freq Est] is applied when the adaptive tuning Tracking Notch function is active. See 0:20 [AdptTune Config] for adaptive tuning modes. Displays the active width of the first torque notch filter. The user entered value 0:26 [Trq NF Width] is applied when the adaptive tuning Tracking Notch function is not active. However, adaptive tuning sets this value between the minimum and maximum widths proportional to the frequency estimate in relation to the high and low frequency limits when the adaptive tuning Tracking Notch function is active. See 0:20 [AdptTune Config] for adaptive tuning modes. Displays the active depth of the first torque notch filter. The user entered value 0:263 [Trq NF Depth] is applied when the adaptive tuning Tracking Notch function is not active. However, adaptive tuning sets this value to 0 when the adaptive tuning Tracking Notch function is active. See 0:20 [AdptTune Config] for adaptive tuning modes. Displays the active gain of the first torque notch filter. The user entered value 0:265 [Trq NF Gain] is applied when the adaptive tuning Tracking Notch function is not active. However, adaptive tuning sets this value to when the adaptive tuning Tracking Notch function is active. See 0:20 [AdptTune Config] for adaptive tuning modes. Displays the active center frequency of the second torque notch filter. The user entered value 0:269 [Trq NF2 Freq] currently passes through adaptive tuning and is applied always. Displays the active width of the second torque notch filter. The user entered value 0:27 [Trq NF2 Width] currently passes through adaptive tuning and is applied always. Displays the active depth of the second torque notch filter. The user entered value 0:273 [Trq NF2 Depth] currently passes through adaptive tuning and is applied always. Displays the active gain of the second torque notch filter. The user entered value 0:275 [Trq NF2 Gain] currently passes through adaptive tuning and is applied always. Rockwell Automation Publication 750-AT006A-EN-P - June

48 Chapter 2 Product Features Table 7 - Adaptive Tuning Outputs 0:280 [Trq NF3 Freq Act] 0:282 [Trq NF3 Wdth Act] 0:284 [Trq NF3 Dpth Act] 0:286 [Trq NF3 Gain Act] 0:290 [Trq NF4 Freq Act] 0:292 [Trq NF4 Wdth Act] 0:294 [Trq NF4 Dpth Act] 0:296 [Trq NF4 Gain Act] Torque Notch Filter 3 Frequency Active Torque Notch Filter 3 Width Active Torque Notch Filter 3 Depth Active Torque Notch Filter 3 Gain Active Torque Notch Filter 4 Frequency Active Torque Notch Filter 4 Width Active Torque Notch Filter 4 Depth Active Torque Notch Filter 4 Gain Active Displays the active center frequency of the third torque notch filter. The user entered value 0:279 [Trq NF3 Freq] currently passes through adaptive tuning and is applied always. Displays the active width of the third torque notch filter. The user entered value 0:28 [Trq NF3 Width] currently passes through adaptive tuning and is applied always. Displays the active depth of the third torque notch filter. The user entered value 0:283 [Trq NF3 Depth] currently passes through adaptive tuning and is applied always. Displays the active gain of the third torque notch filter. The user entered value 0:285 [Trq NF3 Gain] currently passes through adaptive tuning and is applied always. Displays the active center frequency of the fourth torque notch filter. The user entered value 0:289 [Trq NF4 Freq] currently passes through adaptive tuning and is applied always. Displays the active width of the fourth torque notch filter. The user entered value 0:29 [Trq NF4 Width] currently passes through adaptive tuning and is applied always. Displays the active depth of the fourth torque notch filter. The user entered value 0:293 [Trq NF4 Depth] currently passes through adaptive tuning and is applied always. Displays the active gain of the fourth torque notch filter. The user entered value 0:295 [Trq NF4 Gain] currently passes through adaptive tuning and is applied always. Table 8 - Adaptive Tuning Status Bits 0:220 [AdptTune Status] status bits and their descriptions are given in the following table. Name Bit Description Torque Notch Filter Frequency Detected Status Torque Notch Filter Tune Unsuccessful Status Torque Notch Filter Multiple Frequencies Status Torque Notch Filter Frequency Below Limit Status Torque Notch Filter Frequency Above Limit Status 0 This bit is set when resonances are identified between the low and high frequency limits with magnitudes above the tuning threshold. Otherwise, this bit is clear. This bit is also cleared when the drive transitions to the Running state. This bit is set when the tracking notch filters do not compensate all of the identified resonances. Otherwise, this bit is clear. This bit is also cleared when the drive transitions to the Running state or when adaptive tuning transitions from Disabled mode to one of the Tracking Notch modes while in the Running state. 2 This bit is set when multiple resonances are identified between the low and high frequency limits with magnitudes above the tuning threshold. Otherwise, this bit is clear. This bit is also cleared when the drive transitions to the Running state. 3 This bit is set when resonances are identified below the low frequency limit with magnitudes above the tuning threshold. Otherwise, this bit is clear. This bit is also cleared when the drive transitions to the Running state. 4 This bit is set when resonances are identified above the high frequency limit with magnitudes above the tuning threshold. Otherwise, this bit is clear. This bit is also cleared when the drive transitions to the Running state. Adaptive Tune Gain Stabilization Status 5 This bit is set when the gain scaling factor is not equal to one, which indicates that adaptive tuning is controlling the low pass filter and adjusting servo loop gains to stabilize the system. Otherwise, this bit is clear. This bit is also cleared when the drive transitions to the Running state. Adaptive tuning status bits let you create custom logic, trap errors, debug, or manually react to changes. This custom logic is useful for condition monitoring, diagnostics, and preventative maintenance purposes. 48 Rockwell Automation Publication 750-AT006A-EN-P - June 207

49 Product Features Chapter 2 The following table describes when output parameters get reset to their default values. Table 9 - Adaptive Tuning Reset Behavior Parameter Torque Notch Filter Frequency Estimate Torque Notch Filter Magnitude Estimate Torque Low Pass Filter Bandwidth Estimate Adaptive Tuning Gain Scaling Factor When Reset to Default Value Transition to Disabled or Gain Stabilization modes When a resonance is not identified Transition to Disabled or Tracking Notch Filter modes Transition to Disabled or Tracking Notch Filter modes Resonances are characterized in the following way: High frequency (HF) resonances are above the low frequency limit Low frequency (LF) resonances are below the low frequency limit Mid frequency (MF) resonances are slightly above the low frequency limit The following sections describe each adaptive tuning configuration mode in detail. Disabled 0:20 [AdptTune Config] = Disabled (0) Adaptive tuning is always running in the background to identify motor side resonances, even when the feature is disabled. However, no action is taken to compensate for identified resonances in this mode. The result is status only, which allows you react to changes manually or with custom logic. This mode is useful for condition monitoring, diagnostics, and preventative maintenance purposes in tracking HF resonances that change over time. Parts of the control loop structure that are affected by this mode are highlighted below. Figure 36 - Disabled Configuration Position Command s Feed Forwards s Adaptive Tuning Kvff Kaff Torque Loop Filters System Under Control P REF RN PI PI K J LL LP N K T J T s 2 P REG V REG Load Torque Torque Estimate Velocity Feedback Filter Fs Velocity Estimate Load Observer Position Feedback Rockwell Automation Publication 750-AT006A-EN-P - June

50 Chapter 2 Product Features The following figure shows the frequency response of an example resonance to illustrate how it is identified. Figure 37 - Identifying One HF Resonance Frequency response graphs like this are not available to the user. However, adaptive tuning displays all parameters that are shown in the figure. Tracking Notch 0:20 [AdptTune Config] = Tq NF Only () or TqNFGainStb (3) The adaptive tuning Tracking Notch function does three things.. It applies the torque notch filter frequency estimate to the actual torque notch filter frequency instead of the user entered value to suppress resonances if any are identified above the low frequency limit. Parts of the control loop structure affected by this mode are highlighted below. Figure 38 - Tracking Notch Filter Configuration Position Command s Feed Forwards s Adaptive Tuning Kvff Kaff Torque Loop Filters System Under Control P REF RN PI PI K J LL LP N K T J T s 2 P REG V REG Load Torque Torque Estimate Velocity Feedback Filter Fs Velocity Estimate Load Observer Position Feedback 2. It sets the active torque notch filter depth to 0 and the active torque notch filter gain to. These settings ensure that the adaptive tuning Tracking Notch function configures the notch filter as a notch filter during operation regardless of the user entered settings. 50 Rockwell Automation Publication 750-AT006A-EN-P - June 207

51 Product Features Chapter 2 3. It adjusts actual torque notch filter width proportional to the torque notch filter frequency estimate in relation to the high and low frequency limits. Here, the torque notch filter width is set to the minimum width when the torque notch filter frequency estimate is equal to the low frequency limit and the torque notch filter width is set to the maximum width when the torque notch filter frequency estimate is equal to the high frequency limit. This allows tracking notch filters to suppress MF resonances without causing instability. Figure 39 - Identifying One MF Resonance Gain Stabilization 0:20 [AdptTune Config] = GainStb Only (2) or TqNFGainStb (3) The adaptive tuning Gain Stabilization function does two main things.. It enables and tunes the low pass filter to suppress resonances if any are identified above the low frequency limit. Here, the Torque Low Pass Filter Bandwidth Estimate is applied to the torque low pass filter instead of the Torque Low Pass Filter Bandwidth. The bandwidth estimate is incrementally decreased from its default value until the identified HF resonances are suppressed or until a LF resonance or instability occurs. 2. Adaptive tuning detunes control loop gains to suppress any remaining resonances and stabilize the system if necessary. Here, the following gains are scaled by the adaptive tuning Gain Scaling Factor. Load Observer Bandwidth Load Observer Integrator Bandwidth Velocity Loop Bandwidth Velocity Loop Integrator Bandwidth Position Loop Bandwidth Position Loop Integrator Bandwidth This scaling means that the actual control loop gains applied in the regulators are multiplied by the gain scaling factor internally. The scaling factor is incrementally decreased from its default value of until the system is stable. Rockwell Automation Publication 750-AT006A-EN-P - June 207 5

52 Chapter 2 Product Features When Gain Stabilization is not enabled, the scaling factor is reset to its default value so that control loop gains are not affected. Parts of the control loop structure affected by this mode are highlighted below. Figure 40 - Gain Stabilization Configuration Position Command s Feed Forwards s Adaptive Tuning Kvff Kaff Torque Loop Filters System Under Control P REF RN PI PI K J LL LP N K T J T s 2 P REG V REG Load Torque Torque Estimate Velocity Feedback Filter Fs Velocity Estimate Load Observer Position Feedback Gain Stabilization is good for situations where there are more resonances than there are notch filters and for keeping the drive stable. Instability and audible noise are caused from the following situations: HF resonances that are not already suppressed by filters MF resonances that are suppressed by filters but the filter bandwidths are too close to the closed loop bandwidth LF resonances that result when load observer is not applied with the recommended out-of-box settings LF resonances that result from classical instability Figure 4 - Identifying One LF Resonance Other considerations are given: Adaptive tuning Gain Stabilization is not recommended for vertical loads as detuning may cause load drops. 52 Rockwell Automation Publication 750-AT006A-EN-P - June 207

53 Product Features Chapter 2 Tracking Notch and Gain Stabilization 0:20 [AdptTune Config] = TqNFGainStb (3) In this mode, adaptive tuning applies the Tracking Notch function if necessary, followed by Gain Stabilization function if necessary. Parts of the control loop structure affected by this mode are highlighted below. Figure 42 - Tracking Notch Filter and Gain Stabilization Configuration Position Command s Feed Forwards s Adaptive Tuning Kvff Kaff Torque Loop Filters System Under Control P REF RN PI PI K J LL LP N K T J T s 2 P REG V REG Load Torque Torque Estimate Velocity Feedback Filter Fs Velocity Estimate Load Observer Position Feedback First, the adaptive tuning Tracking Notch function sets the actual torque notch filter parameters to suppress a HF resonance if one exists. Next, the adaptive tuning Gain Stabilization function sets the actual torque low pass filter parameters to suppress additional HF resonances if they exist. This is useful for suppressing more HF resonances than there are notch filters. Figure 43 - Identifying More HF Resonances than there are Notch Filters Rockwell Automation Publication 750-AT006A-EN-P - June

54 Chapter 2 Product Features Finally, if the system is unstable, the adaptive tuning Gain Stabilization functions incrementally detunes the position loop, velocity loop, and load observer gains to stabilize the system if necessary. A good example is when a MF resonance is identified. Figure 44 - Identifying One MF Resonance In this example, the torque notch filter is set to suppress the resonance if it is the only HF resonance or if it is the one with the largest magnitude. If not, the low pass filter is set to suppress it and any other HF resonances. Finally, the system is detuned if one or more of the following conditions exist: The torque notch filter is set to suppress the MF resonance, but its width is wide enough or its frequency is close enough to the closed loop bandwidth to cause instability. The torque low pass filter is set to suppress the MF resonance, but its bandwidth is close enough to the closed loop bandwidth to cause instability. Additional un-suppressed resonances are present. Notch Filters Overview Notch filters are represented by the following second order transfer function. G s K 2 s 2 + 2KZ 2 F s + 2 F 2 D K 2 s2 + 2Z 2 F s + 2 F 2 D D D F = =, K s 2 + 2Z 2 F s + 2 F 2 s 2 + 2Z 2 F s + 2 F 2 = F W W D Coefficients are defined in the following table. They feature variable adjustment of width, depth, and gain to allow for a wide range of applications. 54 Rockwell Automation Publication 750-AT006A-EN-P - June 207

55 Product Features Chapter 2 Table 0 - Variable Notch Filter Parameters Parameter Name Minimum Maximum Default F Notch Filter Frequency 0 Hz 3999 Hz 0 Hz K Notch Filter Gain Z W Notch Filter Width 0 = minimum width Z D Notch Filter Depth 0 = maximum depth There are four torque notch filters in the drive. There are two position reference notch filters, two velocity reference notch filters, and two process PID reference notch filters in the drive. These three sets of reference notch filters share common parameters, which allow all command signals to be effected uniformly. The torque notch filters suppress MF and HF motor side resonances and the reference notch filters suppress load side resonances. See Resonances on page 4 for more information on the various type of resonances and how to suppress them. Corresponding reference notch filter parameters are given. 0:942 [Ref NF Freq] Reference Notch Filter Frequency 0:943 [Ref NF Width] Reference Notch Filter Width 0:944 [Ref NF Depth] Reference Notch Filter Depth 0:945 [Ref NF Gain] Reference Notch Filter Gain 0:948 [Ref NF2 Freq] Reference Notch Filter 2 Frequency 0:949 [Ref NF2 Width] Reference Notch Filter 2 Width 0:950 [Ref NF2 Depth] Reference Notch Filter 2 Depth 0:95 [Ref NF2 Gain] Reference Notch Filter 2 Gain Corresponding torque notch filter parameters are given. 0:259 [Trq NF Freq] Torque Notch Filter Frequency 0:26 [Trq NF Width] Torque Notch Filter Width 0:263 [Trq NF Depth] Torque Notch Filter Depth 0:265 [Trq NF Gain] Torque Notch Filter Gain 0:269 [Trq NF2 Freq] Torque Notch Filter 2 Frequency 0:27 [Trq NF2 Width] Torque Notch Filter 2 Width 0:273 [Trq NF2 Depth] Torque Notch Filter 2 Depth 0:275 [Trq NF2 Gain] Torque Notch Filter 2 Gain 0:279 [Trq NF3 Freq] Torque Notch Filter 3 Frequency 0:28 [Trq NF3 Width] Torque Notch Filter 3 Width 0:283 [Trq NF3 Depth] Torque Notch Filter 3 Depth 0:285 [Trq NF3 Gain] Torque Notch Filter 3 Gain Rockwell Automation Publication 750-AT006A-EN-P - June

56 Chapter 2 Product Features 0:289 [Trq NF4 Freq] Torque Notch Filter 4 Frequency 0:29 [Trq NF4 Width] Torque Notch Filter 4 Width 0:293 [Trq NF4 Depth] Torque Notch Filter 4 Depth 0:295 [Trq NF4 Gain] Torque Notch Filter 4 Gain Configuration Notch filters have four modes of operation depending on the gain K setting, similar to lead-lag filters. The difference is that notch filters can only be configured as second order and lead-lag filters can only be configured as first order. See Lead Lag Filters on page 63 for more information. The filter is disabled when F = 0 Hz, regardless of the other settings. Configured as a Notch Filter (K = ) When K =, the notch filter operates as a second order notch filter. This is the recommended and default mode of operation. Also, adaptive tuning sets K = when tracking a notch filter. The variable width and depth functionality of the notch filter provides more flexibility in achieving high performance in some situations. The width can be increased, which allows one filter to suppress two HF resonances that are close to each other. An increased width is also effective in suppressing a resonance that slightly moves as mechanics slowly degrade over time. This reduces the need for retuning. The width of a torque notch filter can be decreased to suppress a MF resonance that is near the closed loop bandwidth. This minimizes the impact of the filter s phase lag on stability margin and reduces the need for detuning. The depth can be adjusted only as deep as required to suppress resonances with magnitudes large and small. This also conserves phase and allows for higher tuning performance. The recommended range for width Z W is from a minimum width of 0 to a maximum width of 4. The default width Z W = produces relatively good performance across applications. The recommended range for depth Z D is from a maximum depth of 0 to a minimum depth of < Z W. The default depth Z D = 0 provides maximum notch attenuation. Also, adaptive tuning sets Z D = 0 when tracking a notch filter. Note that the filter is off when K = and Z D = Z W or when F = 0. Furthermore, the filter becomes an inverse notch (generating resonance) when Z D > Z W. Notch filter width can be approximated in units of [Hz] as the range of frequencies impacted by the filter between -3dB points. 56 Rockwell Automation Publication 750-AT006A-EN-P - June 207

57 Product Features Chapter 2 Width Hz = 2FZ W z 2 0.5z 4 z 6, z = Z D Z W Notch filter depth can be calculated in units of [db] at the center frequency of the filter. Z D Depth db = 20log Z w Bode plots with differing width and depth notch filter settings for K= are shown in the following figures. Figure 45 - Variable Width Notch Filter Configurations Rockwell Automation Publication 750-AT006A-EN-P - June

58 Chapter 2 Product Features Figure 46 - Variable Depth Notch Filter Configurations Configured as a Second Order Low Pass Filter (K = 0) When K = 0, the notch filter is configured to operate as a second order lowpass filter with bandwidth F and damping Z W. Bode plots with differing lowpass filter settings for K = 0 are shown in the following figure. Figure 47 - Low Pass Filter Configurations 58 Rockwell Automation Publication 750-AT006A-EN-P - June 207

59 Product Features Chapter 2 Configured as a Second Order Lag Lead Filter (0 < K < ) When 0 < K <, the notch filter is configured to operate as a second order laglead filter. It has two user configurable poles (lag) at F with damping Z W and two user configurable zeros (lead) at F D = F/K with damping Z D. Bode plots with differing lag-lead filter settings for 0 < K < are shown in the following figure. Figure 48 - Lag-Lead Filter Configurations This configuration can be used to compensate for compliant load mechanics by placing two poles at the anti-resonant frequency and two zeros at the resonant frequency using the following calculations. K = R + F = K F R Z W = B T RJ M F Z D = Z w R + The following definitions are given. J M = 0:900 [Motor Inertia] in units of [kg m 2 ] R = 0:90 [Load Ratio] F R = an audible resonant frequency in units of [Hz] measured by a mobile application, such as ianalyzer Lite. Rockwell Automation Publication 750-AT006A-EN-P - June

60 Chapter 2 Product Features B T = the total mechanical friction in units of [Nm sec/rad]. This value is typically unknown and can be manually tuned. An initial value for a 580J motor is 0.0. However, using the filter in this way can make the drive more sensitive to disturbances. Furthermore, the load observer is recommended to compensate for load compliance and disturbances because it typically does a better job without having to determine these parameters. Configured as a Second Order Lead Lag Filter (K > ) When K >, the notch filter is configured to operate as a second order lead-lag filter. It has two user-configurable zeros (lead) at F D = F/K with damping Z D, and two user configurable poles (lag) at F with damping Z W. Bode plots with differing lead-lag filter settings for K > are shown in the following figure. Figure 49 - Lead-Lag Filter Configurations Initial recommended settings for damping are Z W = and Z D =. Lead-Lag filters have been used to boost velocity or acceleration loop bandwidth in order to improve performance. Low Pass Filters Low pass filters are represented by the following first order transfer function, where F is the filter bandwidth in units of [Hz]. G s 2 F = s + 2 F The filter is disabled when F = 0 Hz. 60 Rockwell Automation Publication 750-AT006A-EN-P - June 207

61 Product Features Chapter 2 These filters pass low frequency signals, but attenuate frequencies above F. Bode plots with differing bandwidth settings are shown in the following figure. Figure 50 - First Order Low Pass Filter The signal is attenuated 3dB at the filter bandwidth F, also known as the cut off frequency. The signal is attenuated 20 db/decade beyond the cut off frequency. Corresponding low pass filter parameters are given. 0:002 [c Vel Fb LPF BW] This parameter is an automatically calculated value for the primary velocity feedback low pass filter bandwidth. This parameter is applied to the drive when 0:905 [System C/U Select] = Calculated (0). It is calculated as a function of the following parameters and it affects the DMTC and 0:906 [System BW] values. 0:407 [Motor Poles] This parameter is the number of motor poles (p). It is calculated as follows: p = round (20 x [Motor NP Hertz] / [Motor NP RPM]) Primary Encoder Resolution The total resolution in edge counts per revolution (EPR) is specified by parameters on the primary feedback option card. Low-Resolution Example: Resolution = 024 pulses per revolution * 4 quadrature edge counts per pulse = 4096 EPR (2-bit). The low-resolution PPR comes directly from a parameter on the option card. When both A and B channels are selected for an incremental encoder, the edge count multiplier is 4. This value is the typical and default setting. When only channel A is selected, then the edge count multiplier is 2. Rockwell Automation Publication 750-AT006A-EN-P - June 207 6

62 Chapter 2 Product Features High-Resolution Example: Resolution = 024 pulses per revolution * 024 edge counts per pulse =,048,576 EPR (20-bit). For high-resolution devices, the overall resolution choices are 20-bit default or an optional 24-bit when the corresponding configuration bit is selected. 0:003 [u Vel Fb LPF BW] This parameter sets the primary velocity feedback low pass filter bandwidth. It is applied to the drive when 0:905 [System C/U Select] = User Entered (). 0:008 [c AltVelFbLPF BW] This parameter is an automatically calculated value for the alternate velocity feedback low pass filter bandwidth. It is applied to the drive when 0:905 [System C/U Select] = Calculated (0). It is calculated as a function of the following parameters and it affects the DMTC and 0:00 [Alt Fb GnScale] values. 0:407 [Motor Poles] This parameter is the number of motor poles (p). It is calculated as follows p = round (20 x [Motor NP Hertz] / [Motor NP RPM]) Alternate Encoder Resolution The total resolution in edge counts per revolution (EPR) is specified by parameters on the alternate feedback option card. Low-Resolution Example: Resolution = 024 pulses per revolution * 4 quadrature edge counts per pulse = 4096 EPR (2-bit). The low-resolution PPR comes directly from a parameter on the option card. When both A and B channels are selected for an incremental encoder, the edge count multiplier is 4. This value is the typical and default setting. When only channel A is selected, then the edge count multiplier is 2. High-Resolution Example: Resolution = 024 pulses per revolution * 024 edge counts per pulse =,048,576 EPR (20-bit). For high-resolution devices, the overall resolution choices are 20-bit default or an optional 24-bit when the corresponding configuration bit is selected. 0:009 [u AltVelFbLPF BW] This parameter sets the alternate velocity feedback low pass filter bandwidth. It is applied to the drive when 0:905 [System C/U Select] = User Entered (). 0:254 [c Trq LPF BW] This parameter is an automatically calculated value for the torque low pass filter bandwidth. It is applied to 0:256 [Trq LPF BW Act] when 0:905 [System C/U Select] = Calculated (0) and the adaptive tuning Gain Stabilization function is disabled. The value = max(kvp, Kop) * 5 when the load observer is enabled. The value = Kvp * 5 when the load observer is disabled. 0:255 [u Trq LPF BW] This parameter sets the torque low pass filter bandwidth. It is applied to 0:256 [Trq LPF BW Act] when 0:905 [System C/U Select] = User Entered () and the adaptive tuning Gain Stabilization function is disabled. 62 Rockwell Automation Publication 750-AT006A-EN-P - June 207

63 Product Features Chapter 2 Lead Lag Filters Lead-lag filters are represented by the following first order transfer function, where F is the filter bandwidth in units of [Hz] and K is the filter gain. 2 F Ks + 2 F s K G s = = K s + 2 F s + 2 F The filter is disabled when F = 0 Hz, regardless of the other settings. The filter has a DC gain of, a user configurable pole (lag) at F, and a userconfigurable zero (lead) at F/K. Bode plots with differing gain settings are shown in the following figure. Figure 5 - First Order Lead Lag Filter Configurations Rockwell Automation Publication 750-AT006A-EN-P - June

64 Chapter 2 Product Features Lead-lag filters have four modes of operation depending on the gain K setting, similar to notch filters. The difference is that lead-lag filters can only be configured as first order and notch filters can only be configured as second order. See Notch Filters on page 54 for more information. When K = 0, the lead-lag filter is configured to operate as a first order low-pass filter. When 0 < K <, the lead-lag filter is configured to operate as a first order lag-lead filter. It can be used to compensate for high frequency gain boost associated with compliant load mechanics. With a known load ratio R and resonant frequency F R, a pole can be placed at the antiresonant frequency and a zero can be placed at the resonant frequency using the following calculations: K = R + F = K F R However, use of the filter in this way can make the drive more sensitive to disturbances. Furthermore, the load observer is recommended to compensate for load compliance, and disturbances because it typically does a better job without having to determine these parameters. When K =, the lead-lag filter is off. This setting is the recommended and default mode of operation. When K >, the lead-lag filter is configured to operate as a first order lead-lag filter. This setting is used to compensate for undesirable dynamics caused by rate transitions between control loops. 64 Rockwell Automation Publication 750-AT006A-EN-P - June 207

65 Chapter 3 Out-of-Box Tuning Topic Page Out-of-Box Tuning 65 Gain Calculation 65 Recommended Default Settings 69 Single Knob Tuning 72 Is Further Tuning Required? 73 In this chapter, the Gain Calculation section describes how gains are calculated out-of-box and when parameters are updated by the drive. The Recommended Default Settings on page 69 describes how to configure a new drive. This outof-box tuning method often yields satisfactory performance where no further tuning intervention is required. Gain Calculation The term out-of-box refers to default control loop gain settings that are preconfigured in a new drive. Since the load is unknown at this point, the motor is assumed to be unloaded, the load ratio R = 0, and the load coupling is compliant. This value also applies to when the load is known to be compliant. This chapter pertains to when the load ratio R = 0. However, when the load is known or an autotune total inertia test has been performed to determine the load ratio, the control loop gains are configured for a load ratio R > 0, and the load coupling is rigid. See Auto Tuning on page 75 for more information. These settings are the primary difference affecting out-of-box and autotune rules. Thus, the term out-of-box implies a compliant load with R = 0 and the term autotune implies a rigid load with R > 0. In a new drive configuration or when the load is unknown, 0:90 [Load Ratio] = 0, but the actual load ratio R > 0. As a result, the torque scaler is lower than what is required and it must be increased by a factor of R+. Rockwell Automation Publication 750-AT006A-EN-P - June

66 Chapter 3 Out-of-Box Tuning This load disconnect has the following effects: The effectiveness of acceleration feed forward is reduced by a factor of R+. The torque scaler borrows a factor of R+ from Kvp, lowering the actual velocity loop bandwidth by a factor of R+ relative to the [VReg Kp] that is applied to the control loops. Reducing Kvp causes the spacing between it and Kvi to become smaller, which reduces the velocity loop damping associated with Kvi. Reducing Kvp also causes the loop spacing between Kvp and Kpp to become smaller, which reduces the position loop damping associated with this reduction in loop spacing. Figure 52 - Effects of Unknown Load R = 0 Loop Spacing Damping Velocity Loop Actual Velocity Loop BW 2 K VP K AFF J K M T R+ System Under Control Motor Electrical K T J M Motor and Load Mechanics ( R ) s 2 Motor Position s 2 K VI Integrator Damping Torque Scalar (K J ) Load Torque To increase damping when Kvp is artificially lowered, a few different methods are applied based on the load observer mode. Relevant variables are defined, followed by descriptions of each method based on the load observer mode. T BW = Torque Loop Bandwidth Z = 0:907 [System Damping] = System Damping System BW = 0:906 [System BW] = System Bandwidth[Hz] Kpp = 0:754 [c PReg Kp] = Position Loop Bandwidth[Hz] Kpi = 0:756 [c PReg Ki] = Position Integrator Bandwidth [Hz] Kvp = 0:955 [c VReg Kp] = Velocity Loop Bandwidth [Hz] Kvi = 0:957 [c VReg Ki] = Velocity Integrator Bandwidth [Hz] Kop = 0:202 [c LdObs Kp] = Load Observer Bandwidth [Hz] Koi = 0:2024 [c LdObs Ki] = Load Observer Integrator Bandwidth [Hz] LPF = 0:254 [c Trq LPF BW] = Torque Low Pass Filter Bandwidth [Hz] Load Observer Disabled When 0:90 [Load Ratio] = 0 and 0:2020 [LdObs Mode] = Disabled (0), the position loop gains are decreased by an additional factor of 0. This reduction increases the loop spacing to allow enough room for Kvp when it is artificially lowered. Kvi is not decreased by an additional factor of 0 because it provides improved stiffness without the risk of instability. The default damping factor is Z =, which produces a loop spacing of 4Z 2 = Rockwell Automation Publication 750-AT006A-EN-P - June 207

67 Out-of-Box Tuning Chapter 3 Figure 53 - Out-of-Box Gain Relationships for Load Observer Disabled 4Z T BW 2 System BW K PP 40Z 2 K VP K OP 5 LPF 0 4Z 2 0 K PI K VI K OI Load Observer Only or Acceleration Feedback When 0:90 [Load Ratio] = 0 and 0:2020 [LdObs Mode] = LdObs Only () or 0:2020 [LdObs Mode] = Accel Fdbk (4), there is no need to decrease the position loop gains by an additional factor of 0 because the load observer accounts for the unknown load to maintain that Kvp is not artificially lowered. Also, Kvi is disabled because the load observer operates as a good velocity integrator. Load Observer Only mode is recommended for velocity control applications. Figure 54 - Out-of-Box Gain Relationships for Load Observer Only or Acceleration Feedback 4Z T BW 2 System BW K PP 4Z 2 K VP K OP 5 LPF K PI K VI K OI Velocity Estimate Only When 0:90 [Load Ratio] = 0 and 0:2020 [LdObs Mode] = Vel Est Only (3), the load observer bandwidth is increased by 4Z 2 because the velocity estimate removes enough phase lag from the closed loop to do so while maintaining stability. Also Kvi is enabled to remove any velocity steady state error that is produced by this load observer mode setting. Rockwell Automation Publication 750-AT006A-EN-P - June

68 Chapter 3 Out-of-Box Tuning Figure 55 - Out-of-Box Gain Relationships for Velocity Estimate Only 4Z T BW 2 System BW K PP 4Z 2 K VP 2 4Z K OP 5 LPF 0 4Z 2 0 K PI K VI K OI Load Observer with Velocity Estimate When 0:90 [Load Ratio] = 0 and 0:2020 [LdObs Mode] = LdObs VelEst (2), the load observer bandwidth is increased by 4Z 2 because the velocity estimate removes enough phase lag from the closed loop to do so while maintaining stability. Also, Kvi is disabled because load observer operates as a good velocity integrator. This mode is recommended for position control applications. Load Observer with Velocity Estimate automatically accounts for the arbitrary load disconnect created by the load, which forces it to function like a mechanically disconnected motor. As a result, standard 4Z 2 spacing sufficient for controlling a mechanically disconnected motor is applied. Figure 56 - Out-of-Box Gain Relationships for Load Observer with Velocity Estimate 4Z T BW 2 System BW K PP 4Z 2 K VP 2 4Z K OP 5 LPF K PI K VI K OI Other considerations are given: When the torque low-pass filter is enabled, set the bandwidth to greater than 5 times Kvp or Kop, whichever is larger. This value prevents additional phase lag created by the low-pass filter from being introduced into the system, which causes instability. When R is entered higher than the actual load ratio, Kvp is artificially lower than its actual bandwidth. When R is entered lower than the actual load ratio, Kvp is artificially higher than its actual bandwidth. 68 Rockwell Automation Publication 750-AT006A-EN-P - June 207

69 Out-of-Box Tuning Chapter 3 Recommended Default Settings This section describes how to configure a new drive. This method often yields satisfactory performance where no further tuning intervention is required. Follow these steps to configure the drive for relatively high performance out of the box.. Enter the following parameters for the motor that is connected to the drive. 0:400 [Motor NP Volts] Enter the motor nameplate rated volts. 0:40 [Motor NP amps] Enter the motor nameplate rated full load amps. 0:402 [Motor NP Hertz] Enter the motor nameplate frequency in units of [Hz]. 0:403 [Motor NP RPM] Enter the motor nameplate speed in units of [RPM]. 0:405 [Mtr NP Pwr Units] Enter the motor nameplate power units [0 = HP, = kw]. 0:406 [Motor NP Power] Enter the motor nameplate power in the respective units from 0:405 [Mtr NP Pwr Units]. 0:407 [Motor Poles] Enter the number of motor poles (p). It is calculated as follows: p = round(20 x [Motor NP Hertz] / [Motor NP RPM]) 0:900 [Motor Inertia] Enter the motor nameplate inertia or from the motor data sheet in units of [kg*m 2 ]. Divide [lb.*ft 2 ] or [WK 2 ] by to convert to [kg*m 2 ]. If data is not available, use the following equation to approximate 0:900 [Motor Inertia]. J M = Motor HP/250 * (Motor HP/500 + ) Divide [kw] by.34 to convert to [HP]. 0:000 [Pri Vel Fb Sel] Enter the 2-digit port location followed by the 4-digit parameter number of the primary feedback device. 0:006 [Alt Vel Fb Sel] Enter the 2-digit port location followed by the 4-digit parameter number of the alternate feedback device. Primary Encoder Resolution The total resolution in edge counts per revolution (EPR) is specified by parameters on the primary feedback option card. Low-Resolution Example: Resolution = 024 pulses per revolution * four quadrature edge counts per pulse = 4096 EPR (2-bit). The lowresolution PPR comes directly from a parameter on the option card. When both A and B channels are selected for an incremental encoder, the edge count multiplier is 4. This value is the typical and the default setting. When only channel A is selected, then the edge count multiplier is 2. Rockwell Automation Publication 750-AT006A-EN-P - June

70 Chapter 3 Out-of-Box Tuning High-Resolution Example: Resolution = 024 pulses per revolution * 024 edge counts per pulse =,048,576 EPR (20-bit). For highresolution devices, the overall resolution choices are 20-bit default or an optional 24-bit when the corresponding configuration bit is selected. Alternate Encoder Resolution The total resolution in edge counts per revolution (EPR) is specified by parameters on the alternate feedback option card. 2. Set the following parameters according to your application. 0:30 [PsnVelTrq Mode A] = Position, velocity, or torque mode 0:00[Vel Fb Taps] = 0 for high-resolution encoders for low-resolution encoders 0:007 [Alt Vel Fb Taps] = 0 for high-resolution encoders for low-resolution encoders 0:2020 [LdObs Mode] = Disabled (0) for torque mode applications LdObs Only () for velocity mode applications LdObs VelEst (2) for position mode applications 0:445 [VCL CReg BW] = 25 when 0:425 [PWM Frequency] =.33 khz 250 when 0:425 [PWM Frequency] = 2 khz 375 when 0:425 [PWM Frequency] = 4 khz 3. Verify or set the following parameters to the values shown. 0:65 [Pri MtrCtrl Mode] = Induction FV (3) 0:229 [Regen Power Lim] = -200% 0:444 [VCL CReg C/U Sel] = Calculated (0) 0:50 [MtrParam C/U Sel]= Calculated (0) 0:90 [Load Ratio] = 0 0:902 [Load Coupling] = Compliant () 0:905 [System C/U Sel] = Calculated (0) 0:907 [System Damping] = 0:20 [AdptTune Config] = Tq NF Only () 0:2 [Trq NF Threshold] = 5 0:22 [Trq NF Freq LLim] = 5 Hz 0:23[Trq NF Freq HLim] = 2000 Hz 0:24 [Trq NF WidthMin] = :25 [Trq NF WidthMax] = :26 [Trq LPF BW LLim] = 4 Hz 0:29 [AdptTuneMinScale] = Run 0:90 [Autotune] = Direction (). 70 Rockwell Automation Publication 750-AT006A-EN-P - June 207

71 Out-of-Box Tuning Chapter 3 5. Measure the motor electrical parameters. a. Set 0:50 [MtrParam C/U Sel] = User Entered (). You can first set it to copy the calculated parameter values over to the user entered parameters. b. Run 0:90 [Autotune] = Rotate MtrID (3) to measure the motor electrical parameters. It initiates motion and rotates the load. c. If you cannot initiate motion to rotate the load, then run 0:90 [Autotune] = Static MtrID (2) to measure the motor electrical parameters. 6. Run 0:90 [Autotune] = InertiaMotor (4) with the load disconnected to measure 0:900 [Motor Inertia]. This step is only an option if the load can be disconnected to run the test. Otherwise, skip this step. 7. Run 0:90 [Autotune] = BW Calc (6). This mode calculates the following parameters based on motor inertia, load ratio, and motor parameters entered in previous steps. Motion is not initiated by this mode. 0:906 [System BW] 0:00 [Alt Fb GnScale] 0:392 [Max Speed Fwd] 0:393 [Max Speed Rev] 0:898 [Vel Limit Pos] 0:899 [Vel Limit Neg] 0:965 [Accel Limit Pos] 0:966 [Accel Limit Neg] 0:2083 [Torque Limit Pos] 0:2084 Torque Limit Neg] 8. Run the drive and adjust 0:906 [System BW] if necessary. It initiates motion and rotates the load. Decreasing 0:906 [System BW] stabilizes the system and increasing it improves performance. Typically, high gain results in a quicker response time, but excessive gain causes system instability. The following parameters are automatically calculated as functions of system bandwidth, load observer mode, and load ratio. 0:754 [c PReg Kp] 0:756 [c PReg Ki] 0:955 [c VReg Kp] 0:957 [c VReg Ki] 0:202 [c LdObs Kp] 0:2024 [c LdObs Ki] 0:254 [c Trq LPF BW] The following parameters are automatically calculated as functions of system bandwidth, alternate feedback gain scale, and motor parameters. 0:002 [c Vel Fb LPF BW] 0:008 [c AltVelFbLPF BW] Rockwell Automation Publication 750-AT006A-EN-P - June 207 7

72 Chapter 3 Out-of-Box Tuning 9. If necessary, reduce the acceleration and deceleration limits to meet application requirements and help protect the drive and motor from overload. a. For velocity mode applications, reduce the velocity ramp acceleration and deceleration times in the velocity reference. b. For position mode applications, reduce the acceleration and deceleration times in the PTP planner. c. As a last resort, reduce the acceleration limits in the velocity regulator. When Load Ratio = 0, these limits are set to their maximum value, which provides the best performance for a motor under no-load conditions. However, the motor is loaded and may not be able to accelerate as fast. As a result, you may have to reduce the acceleration and deceleration limits to meet application requirements. 0. Optional: Use the adaptive tuning Tracking Notch function to set torque notch filters a. Copy the torque notch filter actual values (frequency, gain, width, and depth) to the torque notch filter 2 parameters. b. If adaptive tuning finds a second resonance, copy the torque notch filter actual values to the torque notch filter 3 parameters. c. If adaptive tuning finds a third resonance, copy the torque notch filter actual values to the torque notch filter 4 parameters. d. If adaptive tuning finds a fourth resonance, then the tracking notch filter has already set the torque notch filter actual values to attenuate this resonance. Single Knob Tuning 0:906 [System BW] can be used as a single tuning knob when 0:905 [System C/U Select] = Calculated (0). Decreasing it stabilizes the system and increasing it improves performance. The following parameters are automatically calculated as this parameter is adjusted. 0:754 [c PReg Kp] 0:756 [c PReg Ki] 0:955 [c VReg Kp] 0:957 [c VReg Ki] 0:202 [c LdObs Kp] 0:2024 [c LdObs Ki] 0:254 [c Trq LPF BW] 72 Rockwell Automation Publication 750-AT006A-EN-P - June 207

73 Out-of-Box Tuning Chapter 3 Is Further Tuning Required? Here are some observations that indicate the drive is producing satisfactory performance without additional tuning: Visibly smooth non-oscillatory velocity feedback from a smooth cycle profile Little to no audible noise produced during and after a commanded motion Position and/or velocity errors are repeatable and within the application requirements The reference move type keeps the drive and motor within their thermal limits If the load does not respond as intended, consider these factors before tuning: Investigate proper motor and drive sizing in Online Motion Analyzer. You may also want to compare various design options for gear ratio, load size, coupling configuration, high versus low-resolution feedback devices, reference move types, and so on. Simulate the load in Online Motion Analyzer to conclude how the desired motion can be achieved. Problems are often minimized by creating a more direct and stiff coupling between the motor and load. Quality mechanical components help to achieve this. If the load requires additional tuning to optimize performance, continue on to Auto Tuning on page 75. The following tuning flowchart is given. Rockwell Automation Publication 750-AT006A-EN-P - June

74 Chapter 3 Out-of-Box Tuning Figure 57 - Tuning Flowchart 74 Rockwell Automation Publication 750-AT006A-EN-P - June 207

75 Chapter 4 Auto Tuning Topic Page Auto Tuning 75 General Modes 76 Motor Electrical Parameters 76 Inertia Tests 80 Gain Calculation 82 Is Further Tuning Required? 86 The Autotune function is used to measure motor characteristics. It is composed of several individual tests, each of which is intended to identify one or more motor parameters. These tests require motor nameplate information to be entered into the drive parameters. Although some of the parameter values can be changed manually, measured values of the motor parameters can provide the best performance. Autotune performs four main types of tests that are determined by 0:90 [Autotune] and are summarized below. General Modes Autotune defaults to Ready and performs a direction test. 0:90 [Autotune] = Ready (0) 0:90 [Autotune] = Direction () Motor Electrical Parameters Autotune performs static and dynamic tests to measure motor electrical parameters. 0:90 [Autotune] = Static MtrID (2) 0:90 [Autotune] = Rotate MtrID (3) Inertia Tests Autotune performs a dynamic bump test that initiates momentary motor rotation to measure inertia. 0:90 [Autotune] = InertiaMotor (4) 0:90 [Autotune] = InertiaTotal (5) Gain Calculation Autotune calculates control loop gains, filter bandwidths, and dynamic limits. 0:90 [Autotune] = BW Calc (6) 0:90 [Autotune] = JMtr BW Calc (7) 0:90 [Autotune] = JTotalBWCalc (8) The sections that follow then go into more detail. Rockwell Automation Publication 750-AT006A-EN-P - June

76 Chapter 4 Auto Tuning General Modes This section covers the general Autotune modes. 0:90 [Autotune] = Ready (0) Indicates that the Autotune function is available. Autotune returns to Ready (0) after a test from one of the other settings. After it returns to Ready (0), another start transition is required to operate the drive in Normal mode. 0:90 [Autotune] = Direction () Enables the Autotune function to perform the Direction Test. After selecting this value, you must issue a start command to begin the test. Motor Electrical Parameters Autotune Electrical tests impact motor electrical parameters that are used in the current loop. Calculation of user entered parameters are selected by 0:50 [MtrParam C/U Sel] and 0:444 [VCL CReg C/U Sel] as described in the sections that follow. Configuration 0:50 [MtrParam C/U Sel] allows either calculated or user entered parameters to be applied to motor electrical gains. It also allows you to transfer calculated parameter values to user entered parameter values so that you can use them as a starting point for manual tuning, if desired. The following parameters are applied to motor electrical gains when 0:50 [MtrParam C/U Sel] = Calculated. 0:489 [c Slip RPM atfla] 0:5 [c IM Stator Res] 0:54 [c IM Leakage L] 0:57 [c Flux Cur Ref ] 0:520 [c EncLs AngCmp] 0:523 [c IM StatResComp] These parameters are calculated based on motor nameplate values 0:400 [Motor NP Volts], 0:40 [Motor NP Amps], 0:402 [Motor NP Hertz], 0:403 [Motor NP RPM], 0:405 [Mtr NP Pwr Units], P406 [Motor NP Power], and 0:407 [Motor Poles]. The calculations are automatically updated when one or more motor nameplate values change. The following parameters are applied to the controller when 0:50 [MtrParam C/U Sel] = User Entered. 0:490 [u Slip RPM atfla] 0:52 [u IM Stator Res] 0:55 [u IM Leakage L] 76 Rockwell Automation Publication 750-AT006A-EN-P - June 207

77 Auto Tuning Chapter 4 0:58 [u Flux Cur Ref ] 0:52 [u EndLs AngCmp] 0:524 [IM StatResComp] The static and rotate autotune results go into these parameters. You can enter values for these parameters directly, or you can run Autotune to overwrite them when 0:90 [Autotune] = Static MtrID (2) or when 0:90 [Autotune] = Rotate MtrID (3). The following diagram shows how the Motor Parameter C/U Selection impacts various parameters. Figure 58 - Motor Parameter C/U Selector These parameters only apply when 0:65 [Pri MtrCtrl Mode] = Induction SV (), Induct Econ (2), or Induction FV (3). 0:444 [VCL CReg C/U Select] allows either calculated or user entered parameters to be applied to current loop gains. It also allows you to transfer calculated parameter values to user entered parameter values so you can use them as a starting point for manual tuning if desired. The following parameters are applied as current loop gains when 0:444 [VCL CReg C/U Select] = Calculated. 0:446 [c VCL CReg Kp] 0:448 [c VCL CReg Ki] These parameters are calculated based on 0:445 [VCL CReg BW], 0:5 [c IM Stator Res], and 0:54 [c IM Leakage L] when 0:50 [MtrParam C/U Sel] = Calculated (0) or they are based on 0:445 [VCL CReg BW], 0:52 [u IM Stator Res], and 0:55 [u IM Leakage L] when 0:50 [MtrParam C/U Sel] = User Entered (). The calculations are automatically updated when one or more motor nameplate values change. Rockwell Automation Publication 750-AT006A-EN-P - June

78 Chapter 4 Auto Tuning The following parameters are applied to the controller when 0:444 [VCL CReg C/U Select] = User Entered. 0:447 [u VCL CReg Kp] 0:449 [u VCL CReg Ki] This selection allows you to enter values for these parameters directly. The following diagram shows how the two selectors impact various parameters and ultimately current loop regulator gains. Figure 59 - Current Regulator C/U Selector Modes This section covers the Autotune modes used to measure motor electrical parameters. 0:90 [Autotune] = Static MtrID (2) Enables the Autotune function to perform the Static Motor ID Test. After selecting this value, you must issue a start command to begin the test. Use this test when the motor cannot rotate freely or when it is already coupled to the load. The static test does not initiate momentary motor rotation to measure motor electrical parameters. Instead, it only measures electrical parameters that do not generate motor movement. The rest of the parameters are calculated from a lookup table. A resistance and inductance test updates the following parameters. 0:52 [u IM Stator Res] 0:55 [u IM Leakage L] The following values are also calculated from a lookup table. However, they are measured more accurately during a rotate test. 0:490 [u Slip RPM atfla] 0:58 [u Flux Cur Ref ] 0:52 [u EndLs AngCmp] 0:524 [IM StatResComp] 78 Rockwell Automation Publication 750-AT006A-EN-P - June 207

79 Auto Tuning Chapter 4 0:90 [Autotune] = Rotate MtrID (3) Enables the Autotune function to perform the Rotating Motor ID Test. After selecting this value, you must issue a start command to begin the test. Use this test when the motor can rotate freely and when it is not coupled to the load. The rotate test initiates momentary motor rotation to measure motor electrical parameters. This test is more accurate than the static test. During the rotate test, a static test is first performed to calculate the following parameters. 0:52 [u IM Stator Res] 0:55 [u IM Leakage L] Then the rotate test calculates the following parameters. 0:490 [u Slip RPM atfla] 0:58 [u Flux Cur Ref ] 0:52 [u EndLs AngCmp] 0:524 [IM StatResComp] In order for these parameters to be applied, set 0:50 [MtrParam C/U Sel] = User Entered (). The following diagram shows how Autotune Electrical modes impact various parameters. Figure 60 - Autotune Motor Electrical Parameters Run Static or Rotate Autotune when an LCL filter, dvdt filter, sine-wave filter, or line reactor is placed between the drive and motor. Rockwell Automation Publication 750-AT006A-EN-P - June

80 Chapter 4 Auto Tuning The following parameters allow you to set dynamic limits on the move profile that is used during rotate tests. 0:9 [Autotune Psn Lim] 0:93 [Autotune Trq Lim] IMPORTANT If rotate tune for a Sensor-less Vector mode is used, uncouple the motor from the load or results can be invalid. In Flux Vector mode, either a coupled or uncoupled load produces valid results. Caution must be used when the load is connected to the motor shaft and an Autotune is performed. Rotation during the tuning process may exceed machine limits. Inertia Tests Autotune inertia tests are used to measure the following parameters accurately. 0:900 [Motor Inertia] 0:90 [Load Ratio] IMPORTANT In legacy PF755 drives, total inertia required accurate measurement using autotune. In this drive, total inertia is calculated from 0:900 [Motor Inertia], which now requires accuracy. As a result, a total inertia autotune test is often not necessary. 0:900 [Motor Inertia] Four ways to determine this value are given in order from most to least effective. Data Sheet: Enter a value from the motor nameplate or data sheet in units of [kg*m 2 ]. Divide [lb.*ft 2 ] or [WK 2 ] by to convert to [kg*m 2 ]. Measured: Run Autotune with 0:90 [Autotune] = InertiaMotor (4) or JMtr BW Calc (7) with the load disconnected to dynamically measure motor inertia. This method is only an option if the load can be disconnected to run this test. Estimated: If data is not available or a motor inertia test is not possible, use the following equation to approximate 0:900 [Motor Inertia] based on motor nameplate horsepower (HP): J M = HP/250*(HP/500+) Divide [kw] by.34 to convert to [HP]. Default Value: This value is based on a motor power rating equal to the drive power rating. See Inertia on page 8 for more information. 80 Rockwell Automation Publication 750-AT006A-EN-P - June 207

81 Auto Tuning Chapter 4 0:90 [Load Ratio] and 0:900 [Motor Inertia] are used to calculate the torque scaler K J, an internal parameter that compensates for the effects of inertia and affects overall tuning. 0:90[Load Ratio] is also used to calculate controller gains, however the effectiveness of this calculation is largely based on the accuracy of 0:900 [Motor Inertia]. See Gain Calculation on page 82 for more information. Figure 6 - Autotune Inertia Parameters When Autotune performs an inertia test, a momentary tuning torque profile is applied to the motor and the acceleration and deceleration times are measured. As a result, the motor shaft moves, which ramps the speed up and down. The following parameters allow you to set dynamic limits on the torque profiles that are used during the rotate tests. 0:9 [Autotune Psn Lim] 0:93 [Autotune Trq Lim] IMPORTANT Rotation during the tune process may exceed machine limits. Caution must be used when connecting and disconnecting the load and when setting these autotune limits, so that the autotune motor rotation does not damage the machine. Note that systems with mechanical restrictions or travel limits may not complete the autotune test. As a result, these parameters may require readjustment. 0:90 [Autotune] = InertiaMotor (4) Enables the Autotune function to perform the Motor Inertia Test. After selecting this value, you must issue a start command to begin the test. This test initiates momentary motor rotation to measure and update 0:900 [Motor Inertia]. Also, 0:90 [Load Ratio] is set to zero. IMPORTANT This test is only an option if the load can be disconnected to run the test. It must be performed with the load disconnected from the motor to generate accurate results. Otherwise, the motor inertia is calculated as the total inertia of the motor and load together. 0:90 [Autotune] = InertiaTotal (5) Enables the Autotune function to perform the Total Inertia Test. After selecting this value, you must issue a start command to begin the test. This test initiates momentary rotation of the motor and load to measure total inertia and calculate 0:90 [Load Ratio]. Rockwell Automation Publication 750-AT006A-EN-P - June 207 8

82 Chapter 4 Auto Tuning When possible, perform Autotune at the point of lowest mechanical inertia on variable inertia loads, such as winders. IMPORTANT This test is only an option if motion can be initiated to rotate the load during the test. It must be performed with the load connected to the motor to generate accurate results. Otherwise the test measures motor inertia instead of total inertia and incorrectly uses it to calculate load ratio. Also, the calculation assumes 0:900 [Motor Inertia] is accurate. If you get an error that results from a negative load ratio, the motor inertia is incorrect and must be reduced. The total inertia test is an effective way to determine load ratio for existing applications where R > 0 is known or manually calculated. However, a load ratio R = 0 is often all that is required to achieve relatively high performance and this test is often not required. As a result, it is strongly advised to first run the drive with the recommended default settings before this test is conducted, as it typically produces desired performance. See Recommended Default Settings on page 69 for more information. Then run this test if the load requires additional tuning to further optimize performance. Gain Calculation The term out-of-box refers to default control loop gain settings that are preconfigured in a new drive. Since the load is unknown at this point, the motor is assumed to be unloaded, the load ratio R = 0, and the load coupling is compliant. This value also applies to when the load is known to be compliant. See Out-of-Box Tuning on page 65 for more information when R = 0. However, when the load is known or an autotune total inertia test has been performed to determine the load ratio, the control loop gains are configured for a load ratio R > 0 and the load coupling is rigid. This chapter pertains to when the load ratio R > 0. These settings are the primary difference affecting out-of-box and autotune rules. Thus, the term out-of-box implies a compliant load with R = 0 and the term autotune implies a rigid load with R > 0. Modes The following Autotune modes perform Bandwidth Calculations. Some modes perform inertia tests before bandwidth calculations because bandwidth calculations use 0:900 [Motor Inertia] and 0:90 [Load Ratio] in their computation. 0:90 [Autotune] = BW Calc (6) Enables the Autotune function to perform Bandwidth Calculations, which means that the drive calculates control loop gains and dynamic limits. When R = 0, the gains are calculated according to Out-of-Box Tuning Gain Calculation on page 65. When R > 0, 82 Rockwell Automation Publication 750-AT006A-EN-P - June 207

83 Auto Tuning Chapter 4 the gains are calculated according to Auto Tuning Gain Calculation on page 82. 0:90 [Autotune] = JMtr BW Calc (7) Enables the Autotune function to perform a Motor Inertia Test followed by the Bandwidth Calculations. After selecting this value, you must issue a start command to begin the test. Perform this test with the load disconnected from the motor. Since the Motor Inertia test sets R = 0, the gains are calculated according to Out-of-Box Tuning Gain Calculation on page 65. This test works well for most applications. 0:90 [Autotune] = JTotalBWCalc (8) Enables the Autotune function to perform a Total Inertia Test followed by the Bandwidth Calculations. After selecting this value, you must issue a start command to begin the test. Perform this test with the load connected to the motor. Since the Total Inertia Test sets R > 0, the gains are calculated according to Gain Calculation on page 82. This test works well for high performance rigid loads. Configuration Autotune bandwidth calculations consist of two things:. The following dynamic limits are calculated based on 0:900 [Motor Inertia], 0:90 [Load Ratio], and motor nameplate parameters. 0:392 [Max Speed Fwd] This parameter is the forward speed limit that is used by the position PTP planner. 0:393 [Max Speed Rev] This parameter is the reverse speed limit that is used by the position PTP planner. 0:898 [Vel Limit Pos] This parameter is the positive velocity limit that is placed on the velocity reference. 0:899 [Vel Limit Neg] This parameter is the negative velocity limit that is placed on the velocity reference. 0:965 [Accel Limit Pos] This parameter is the positive acceleration limit on the velocity regulator output. 0:966 [Accel Limit Neg] This parameter is the negative acceleration limit on the velocity regulator output. 0:2083 [Torque Limit Pos] This parameter is the positive limit that is placed after the torque reference filters. 0:2084 [Torque Limit Neg] This parameter is the negative limit that is placed after the torque reference filters. 2. 0:906 [System BW] and 0:00 [Alt Fb Gain Scale] are calculated, which triggers an automatic calculation of control loop gains. The following parameters are used to calculate 0:906 [System BW], 0:00 [Alt Fb Gain Scale], and the control loop gains. Rockwell Automation Publication 750-AT006A-EN-P - June

84 Chapter 4 Auto Tuning 0:907 [System Damping] This parameter sets system damping (Z) which adjusts position, velocity, and torque loop spacing of the calculated control loop gains and load observer gains. It also adjusts the integrator spacing of the calculated gain parameters to generate the required responsiveness. 0:902 [Load Coupling] This parameter reflects the type of mechanical coupling between the motor and load. It adjusts calculated control loop gains during an autotune bandwidth calculation test. 0:902 [Load Coupling] = Rigid (0) is where the load consists of few mechanical components with direct connection to the motor shaft. The components are high performance, have no flex or twist, and misalignment is not likely to occur. This setting is for high performance machines. A loop spacing of 4Z 2 is sufficient for rigid loads. 0:902 [Load Coupling] = Compliant () is a non-rigid load where position misalignment, backlash, and flexing of couplings, gearboxes, belts, and shafts can occur which creates an indirect connection of the load to the motor shaft. This setting is common for most machines. As a result, all gains are reduced by an additional factor of (R+) to achieve a conservative stability margin over a large class of possible systems. This setting can lead to a low system bandwidth which can then be manually increased to achieve higher performance. 0:2020 [LdObs Mode] Configures the load observer feature. The following recommended settings are given based on control mode. 0:2020 [LdObs Mode] = Disabled (0) for torque mode applications 0:2020 [LdObs Mode] = LdObs Only () for velocity mode applications 0:2020 [LdObs Mode] = LdObs VelEst (2) for position mode applications Load Observer mode affects the automatic calculation of calculated control loop gains and load observer gains. A few different methods are applied based on the load observer mode. Relevant variables are defined, followed by descriptions of each method based on load observer mode. T BW = Torque Loop Bandwidth Z = 0:907 [System Damping] = System Damping System BW = 0:906 [System BW] = System Bandwidth[Hz] Kpp = 0:754 [c PReg Kp] = Position Loop Bandwidth[Hz] Kpi = 0:756 [c PReg Ki] = Position Integrator Bandwidth [Hz] Kvp = 0:955 [c VReg Kp] = Velocity Loop Bandwidth [Hz] Kvi = 0:957 [c VReg Ki] = Velocity Integrator Bandwidth [Hz] Kop = 0:202 [c LdObs Kp] = Load Observer Bandwidth [Hz] Koi = 0:2024 [c LdObs Ki] = Load Observer Integrator Bandwidth [Hz] 84 Rockwell Automation Publication 750-AT006A-EN-P - June 207

85 Auto Tuning Chapter 4 LPF = 0:254 [c Trq LPF BW] = Torque Low Pass Filter Bandwidth [Hz] Load Observer Disabled or Velocity Estimate Only When 0:90 [Load Ratio] > 0 and 0:2020 [LdObs Mode] = Disabled (0) or 0:2020 [LdObs Mode] = Vel Est Only (3), Kvi is enabled to remove any velocity steady state error that is produced by these load observer mode settings. When R > 0, the torque scaler K J presents a high gain in the control loop forward path. As a result, the load observer bandwidth equals the velocity loop bandwidth to maintain stability. Figure 62 - Autotune Gain Relationships for Load Observer Disabled or Velocity Estimate Only ( Rigid) 2 4Z T BW ( Compliant) 2 4Z ( R ) K PP 4Z 2 System BW K VP K OP 5 LPF 0 4Z 2 0 K PI K VI K OI Load Observer Only, Load Observer with Velocity Estimate, or Acceleration Feedback When 0:90 [Load Ratio] > 0 and 0:2020 [LdObs Mode] = LdObs Only () or 0:2020 [LdObs Mode] = LdObs VelEst (2) or 0:2020 [LdObs Mode] = Accel Fdbk (4), Kvi is disabled because load observer operates as a good velocity integrator. Figure 63 - Autotune Gain Relationships for Load Observer Only, Load Observer with Velocity Estimate, or Acceleration Feedback ( Rigid) 2 4Z T BW ( Compliant) 2 4Z ( R ) K PP 4Z 2 System BW K VP K OP 5 LPF K PI K VI K OI Rockwell Automation Publication 750-AT006A-EN-P - June

86 Chapter 4 Auto Tuning The following diagram shows how autotune inertia and bandwidth tests interact with various parameters to affect the control loop gains. Autotune updates the parameters in red, which then triggers automatic calculations upon any change of their input parameters. If inertia tests are performed, always initiate them before running bandwidth calculations. However, inertia tests are not always required before running bandwidth calculations, for example, when motor inertia is known and R = 0. Figure 64 - Autotune Bandwidth Calculations Is Further Tuning Required? Here are some observations to indicate that the drive does not require additional tuning: Visibly smooth non-oscillatory velocity feedback from a smooth cycle profile Little to no audible noise produced during and after a commanded motion Position and/or velocity errors are repeatable and within the application requirements The reference move type keeps the drive and motor within their thermal limits If the load does not respond as intended, consider these factors before tuning: Investigate proper motor and drive sizing in Online Motion Analyzer. You may also want to compare various design options for gear ratio, load size, coupling configuration, high versus low-resolution feedback devices, reference move types, and so on. Simulate the load in Online Motion Analyzer software to conclude that the desired motion can be achieved. Problems are often minimized by creating a more direct and stiff coupling between the motor and load. Quality mechanical components, such as couplings, gearboxes, actuators, and guides help to achieve this. If the load requires additional tuning, continue on to Manual Tuning on page Rockwell Automation Publication 750-AT006A-EN-P - June 207

87 Chapter 5 Manual Tuning Topic Page Manual Tuning 87 Initial Configuration 88 Tune the Current Loop (Optional) 88 Tune the Velocity Loop 89 Tune the Position Loop (Optional) 92 This chapter provides a common method of manually tuning a drive. It is often referred to as inside out tuning, where the inner current loop is tuned, then the velocity loop is tuned, followed by the outer position loop if the drive is in position mode. It involves incremental increases in control loop gains to the point of marginal stability, then backing them off by a given percentage. Typical ranges for various gains are also given to provide guidelines. The out-of-box and autotune rigid methods achieve relatively high performance. However, if you are comfortable with tuning, this method can help to optimize performance from autotune compliant settings or if maximum performance is required. The method can be used in the following situations where previously described methods do not produce the required level of performance: To optimize performance starting with out-of-box or autotune settings To tune a difficult load from a stable initial state Other requirements include the following: Start with a set of gains that produces stable operation The load is connected to the motor while tuning Rockwell Automation Publication 750-AT006A-EN-P - June

88 Chapter 5 Manual Tuning Initial Configuration If you have configured the drive with the out-of-box recommended default settings or with autotune that is outlined in the previous chapters, but you still require more performance, skip this section and proceed to Tune the Current Loop (Optional) on page 88. If the mechanical load is unstable or your control loop gain values are undesirable, then start with this section. However, we advise that you first follow the tuning flowchart to yield a good stable starting point. See Is Further Tuning Required? on page 86 for more information. Follow these steps to bring the drive back to an initial state with default settings.. Configure drive parameters by following the Recommended Default Settings on page Run the drive and adjust 0:906 [System BW] to achieve a desired performance. 3. If desired performance is not yet achieved, do the following: a. Run 0:90 [Autotune] = JMtr BW Calc (7) to measure the motor inertia and reset the control loop gains for R = 0. Perform this test with the load disconnected from the motor. b. Or enter a data sheet value for 0:900 [Motor Inertia] and run 0:90 [Autotune] = BW Calc (6) if the load cannot be disconnected. c. Run the drive and adjust 0:906 [System BW] to achieve a desired performance. 4. If desired performance is not yet achieved, do the following: a. Run 0:90 [Autotune] = JTotalBWCalc (8) to measure the total inertia and reset the control loop gains for R > 0. Perform this test with the load connected to the motor. b. Run the drive and adjust 0:906 [System BW] to achieve a desired performance. 5. If desired performance is not yet achieved, continue to the next section. Tune the Current Loop (Optional) Follow these steps to manually tune the current loop.. Set 0:425 [PWM Frequency] as high as your power structure allows. 2. Run the drive and adjust 0:445 [VCL CReg BW] to achieve a desired performance. IMPORTANT Torque loop performance increases as 0:445 [VCL CReg BW] increases. However, torque ripple and noise also increase. Excess noise causes the velocity loop to be tuned lower. As a result, a trade-off exists in setting the bandwidth to balance performance with torque ripple and noise. 88 Rockwell Automation Publication 750-AT006A-EN-P - June 207

89 Manual Tuning Chapter 5 3. If the desired current/torque loop performance is not yet achieved, do the following: a. Set 0:50 [MtrParam C/U Sel] = User Entered (). You can first set it to copy the calculated parameter values over to the user entered parameters. b. Run 0:90 [Autotune] = Rotate MtrID (3) to measure the motor electrical parameters. If you cannot rotate the load, then run 0:90 [Autotune] = Static MtrID (2) to measure the motor electrical parameters. c. Run the drive and adjust 0:445 [VCL CReg BW] to achieve a desired performance. 4. If the desired current/torque loop performance is not yet achieved, do the following: a. Set 0:444 [VCL CReg C/U Sel] = User Entered (). You can first set it to copy the calculated parameter values over to a user entered parameters. b. Manually adjust 0:447 [u VCL CReg Kp] and 0:449 [u VCL CReg Ki] to achieve the best possible performance. Tune the Velocity Loop Follow these steps to manually tune the velocity loop.. Set 0:905 [System C/U Select] = User Entered (). You can first set it to copy the calculated parameter values over to the user entered parameters. 2. Temporarily isolate the velocity loop. This step is only required if the drive is in position mode. a. Record the values of the following parameters. They are temporarily disabled to isolate the velocity loop and later restored to these original values. 0:755 [u PReg Kp] 0:757 [u PReg Ki] 0:255 [u Trq LPF BW] 0:974 [u Accel FF Gain] b. Temporarily set these parameters to zero. c. Set all torque notch filter bandwidths to zero. 3. Run the drive (command motion) and slowly increase 0:956 [u VReg Kp] while looking for a reduction in velocity error. a. If the 0:90 [Load Ratio] = 0 and 0:2020 [LdObs Mode] = LdObs VelEst (2) or Vel Est Only (3), set 0:2022 [u LdObs Kp] = 4 * 0:956 [u VReg Kp]. If 0:2020 [LdObs Mode] = Disabled (0), there is no need to adjust 0:2022 [u LdObs Kp]. Rockwell Automation Publication 750-AT006A-EN-P - June

90 Chapter 5 Manual Tuning b. Otherwise, set 0:2022 [u LdObs Kp] = * 0:956 [u VReg Kp]. IMPORTANT Every time that you increase or decrease 0:956 [u VReg Kp] in the following steps, also increase or decrease 0:2022 [u LdObs Kp] to keep the or 4 times ratio between them constant. c. Continue to increase 0:956 [u VReg Kp]until ringing occurs in torque signals or an audible noise exists at any time while tuning. d. Stop and restart motion after each modification of 0:956 [u VReg Kp]. 4. Manually compensate for resonances. a. Set 0:20 [AdptTune Config] = Disabled (0) to manually tune the torque low pass and notch filters. b. Enable and jog the drive (momentarily command motion) to listen for audible resonance. Jogging the drive is unnecessary if audible resonance is heard while enabling the drive. c. Determine if a high frequency resonance exists in your application. If an audible high frequency resonance is not present during the previous step, skip the remaining steps, and velocity loop tuning is complete. If an audible high frequency resonance is present during the previous step, use a mobile application to identify resonances. An ianalyzer Lite example shows that the dominant resonance frequency has the largest peak. Figure 65 - ianalyzer Lite Example d. If resonances are below the torque loop bandwidth or a low pitch growling sound is present, then instability is present and you must decrease 0:956 [u VReg Kp] and 0:2022 [u LdObs Kp] before continuing to the next steps. e. Set the first torque notch filter frequency to the resonant frequency with the largest magnitude. f. If multiple resonances have nearly the same magnitude, set the torque notch filter frequency to the lowest resonant frequency. 90 Rockwell Automation Publication 750-AT006A-EN-P - June 207