2

*Edited, AGAIN, in an attempt to remove any confusion

Settle this debate!

I am designing a Harness for testing the functionality of a PDU (Power Drive Unit), operated using CAN (ARINC-825). Prior to my time here, they used just regular wire (NOT BARE) with a resistor between them, for CAN Hi/Lo...no twisted pair, no shielding...in the same harness as the power lines (typically an A, B & C 115/200v, 400Hz and a +28VDC). Our contracted CAN Specialists insists that, due to the distance (~8ft), shielding is not necessary, nor is being a twisted-pair. Another consultant insists that both are necessary to ensure signal integrity... the consensus is that being a twisted-pair is not up for debate, so the ONLY question now is: Is there a possibility that running unshielded CAN lines, with the high and low voltage lines, result in communication interference?

NOTE:

I'm not sure why some say it "depends"....seems like a simple yes/no question...maybe "simple" isn't the best word.

Details:
8ft long Harness for Bench Testing PDU's (it will not move once connected) No Safety Requirements

System Power:

115/200VAC @ 400Hz, 3A (Max)
28VDC, 1A (Max)

CAN:

(using Peak PCAN View - PCAN-USB)
125k Baud
29-bit
50ms Message Frequency

Edit 1:

I would LOVE some supporting documentation.

Edit 2:

this is the CAN Cable that is specified by the manufacturer to be used with the PDU when in service.

Thanks Everyone!

joel cox
  • 41
  • 4
  • 1
    I get the no shielded part. Could work. Not twisted sounds suspicious - twisted or not, the impedance matters. No insulation, that's not what you really mean, do you? – Justme May 24 '23 at 19:07
  • By "Insulation" I'm referring to PTFE, etc. I don't mean bare wire. most shielded wire has the twisted pair, a insulation (PTFE, etc) and a shield (copper mesh, silver mesh, etc) to protect from EMI. – joel cox May 24 '23 at 19:35
  • By "insulation" I assume you mean an overall insulation jacket over the twisted pair ,as compared to a plain twisted pair - in either case, the individual wires would be insulated. – Peter Bennett May 24 '23 at 19:42
  • @joelcox Regarding the "simple" yes/no rather than "it depends". It depends. Without knowing the detailed specifications (Electric shematics, application demands, safety requirements, downtime requirements, etc....) no one can say for sure or even judge by knowledge/experience. If you want a "simple" answer: Use a proper 120ohm twisted pair cable made by a know brand for CAN. This is why industrie uses them: You can do nothing wrong....but you pay for that convenience! If you insist on "making it perfect" with "some" cable... Trouble can be ahead - but nobody can tell for sure. – ElectronicsStudent May 25 '23 at 14:36
  • @joelcox ".. possibility ... unshielded... interference" jup, there is. The questions is: What kind of interference? Can your slaves/controller recover the signal? Do they go into error mode? Do they have some sort of "re-send" mechanism implemented? How much downtime is allowed? What are the implications of a break-down? Can people get injured? Can machinery begin to burn/explode? Nobody knows except you. My approach is: Buy the "most known, best documented and most used, highest quality of cable"...Money is no problem, as you could save maybe 100$ but waste many thousands. – ElectronicsStudent May 25 '23 at 14:43
  • @ElectronicsStudent why would someone need schematics to make a decision? its 3 115/200v & (1) 28V power wires with a maximum load of 3A. Safety req. & downtime are both completely irrelevant as they have no bearing on whether or not that much AC & DC Voltages could cause EI... – joel cox May 25 '23 at 17:58
  • @joelcox The schematics can help understand what type of cabeling is used for the 115/200v bus - shielded and terminated on both ends? Safety grounded or functional grounding? TErminated only on one end? Safety req. & downtime are not irrelevant, as electrical interferenc is a statistical phenomena... Even the best effort will give you bit-errors for a given environment. The question now becomes: What intensity/likelyhood is acceptable? (Downtime and safety). These metrics will also govern the choices ahead: Are standards to meet? Is downtime critical - so opt for 10$ more in machinery? – ElectronicsStudent May 25 '23 at 18:04
  • @ElectronicsStudent ok lets pretend that safety and downtime would change whether or not there would be Electromagnetic Interference... and lets assume that all connections and terminations meet or exceed the most stringent code.. AND that no false errors are allowed due to interference... the question now is..... will unshielded suffice? – joel cox May 25 '23 at 18:22
  • @ElectronicsStudent also, Im not asking this to save money. I'm of the opinion that it is necessary, but I need more than myself and one other engineer, when a "specialist" is saying otherwise... i didnt add the myriad of details because they are all irrelevant to the question Im asking so I kept it as simple and straight forward as I could. – joel cox May 25 '23 at 18:35
  • @joelcox "Safety and downtime do NOT change .... Electromagnetic Interference". Signal integrity requirements (Shielding, cabeling used, components used, style of termination, type of grounding... ) change for a given Electromagnetic Interference based on the desired Safety and downtime. "also, Im not asking this to save money" - i read your comments, as if you would: E.g "will unshielded suffice?" Nobody knows for sure without measurements/simulation or blunt try/error. I'd recommend: Go with a proper twisted pair 120ohm shielded cable for CAN.The extra cost buys you: i don't care about EMI. – ElectronicsStudent May 25 '23 at 18:42
  • @joelcox If your specialist says: Unshielded will suffice and management doesn't want to spend the extra $: Go with it. Make clear, that the extra $ saves in: Problems at the customers site, problems during operation, downtime, engineering time spend, "looking like cheep equipment" and many more. I personally had this discussion only once: The feedback at my company was: Yeah, basically we don't care... use the "usual high quality stuff" and move on. I highly recommend this approach. If your specialist insists on cheap stuff, make the management understand that this is a slippery slope to go. – ElectronicsStudent May 25 '23 at 18:52

2 Answers2

3

I can give some advice from a EE-Job in the automotive industries (German plant, EE for machine and automation design).

  • Seen 250m runs with ~10 slaves on standard 120ohm cabeling repurposed from RS485 (Profibus) systems. No Problems.
  • Seen 10m runs with ~20 slaves on bulk-standard .5mm² cabeling. No Problems.
  • Seen 1km runs with specialized cable and slaves for lightning protection - expensive stuff. No Problems.

Essence:

Depens on the use case.

Assumptions:

  • You want to control a few Inverters/Motor Drivers from a PLC with a cable run of ~20m
  • Your cables should be run in the same harness as your 400V power bus
  • The harness is not within a drag chain
  • The harness is run through 'common industrial spaces'
  • No chemical standards need to be fullfilled
  • No Safety-ratings need to be fullfilled

I'd recommend:

  • Use a cable according to ISO 11898-2
  • Spend a few $ more per meter
  • Stick to a proper vendor known in industry

Why:

  • The 10$ you saved on the cable is nice...your 1000$ extra work-time is not!
  • You can stock a single "type of cable" for every machine/application you build.
  • Your customers will likely have the same cable stocked already
  • 10$ saved in a cable is nice ... or approx. 0.01s downtime in some applications.

NOTE: This link is not a product recommendation. But, it shows a products used often in automation.

ElectronicsStudent
  • 2,746
  • 5
  • 18
1

I would hope they had insulation on the wires in the first installation. That said, your first installation probably was getting some interference from the power cabling but I doubt that was the cause of the issues you were seeing. CAN bus uses impedance controlled twisted pair wiring with termination. If they were using "regular wire", then it was not twisted, not impedance controlled, and probably not properly terminated either. Twisting reduces the received interference by reducing the open loop area of the wire run, but the rampant impedance mismatch was likely the bigger factor. Differential signals on twisted pair can be surprisingly resistant to EMI.

vir
  • 14,718
  • 13
  • 28