USCG QUESTIONS RELATED TO PID CONTROL

Transcription

1 USCG questions related to PID control by Frank Owen Maine Maritime Academy 4 December 2018 Control loops on the USCG exam are often called analog control. Much of the material is about legacy (i.e. old-fashioned) systems, but you may run across them too in your work. There are many old systems still out there running. Op-amp-based P-only control This is, I believe, a P-only controller made with op amps. On the left you see an input from a sensor and also a potentiometer input that constitutes the setpoint. The 1 kw pot for the setpoint is connected to a knob that you turn or a slider that you move to select some voltage between 0 and 5 VDC. On the dial or slider panel will be a scale in whatever variable you are setting the setpoint for. The Sensor input is on top here, whereas in our loops, we ve had the setpoint coming in from the left and the sensor input below it, coming back from the feedback part of the loop. See the cover of the book for that diagram. This seems to be a P-only controller because there is only the Proportional gain amplifier. There is no evidence of an integral nor derivative path. The Inverter at the output of the electronic mess is probably used to invert an electrical voltage that was inverted in the comparator process of subtracting the sensor input from the setpoint. I have seen such a system at Cal Poly. The output from the op-amp summing operation is inverted during the summing, so it needs to be inverted back. The output from the Proportional gain amplifier is indeed the error signal amplified by multiplying it by the proportional gain. The first two options would be done by U 1. The last option is done by U 3. 1 P a g e

2 2 P a g e

3 Op-amp-based PID control This figure is similar to the previous figure, but here there are all three parts of the PID. On the left, the Process variable from the sensor is fed back and compared with the setpoint. We usually see in our loop diagrams the setpoint coming in from the left and the process variable coming in from the sensor below at the summing junction. See front cover of book. Here there is no inverter on the output. But there is a Summer and Multiplier. The Multiplier indicates that perhaps this loop has the configuration of the PID shown as the second possibility in Table 9.1 in the textbook the PID with K PID, T I, and T D. U1 is the comparator in the control loop, which compares the setpoint with the signal coming in the feedback path from the sensor. The proportional action in a PID controller is just proportional to the error (the difference between the setpoint and the sensed value from the sensor. 3 P a g e

4 Integral action is the area under the error curve from the beginning of the control to the present instant of time. Derivative action is made by multiplying the slope of the error curve at the present time by a constant. The slope is the rate of change of the error curve. So the derivative action is proportional to the slope (rate of change) of the error curve at the current instant of time. 4 P a g e

5 PID function block There is a PID function block that implements the function of a PID controller: This is a figure from the USCG exam. The various inputs are mostly identifiable: Auto Flag to tell whether PID is active or not (if in manual mode, Auto = 0) PV process variable, i.e. the measured value SP Setpoint, the desired value XO??? KP Proportional gain TR Reset time (related to K I) TD Derivative time (related to K D) CYCLE??? XOUT is the controller command output to the actuator The values for the controller gains come from other memory locations and are fed into the PID block from outside; i.e. they are not stored internally in the block. Now that we see what the blocks look like and how they function, let s look at a simple example of an entire function block diagram for an application. 5 P a g e

6 A is the correct answer. PV stands for process variable, which is another name for C (actual value) in the world of process controls. The RA values stand for register address, which are areas in the PLC memory that contain certain values. Noteworthy here is that you can mix the two types of control that we ve taken pains to separate: 1) PLC ON/OFF control and 2) PID continuous (also called analog ) control. In other words, you can run a PID algorithm on a PLC platform, or, said yet another way, the PLC is capable not only of ladder-logic decisions but also of hosting continuous-control (PID) algorithms. Where these two types of control were separate before, the growth in processing power of the PLC has made it possible for it to serve also as a PID controller. Other register contents are Register address 0325 Setpoint (desired value) Contents 6 P a g e

7 0326 Process variable (actual value) 0330 Error 0331 PID output (U) 0327 K P 0328 T I (or reset) 0329 T D (or rate) Note that the register addresses don t mean anything. They are arbitrary. Here they show simply that there is a different address for the various quantities. The PID algorithm needs to have various values to operate, and these addresses say where those values are stored. 7 P a g e

8 Advisory vs. supervisory control (EL-0094) 8 P a g e

9 C) is the correct answer.. 9 P a g e

10 10 P a g e

11 The correct answer is B. The USCG counts C as the correct answer, which is not correct. I think what I would do here is answer B and then challenge the results if it is counted wrong. 11 P a g e

12 12 P a g e

13 D/A and A/D conversion To read an analog signal (voltage), which is continuously and continually changing, it must be discretized or digitized. This is called A-to-D conversion (ADC). The converter contains a number of switches that are off or on, depending on whether or not the voltage measured at an instant of time has arrived at a certain level or not. The figure below illustrates this. The switches are represented by the binary numbers on the left. At any instant of time, the voltage is measured. It will be converted only into a certain voltage level if that level has been reached. This ADC has a 4-bit resolution, which means that it can take on 8 different voltage levels. This means that only steps in voltage of 1/8 the range measured are possible. This is V in the figure. Thus the smoothly changing signal is seen by the digital system as the rough, stairstep signal shown. Of course the digital version of the signal can be improved by adding off/on switches to increase the resolution. With each added bit, the resolution doubles. The figure below shows the improvement of the signal by adding an additional bit, so that the converter has a 5-bit resolution. Thus, the more bits dedicated to the ADC, the better the digital version of the curve matches the real curve. The other thing that figures into how well or badly the digital curve matches the smooth, analog curve is how fast the sampling is done. To decide on which switches to turn off or on, the voltage must be sampled or sensed, held, then converted. This is called sample-on-hold. The time it takes to sample and set the bits is called the sampling time ( t above) or sampling rate. If the sampling rate can be increased (the time between samples reduced), this also improves the accuracy of the digital version of the signal. The figure below illustrates what would happen to the above signal if the sampling rate were doubled. 13 P a g e

14 Thus the greater the resolution (number of bits of the ADC) and the faster the sampling rate, the better the digital signal represents the smooth, analog curve. These examples use very coarse, very slow sampling, what might have been seen in the early days of ADCs. Today s ADCS have high bit resolutions and very fast conversions to digital format. The stairsteps are still there, but they are very, very fine, so that the curve is represented very well compared with the examples shown here. 14 P a g e

15 You need not go into the details of how Block A or Block B above works. What you need to know is that the digital representation of an analog voltage is a set of bits which form a digital word. In the above cases, there are 8 bits dedicated to the digital representation of a voltage, so in the range of measurement, the analog voltage can have 2 8 different values, that is 256 different values. In the top figure, an analog voltage is being converted into its digital representation. In the bottom figure, the opposite is taking place: a digital representation is being turned into an analog voltage. Thus the top is an ADC (analog-to-digital converter) and the bottom is a DAC (digital-to-analog converter). 15 P a g e

16 Ship steering modes This seems to be a ship steering system, judging from the input from the Steering Wheel. It is a redundant system, in that it can operate either from the Port system or from the Stbd system. The selector switch for that is in the center of the diagram. If we look at one of the Control Potentiometers, what is happening here is that the Steering Wheel is delivering the setpoint of direction (the desired azimuth setting) to the potentiometer, and the gyrocompass is delivering the actual azimuth setting to the Control potentiometer. These two signals are compared, and the output (the error in a control loop) is delivered to an amplifier that then goes to a power unit to actuate the ship s rudder. At the bottom of the diagram, there is a selector switch for the operating mode. Hand means that the control loop is disabled, i.e. is open, and the Port or Stbd Power Unit is getting a signal from manual control. Gyro here indicates that the signal to the power unit is coming from a gyroscope that is used to get the ship s current heading. NFU stands for non-follow-up. This is an open-loop mode where there is no mechanical feedback to shut-off a hydraulically actuated valve when the thing that it actuates changes position. The question is not well posed. The Controller position is ambiguous. Do they mean the NFU Controller position? Apparently, this is what they mean. In that case, what is active in that position, the one shown, is the non-follow-up controller, so (b) is the correct answer. See discussion below on the various steering modes of a ship. 16 P a g e

17 These are various modes of steering a ship. I do not know what DIFF is. HAND is actually the fly-by-wire (or steer-by-wire system) when the ship s wheel, when turned, commands a corresponding deflection of the rudder. The rudder angle is sensed and fed back to the controller so that the angle commanded by the wheel is sought by the actuators turning the rudder. When the rudder reaches the commanded angle, it stops. HAND mode is also known as follow-up (FU) mode, because the feedback loop follows up the shifting of the hydraulic servo valve and shifts this valve to its neutral position when the desired rudder angle is reached. If there were no follow-up function, shifting the valve would give the rudder a rotational velocity, and it would continue to turn until it reached its stops. GYRO is an autopilot. A certain azimuth angle of travel is selected by the user, and the ship s autopilot system ensures that the ship tracks this angle. Or perhaps the desired ship s track is set, and the rudder angle is changed automatically by the controller to ensure that the ship follows this track. Before the days of GPS systems, I think that maybe autopilots simply used a gyrocompass to maintain a desired heading. With GPS systems it is now possible to put in a desired track and, through GPS feedback, have the ship follow that track. NFU mode is non-follow-up mode. Here the controller just directly operates the servo valve that moves the rudder. This is an open-loop system, so the rudder angle sensed by the angle sensor is not fed back and used by the controller. This is a direct, non-feedback deflection of the rudder by holding the servo valve open to a specific deflection. As mentioned above, shifting the servo valve manually will make hydraulic fluid continue to flow in a certain direction into the cylinders that move the rudder. Thus what results from such a shift is that the rudder continues to rotate until it comes up against its stop. Turning a rudder in NFU mode produces a rotational velocity of the rudder, not just a rotational angle. Otherwise said: in follow-up mode the turn of the wheel produces a specified, commanded rudder angle. In non-follow-up mode, turning the wheel produces a commanded rotational velocity of the rudder. Hydraulic valves, whether they be directional-control valves or servo valves, are usually self-centering. That means that if the shifting signal is taken away, the valve returns to its neutral, no-flow/no-motion position. I read on the web that traditionally in NFU mode, when you command the rudder to turn, you have to hold the wheel or joystick or whatever the command device is over from its centered position. If you don t, it will return back to the neutral position, where the motion stops. This is a good feature for NFU mode. Without the centering function, it would be less safe and also a burden to have to find the center position manually, especially if you were rocking and rolling in a heavy seaway. In conclusion, if we were to rank the three modes (GYRO, HAND, and NFU) in their order of automation, they would be ranked in that order. GYRO just adjusts the rudder automatically as needed to follow a commanded track; the ship steers itself. In HAND (or FU) mode, the wheel commands the rudder to go to a certain angle. The rudder angle is sensed and fed back in the control loop until the actual rudder 17 P a g e

18 angle matches the commanded rudder angle. In NFU mode, turning the wheel just opens the hydraulic valve that moves the rudder, and the rudder continues to move until it reaches its stop. For the above question, NFU is the dumbest sort of steering mode. It simply shifts a valve without regard for controlling anything else, like rate-of-turn in the above question. See for a discussion of autopilot (GYRO), FU (HAND), and NFU (manual) modes of steering. For the top question, non-follow-up mode does not feed the rudder angle back, so the second and third possibilities can be eliminated. Gyro-compass certainly does feed the rudder angle back. So does hand steering; it s a fly-by-wire system that takes a desired rudder angle and actuates the system until the actual rudder angle is the same as the desired rudder angle. Don t know what synchronizing is. Anyway, the first answer is the correct answer. This question is ambiguously posed in my opinion. Does putting the steering mode in NFU mode disable the Steering Wheel? Apparently, according to the diagram it does. Choosing NFU mode on the bottom also puts the System Selector Switch to Off, and this disables any commands coming from the helm (Steering Wheel). This is seen clearly in the drawing. What should be shown on the drawing and is not is the direction of signal flow, shown with small arrows on each signal line. 18 P a g e

19 Steer-by-wire rudder control This is a steer-by-wire system used to position a ship s rudder. The PID error signal input on the left is, in my opinion, really the rudder-angle setpoint. The Rudder error amplifier is, in my opinion, really a P- only controller. The output of the Rudder error amplifier (really the controller) can be sent to one of three different places NFU, FU, or Auto but it is not clear what these mean. RLA and RLB refer, I think, to relays. The relays are used to activate the Rudder Prime mover, often a hydraulic unit, to move the rudder clockwise or counterclockwise. The Rudder translator is a potentiometer mechanically connected to the Rudder linkage to measure the actual position of the Rudder. This is the sensor in the loop. The signal from this pot is set back along the feedback path to be compared with the setpoint. 19 P a g e

20 Legacy steering system The figure below shows a legacy (i.e. old fashioned ) steering system, prior to the days of computer control. The system can be run in a number of different modes. One of the modes steer-by-wire is simply a force-magnification system. To move the rudder on a big ship, a lot of torque is needed to deflect the rudder into the passing water stream. Ships, like airplanes, have usually used a hydraulic system to provide the torque to turn the rudder. Thus the helmsman turns the wheel on the bridge. This does nothing more than set the setpoint for the rudder angle. Turning the ship s wheel through a certain angle merely expresses the wish to have the rudder turn through a certain angle. The hydraulic system goes into action, turning the rudder, until the measured angle is equal to the desired angle. A sensor is mounted on the rudder to provide the actual rudder position to the controller in a feedback path. From a Controls standpoint, in steer-by-wire mode, the control-loop gain is much less than 1: the ship s wheel can be turned through many degrees, but the rudder is limited to around ±30 of travel. 20 P a g e

21 With this basic understanding of the steer-by-wire mode, let s look at the system and try to identify the function of its parts. Let s start by focusing on how the rudder is moved. Below is shown the floating lever from the above drawing. This component has this name because it is not pivoted about any fixed point. The steering engine is what actually moves the rudder. The two rods coming out from it to the tiller, the component atop the rudder, drive the rudder back and forth. These rods are commanded by the input rod atop the steering engine, entering the steering engine from its forward side. Internal hydraulic passages and spool valves take this input and convert it into the commands to the output rods. As the rudder turns, the follow-up linkage reflects this movement mechanically back to the floating lever. This will cause in the hydraulic ram and in the telemotor a movement that will shut-off hydraulic command flow to the rudder to make it continue to move. The follow-up function is a way to convert a hydraulic input signal, which usually produces a rotational velocity, to produce instead a rotation itself. To understand fully how mechanical follow-up works with a hydraulic valve, one needs to look at the hydraulic valve and see the mechanical feedback that produces this effect. This is not considered here, as it is a specialized topic. One needs only to know that mechanical follow-up allows an input angle to produce an output angle in lieu of an output velocity. The two inputs to the floating lever are the hydraulic ram and the telemotor. Either can command the rudder to move, but in different modes. When the telemotor is active, the floating lever rotates about the hydraulic ram pivot point; when the hydraulic ram is active, the floating lever rotates about the telemotor pivot point. The telemotor seems to be driven by the ship s wheel. The wheel turns a pinion that moves two racks, one up and one down. The racks are connected to hydraulic pistons that then move a piston in the telemotor receiver forward or aft. A clockwise turn of the wheel will drive the starboard piston down and the port piston up. This will in turn drive the piston in the telemotor aft, which will push the input rod into the steering engine, and the rudder will rotate counter-clockwise (from above) to turn the ship to starboard. Thus in this mode, the ship s wheel acts like a little pump to deflect the floating lever and move the rudder. This mode of operation is purely hydro-mechanical. There is no electrical signal involved in moving the rudder. The rudder angle transmitter does sense the rudder angle and feed it back to the rudder angle indicator, but this has no control function in an automatic loop. The helmsman sees the rudder angle and adjusts his/her input to the ship s wheel, thus serving as the controller in this hydro-mechanical mode of operation. The by pass valve control is a little valve that, when opened, disables the wheel. When opened, the flow from the down-going little rack-mounted pistons simply flows through the bypass into the chamber for the up-going piston. The telemotor receives no flow when the bypass is opened. The interlock switch 21 P a g e

22 senses the position of the by pass valve and prevents operation of any electrical actuation if the bypass valve is closed and the wheel/telemotor system is active. The steer-by-wire mode is also a hand operating mode that uses electrical transmission of the bridge commands to move the rudder. Besides this Hand steering mode, there is a Gyro mode. This is an autopilot mode whose purpose is to maintain the ship s heading despite disturbances of wind and waves. The choice between modes is made by selector lever the control arm in the illustration below. It is set into either Hand or Gyro mode. In Hand mode, the Hand steering potentiometer is used to set the rudder-angle setpoint. This signal is amplified and sent to the Control panel, which seems to be simply the controller. No details are given on its operation or even what type of device it is. Notice also that the control panel gets a signal coming in from below that comes from the Rudder repeat back potentiometer which gives the actual position of the rudder. Signals from the control panel to the Servo valve actuate it. The servo valve controls flow into and out of the hydraulic ram, which moves the floating lever. The floating lever pivots about the locked telemotor receiver, so the input rod to the steering engine moves in or out to cause rotation of the rudder. Thus in Hand mode, the helmsman turns a knob or a small wheel connected to the Hand steering potentiometer and the electrical signal from there eventually leads to rudder rotation. The Gyro mode is an autopilot. The Gyro steering potentiometer sets the desired heading to steer. It is turned by a knob or wheel on a panel whose readout around the knob or wheel is desired heading angle. The actual heading angle coms in from the compass transmitter. Thus the helmsman sets a heading, and the steering system in this mode turns the rudder to maintain that heading. Nowadays the automatic mode doesn t just maintain a heading. With GPS it is possible to lay out a track that the helmsman wants the ship to follow. The position of the ship is detected via GPS. The rudder is automatically controller to steer the ship to stay on that course line. The heading of the ship is 22 P a g e

23 unimportant. With wind, waves, and current the heading angle of the ship may not be the course heading. That is, the ship may need to crab somewhat to maintain a course at a desired heading. Now, with all that background, let s see what the Coast Guard question relative to this detailed figure is. Here the helm is the ship s wheel, apparently. Thus the manual, hydro-mechanical mode is what is being referred too, i.e. the transmission of the turning of the ship s wheel to the telemotor. 23 P a g e

24 PID control of steering This is a strange question. I suspect that it was written by someone who does not know PID control very well. The top scenario is described as a P-only control system to maintain a certain course. The bottom example is a PD controller. The middle scenario is a D-only controller, something I have never encountered in many years involvement with control systems. What the control aim is here is position control, to have the ship follow a certain laid-out track. As long as the ship is on that prescribed track, the error is 0. Regarding the first scenario P-only control it is not true that a ship under P-only control will continuously oscillate around a given track. it will only do this if K P is turned up too high. With a lower KP, the ship will not oscillate around the course. It may be offset from the course because of the lack of integral control, but it need not be the case that under P-control the ship s course will oscillate. Regarding the D-only controller, it is true that if the ship s course is not the same as the desired course and not parallel to the desired course, the error will be changing, and thus the derivative action of a controller will be active. But usually derivative control is part of a PD or a PID controller. It is also true 24 P a g e

25 that if the actual course is parallel to the true course, the error is not changing, i.e. is constant, so the derivative action will be 0. But with P-control, you can still have this offset. The way to get rid of offset is with integral control. Thus the third scenario should rather be a PI controller than a PD controller. Integral action is what gets rid of offset. There is no question given with this figure. 25 P a g e

26 Gyro-compass function block diagram (EL-0194) Details of operation of gyro-compass Rate-of-turn specifications Question: 26 P a g e

27 According to the specifications shown with the figure, the rate-of-turn sensor puts out a voltage proportional to the rate-of-turn, the speed at which the ship s heading is changing. Rate-of-turn has units deg/min. Thus in a block diagram: with the additional specification that if a positive voltage indicates a turn to starboard (and a negative voltage indicates a turn to port). So if the turn is 30 deg/min to port, then the sensor will read = 30 / 50 / = 1500 = P a g e

28 Gyro-compass trouble-shooting diagram (EL-0195) If you follow the logic through the flowchart with the symptoms described, you wind up at the box that says Replace CPU Assembly. So the gyro-compass CPU (central processing unit) is probably bad ( ) Using the trouble analysis chart and faults table provided in the illustration, if the gyrocompass was malfunctioning, but no fault codes are present on the display unit, what is most likely the problem if the DC/DC converter LED status indicator is functioning properly, but the CPU LED status indicator is not blinking? Illustration EL-0195 o (a) The AC/DC power supply is malfunctioning. o (b) Ship's power is not available. o (c) The DC/DC converter is malfunctioning. (d) The CPU assembly is malfunctioning. 28 P a g e

29 Ship s DCS (distributed control system) This diagram shows an integrated system for a ship s automation system. Some features: The computers above with ROS on them are remote operating stations (ROS). They are connected to a dual network, probably for redundancy purposes. This is indicated by the two horizontal lines below these stations. There are operating stations for various functions on the ship. Underneath the system dual network bus, you see Dual CAN. CAN stands for Controller Area Network and is a bus system to connect hardware for sensing and actuating. CAN systems were first developed for automobiles and trucks by Bosch in the 1980s. 29 P a g e

30 The LOSs are some type of operating stations. I don t know what L stands for. Maybe local. Thus an operator or troubleshooter could go out into the plant with a hand-held unit and diagnose or change operating parameters of equipment while observing its operation. DPU probably stands for Distributed Processing Unit. These are local units that handle much of the control themselves, with the higher level monitoring of the system taking place on the ROSs at the top of the diagram. Swbd stands for switchboard, used for electrical power. The CAN bus is broken down into modules used for various functions on board as indicated by the captions at the bottom of the diagram. 30 P a g e

31 Ship s DCS This diagram seems to represent an overall view of a distributed control system. The Computer (CPU Central Processing Unit) is where the control of various Plant Processes is done. This would be where the PID functions reside. Input comes into the computer from the right, is operated on (decisions are made), and then control signals are sent to the left, out from the CPU to the field for actuation. The drawing emphasized the conversion of signals from the field into computer-readable format and then the conversion of computer signals back into output that can be used to actuate things in the field. Note that the input signals are either digital (OFF/ON) or analog (a range of voltages or currents). The output signals are also digital or analog. This is not a single-input/single-output loop. It contains many loops. Many things are controlled and sensed, as indicated by the parallel arrows into and out of the Plant Processes at the top of the diagram. The Operator Console is used to monitor the processes being controlled and even to manipulate these processes by changing setpoints or manually operating devices in hand (manual) mode. 31 P a g e

32 Off/On rudder positioning system This is a bad question. The deadband is the space in the middle of the control band where nothing turns on. Thus the deadband extends from 14 to 20, i.e. is 6 wide, not the 8 shown. Here s what s happening: The rudder is commanded to go to the position represented by V c. The V p voltage is the voltage coming back from a pot mounted on the rudder so that it gives rudder position. Thus the position commanded is about Because of the relays, it seems that the rudder-positioning mechanism, most likely a hydraulic positioner, is either on or off. I.e. this is not a hydraulic servo system. We are familiar with on/off controllers: the common thermostat in a house is such a system. For heating, when the temperature falls to a certain level (T low ), the heating system turns on. It stays on until the temperature climbs to a pre-set level (T high ), whereupon it turns off. If T high is not so much greater than T low, then the heating system will maintain a more constant temperature, but the heating system will turn off and on frequently. The above system has the added complexity that the rudder motor turns on to swing the rudder toward the commanded position. But then as it approaches this position, the moving mechanism turns off. The rudder, however, continues to move because of the rotational momentum imparted to it when the rudder motor was on. The rudder continues to rotate, due to this rotational inertia, until it moves so far beyond the commanded position. Then the opposite rotation is commanded to bring the rudder back to the commanded position The rudder overshoots this in the opposite direction, and the original rotation is again commanded to push the rudder back toward its commanded position. This seesawing back and forth often called hunting continues indefinitely. The positioning of the rudder is inherently unstable according to the USCG. 32 P a g e

33 From a Controls standpoint, this is not true. Inherent instability means that the rotational swings of the rudder would get bigger and bigger with each oscillation. Also, with the relays, this is on/off control. What this is is called a limit cycle in the jargon of non-linear controls. What is possibly wrong here is that the rudder has too much angular momentum, so once it starts rotating, it is difficult to stop. Another possibility is that the actuating mechanism for the rudder is too powerful. Too much of a torque is imparted on the rudder by the actuating mechanism, and this gives it too much and too sudden a motion. So lightening the rudder or not giving such big rotational actuation commands will prevent this behavior from occurring. 33 P a g e

34 Ship steering phantom rudder Starboard relay is shut off because phantom rudder says that it s arrived at the cut-off point even though real rudder is not yet at that point It is not clear why phantom rudder takes this trajectory Phantom rudder: Since the rudder has so much rotational momentum once in motion, this is a strategy for anticipating where it will be before it gets there and basing control commands on this predicted position. What is known is the steady rate of increase in the rudder angle once it gets moving i.e. its rotational velocity. It takes the rudder a bit of time to get up to that rotational velocity, but the algorithm pretends that it goes to this rate immediately. The position as determined by this sudden velocity are used to turn the rudder relays on and off. 34 P a g e