Requirements for Industrial Cabling Supporting Gigabit Applications for Control, A study of Noise

Transcription

1 Requirements for Industrial Cabling Supporting Gigabit Applications for Control, A study of Noise Bob Lounsbury Principal Engineer, Rockwell Automation Rockwell Automation Presented at the ODVA 2015 Industry Conference & 17th Annual Meeting October 13-15, 2015 Frisco, Texas, USA Abstract This paper is the result of over 700 man hours of cable measurements, noise testing and analysis resulting in defining cabling parameters supporting 1Gig cabling requirements for harsh industrial environments. In the early 2000 a study was performed by several ODVA member companies. The goal of this study was to define the requirements for 10/100mb/s twisted pair cabling for industrial environments. This eventually resulted in the creation of national and international standards defining parameters for cabling used on the factory floor and industrial machine areas. Once again with the need for additional bandwidth we find the need to define the parameters required to support of 1gig data rates on 4 pair industrial cabling. This paper will discuss the approach used, the testing, analysis and the results of the study. The testing, is based on standardized transmission line testing techniques and environmental noise testing as defined by ISO/IEC and IEC standards. In addition this paper will discuss the impact of the new limits to the parameters from a technical perspective. Since industrial cabling is the subject to national and international standards this paper will briefly cover the approach to standardization, including the EtherNet/IP Specification. The paper will conclude with a complete set of cabling component specifications defining the communications channel supporting robust data communications for control applications. The purpose of this project is to determine the required cable performance needed to support 1000 Base- TX for industrial control applications. This study and white paper does not consider applications using office grade cabling in office (MICE 1) environments. This report documents the tests, test results and conclusions of cable transmission and EMC testing. Both unshielded and shielded cables were considered and evaluated. In the case of shielded cables additional testing is performed to determine the shielding effectiveness since it is expected that shielding has a positive impact on cable s noise rejection. The foundation of this project is based on examining the performance of off-the-shelf cables and engineering samples designed for both office and industrial applications. The design of the cabling targeted for Industrial applications must be carefully considered where the environments are MICE - E 2 and MICE - E3. Keywords TCL, Balance, Coupling Attenuation, Channel, Industrial Cabling, EMC, BER, Noise, 2015 ODVA Industry Conference ODVA, Inc.

2 Acronyms TCL Transverse Conversion Loss (cable balance) BER Bit Error Rate Coupling Attenuation MICE Mechanical Ingress Climatic/Chemical Electromagnetic EMC Electromagnetic Compatibility Vn_Diff Differential Noise Voltage Process and Approach 31 cable samples were obtained from various vendors. Cable types ranged from UTP to Shielded, both solid and stranded designs. In this paper, I use Shield as a generic term for all foil and or screened designs. The cable designs were a mix of commercial and industrial designs. One problem with the term industrial is that there is no real definition as to what makes a product industrial rated. In this paper I use this term to loosely describe any attribute that is consistent with industrial. For example a 70 degree cable design is industrial. However later in this paper I will use this term to describe a more definitive and complete set of attributes that makes up industrial The following flow Figure 1 helps to describe the testing approach. Data collection and file management became a large factor in the efficient success of this project. For test consistency all test programs have been automated using HP-VEE. Visual Basic was used to compile the data. Finally, MathCad was used to create the performance equations and limit lines based on the compiled data and calculations. All in all there were some 350 data files created across the 31 test samples ODVA Industry Conference ODVA, Inc.

3 Figure 1 Project Plan Test Flow Start Test Plan and Cable Data Sheets Transmission Line Testing UTP and Shielded Data S4p and CSV CA Testing Shielded Cables Only Data XLSX Conducted Immunity Testing UTP/Shielded IEC Differential Noise Voltage Data XLSX IXIA BER Data XLSX Common Mode Impedance Testing UTP/Shielded Data XLSX EFT/B Testing UTP/Shielded IEC Differential Noise Voltage Data XLSX Data Compiled and Calculations Performed IXIA BER Data XLSX Report End 2015 ODVA Industry Conference ODVA, Inc.

4 The transmission performance of the 31 cable samples were obtained by using an Enhanced TDR (ENA) with two 28 port RF switches in accordance with IEC test procedures. For Shielded cable designs the coupling attenuation was captured using the Triaxial test method in accordance with IEC and IEC The noise performance testing was performed in accordance with IEC and IEC Background As was done for 10/100 Mb/s testing the goal was to correlate the cabling transmission performance to the noise rejection performance in EMC testing. Ethernet cabling systems use balanced twisted pair cables to achieve a high degree of common mode noise rejection. However in the early standardization of Category cables Balance (known today as TCL and ELTCTL) was not specified. The standards relied on the inherent properties of the twisted pairs and cross talk specifications to minimize noise in the channel. As data rates increased and elaborate encoding schemes were developed the sensitivity to noise increased. Armed with noise test results and analysis, industrial Consortia s entered into the commercial and generic cabling standards committees. The outcome was a set of balance specifications that finally became part of the standards in The balance specifications for industrial were first published in a set of Technical reports published by Working Group 3 of ISO/IEC/JTC1/SC25 as ISO/IEC and ISO/IEC TR Today balance requirements are specified for all categories greater than Cat 5 (Cat 5E TIA). With requirements specified for Class D channels in the aforementioned ISO/IEC standards. With Gigabit now moving into the industrial space and being used for machine control we once again must look at balance and other parameters to a higher frequency than before. We must keep in mind that from a signal encoding perspective we have 6dB less immunity over 100Mb Ethernet encoding methods. Yet we hope to get the same performance in noise as in 10/100 Mb/s Ethernet systems. Since ODVA networks require support for both UTP and Shielded cables both balance and shielding are important. As a result, this paper focuses on the main contributor to noise immunity which is cabling balance and shielding properties. First Analysis of Transmission and Noise Results Figure 1 above shows the test flow for the project. This project has taken about 700 man hours to date to create the tests, perform the tests and analyze the test results. In addition the project generated some 300 test data files. The end result provides a set of parameters needed to achieve a low noise channel in industrial noise environments. The end analysis consisted of the creation of limit lines based on measured cable performance and mode conversion calculations. The limit lines do not consider the feasibility of such a channel based on components. Correlation of Differential Noise Voltage to BER in Conducted Immunity As with the 10/100 Fast Ethernet study it was quickly determined that differential noise in the channel impacts the BER due to S/N degradation when noise is applied ODVA Industry Conference ODVA, Inc.

5 TCTL(dB) TCTL(dB) Cable A Good Packets Versus Vn Diff CH1 P1 PRBS F RX Sum Poly. (Sum) Cable A Equal_Mode Conversion UTP Frequency Figure 2 UTP Cable A The mode conversion for Cable A is less 40dB to 36.5dB, the resultant Vn_Diff is just over 0.03 Volts. The missing packets were moderately low Cable B Good Packets Versus Vn Diff Cable B Equal_Mode Conversion UTP CH1 P1 PRBS F RX Sum Poly. (Sum) Frequency Figure 3 UTP Cable B Cable B has lower balance performance resulting in higher Vn_Diff and higher error rates. This was found to be the case for most of the UTP samples tested. When the differential noise (Vn_Diff) exceeds 40mV the lost packets increase rapidly. There were no missing packets for the screened cables tested. In some cases the balance was not any better than the UTP when tested. However Vn_Diff was below volts and there were no lost packets. This is largely due to the high coupling attenuation of the cable s shield ODVA Industry Conference ODVA, Inc.

6 TCTL(dB) TCTL(dB) Cable C Good Packets Verses Vn Diff CH1 P1 PRBS F RX Sum Poly. (Sum) Cable C Equal_Mode Conversion Screened Cables Frequency Figure 4 Screened Cable C Cable D Good Packets Vurses Vn Diff Cable D Equal_Mode Conversion Screened Cable CH1 P1 PRBS F RX Sum Poly. (Sum) Frequency Figure 5 Screened Cable D The balance for Cable C (Figure 4) is measured as -35dB to -32.5dB at the highest frequency. The Vn_Diff is 0.02V at the highest frequency. For Cable D the balance is roughly 6dB better than Cable D and Vn_Diff is about 5mV less at 80MHz. This study concluded that for a standard PHY, if Vn is above 40mV we see lost packet error rate increase dramatically. Further the study shows us that the knee for acceptable performance and un-acceptable performance is very sharp. Measured Cable Performance The results are divided into two groups, UTP and Screened Cables. Screened cables have one major advantage over UTP in that the screen, when properly applied and terminated provides noise attenuation for the core of the cable. The screening attenuation can be subtracted from the balance numbers allowing for a lower TCL and TCTL cable value. Below is the distribution of the two cable designs. The distributions are created by trending the calculated TCL. Figure 6 is a graph of all the TCL trend lines for UTP cables. The graph shows a range of 5.8dB at the low end of the frequency spectrum to 15dB at 500 MHz. The range is greater for Screened cables, ranging from 21.2dB at 2.5 MHz to 11.56dB at 500 MHz. Figure 8 shows the range for the Screened cables ODVA Industry Conference ODVA, Inc.

7 Figure 6 TCL Range for UTP Cables Looking at the differential noise during conducted testing we see a large destitution in conversion from common mode to differential mode voltages of Vn. Figure 7 Vn_Diff for UTP Cables up to 80MHz 2015 ODVA Industry Conference ODVA, Inc.

8 Figure 8 TCL Range for Screened Cables The cables that exhibited poor TCL trend lines are relying on the shield properties for noise immunity. Figure 9 Vn_Diff for Screened Cables up to 80MHz In the case of UTP cables we see that Vn in conducted immunity testing produces differential voltages 87mV. For Screened cables we see differential voltages at less than 25mV. The required PHY sensitivity is less than or equal to 40mV. Therefore it is expected that the Screened cables will not experience high bit or packet errors since the Vn is less than RX_min threshold of the receiver. By analyzing the differential noise voltages and balance of 6 UTP cables the correlation between cable balance and Vn can be made. Figure 10 shows the relationship of Vn to balance of the UTP test 2015 ODVA Industry Conference ODVA, Inc.

9 samples. Vdiff is the differential noise voltage measured on the cable with 10V RMS applied through the conducted immunity coupling clamp. The graph line of equation Exp (Balance) shows the exponential equivalent of the TCL from the cable measurements according to the following Equation 1. Figure 10 Correlation between cable balance and Vn in UTP Cables TCTL 20 Exp( Balance ) 10 Equation 1 Exponential Equivalent of TCTL (ratio) Establishing the Limit Line for TCL The Channel limit line can be established by calculating Vn for a range o values over frequency. The component make up of the channel can be derived from partitioning the TCL between the components that make up the channel. First we look at the TCL provided by the standards for the three MICE noise levels defined by ISO/IEC Figure 11 shows us the limit lines defined by the standards for the three MICE E levels E1, E2 and E TCL_Chan_24702E1f TCL TCL_Chan_24702E2 40 TCL_Chan_24702E Figure 11 ISO/IEC TCL for E1, E2 and E3 MICE Levels 2015 ODVA Industry Conference ODVA, Inc.

10 Where; TCL_Chan_24702E1 = ISO/IEC E1 TCL TCL_Chan_24702E2 = ISO/IEC E2 TCL TCL_Chan_24702E3 = ISO/IEC E3 TCL The predicted Vn_Diff for each of the three TCL equations was calculated and graphed in Figure VnE1 VnE2 VnE Figure 12 Calculated Vn_Diff for ISO/IEC E1, E2 and E3 Levels Where; VnE1 = ISO/IEC E1 Vn_Diff VnE2 = ISO/IEC E2 Vn_Diff VnE3 is ISO/IEC E3 Vn_Diff Error! Reference source not found. shows us that for all three MICE E levels that we can expect the Vn_Diff to exceed the 40mV RX_min Threshold of the PHY s receiver at frequencies above 2.25MHz. The established TCL for industrial cables is insufficient to assure that Vn_Diff is less than 40mV at the receiver when the cabling system is subjected to 10V RMS noise. Based on testing results, the following channel limit line has been created by analyzing the test cable s performance in noise, see Equation 2 below. Equation 2 TCL for E3 Environments Where; MICE6,0 = 77 and MICE6,1 = 86.6 FChan is in MHz Indust_TCL_E3 ( fchan ) y MICE 15 log( fchan ) if fchan Figure 13 shows the comparison between the established TCL values set by ISO/IEC and the recommended E3 level for industrial gigabit cabling. 20 log( fchan ) y MICE 61 otherwise 2015 ODVA Industry Conference ODVA, Inc.

11 TCL_Chan_24702E1 20 TCL_Chan_24702E2 TCL_Chan_24702E3f 40 TCL Indust_TCL_E Figure 13 Comparison of ISO/IEC TCL to Industrial TCL for 1Gig Cabling Where; TCL_Chan_24702E1 = ISO/IEC E1 TCL TCL_Chan_24702E2 = ISO/IEC E2 TCL TCL_Chan_24702E3 = ISO/IEC E3 TCL Indust_TCL_E3 = Required E3 TCL needed to achieve a low noise channel Figure 14 shows the comparison of the current standards limits for TCL and the new proposed limit line for 1 Gig Channels. The calculated noise voltage for the industrial E3 limit line in conducted immunity is then less than 40mV up to 85MHz, whereas the ISO/IEC E3 limit line approaches 40mV at 42MHz. 1 VnE3 Indust_VnE Figure 14 Calculated Vn for 10V Applied ISO/IEC E3 and Industrial E3 Where; VnE3 = ISO/IEC E3 Vn_Diff Indust_VnE3 = Vn_Diff 2015 ODVA Industry Conference ODVA, Inc.

12 Component Contribution to Channel TCL The individual contribution of the channel elements (cable and connectors) make up the total Channel TCL. The relationship is found to be according to the following Equation 3; 20log10 TCL_Chan_Combined Equation 3 Component Contribution to Channel TCL TCL_Cable 20 n_conn 10 TCL Conn_sim 20 Where; ftcl is in MHz n_conn = the number of connections in the channel The goal is to propose new component limit lines that make up an acceptable channel TCL. Figure 15 shows the individual TCL contribution of the connectors and cable to the channel TCL. 20 TCL_Chan_Combined 40 TCL_Cable TCL Conn_sim Indust_TCL_E Analysis of Channel Element Contributions to Channel TCL Combined TCL Cable TCL Connector TCL Industrial E3 TCL Figure 15 Component Contribution to TCL Where; TCL_Chan_Combined = Calculated combined TCL of Cables and Connectors TCL_Cable = Cable TCL TCLConn_sim = Connector TCL Indust_TCL_E3 = Required TCL level to achieve a low noise channel The TCL for the connector implies that a better balance is required than offered by IEC series standard for 8-Way Modular connections. This may only be achieved through the use of M12 series of connectors. The requirement for cable TCL is very high and may require special manufacturing processes to achieve. The combined TCL meets the limit line of the industrial TCL limit line. The following equations describe the component TCL requirements to achieve the Industrial Channel TCL. y TCL_Ancor 15 log TCL_Ancor 7.4 TCL_Cable if 30 if TCL_Ancor 15 log if 30 y TCL_Limitotherwise Equation 4 Industrial Cable TCL ( ) 20 log f TCL otherwise ( TCL_Ancor 7.4) 20 log f TCL otherwise TCL_Limit 2015 ODVA Industry Conference ODVA, Inc.

13 TCL Conn_sim f comp TCL_Conn Ancor 20 logf comp TCL_Limit conn Equation 5 Industrial Connector TCL if TCL_Conn Ancor 20 log f comp otherwise TCL_Limit conn Where; TCL_Ancor = 84 and 91.4 TCL_ConnAncor = 68 FComp is in MHz The new Vn_Diff is calculated to exceed the 40mV at 60MHz and has a steep slope upwards as shown in Figure 16. Vn Differential TCL_Ratio RX_Limit Vn RX Sens Differential Noise (Mode Conversion) Figure 16 Vn_Diff for Industrial Channel Frequency (M Hz) Summary This white paper is a summary of the testing and results of the cabling research project. The Slide set dives deeper into the technical details of the testing and testing results. While there are many other factors that influence the communications system s performance in noise, this paper only focused on cabling balance. The testing showed that packet loss is low to non-existent when the differential noise voltage is below 40mV. From the 31 cable test samples a performance limit lines were determined. This limit line produced a very stringent specification for the UTP constructed channel that may not be possible with standard connectors and cables. Calculations were performed to create limit lines for the contributing components in the channel. These calculations then were combined to simulate the performance of a typical channel. The resulting component limit lines tell us that special constructions of UTP cables are needed and that the typical RJ 45 connectivity may not be optimal for a low noise channel. The test results for screened cables showed us that in environments where the noise is conducted that the shields provide enough isolation from the noise for most installations. This paper did not cover EFT/B type noises. However it has been demonstrated that screened cables alone are not sufficient to produce a low error channel for control in EFT/B environments. Additional work is needed in concert with the cable manufactures to determine the impact of cable designs meeting the higher TCL and ELTCTL values. It is expected that tradeoffs will be made in other cable parameters will be needed. The total channel length is not expected to decrease below the current limits defined today. Equations 4 and 5 provide the new TCL limits for the cable and connectors, that when combined meet the channel requirements ODVA Industry Conference ODVA, Inc.

14 ****************************************************************************************************************************************************** The ideas, opinions, and recommendations expressed herein are intended to describe concepts of the author(s) for the possible use of ODVA technologies and do not reflect the ideas, opinions, and recommendation of ODVA per se. Because ODVA technologies may be applied in many diverse situations and in conjunction with products and systems from multiple vendors, the reader and those responsible for specifying ODVA networks must determine for themselves the suitability and the suitability of ideas, opinions, and recommendations expressed herein for intended use. Copyright 2015 ODVA, Inc. All rights reserved. For permission to reproduce excerpts of this material, with appropriate attribution to the author(s), please contact ODVA on: TEL FAX WEB CIP, Common Industrial Protocol, CIP Energy, CIP Motion, CIP Safety, CIP Sync, CompoNet, ControlNet, DeviceNet, and EtherNet/IP are trademarks of ODVA, Inc. All other trademarks are property of their respective owners ODVA Industry Conference ODVA, Inc.