2
\$\begingroup\$

I'm seeing several pre-fabricated "CAN-Bus" cables with (2) DB9, 120 ohm terminal resistor connections, while the cable's impedance is 100 ohms.

A. What is the purpose for this mismatch?

B. What potential problems could you see if used in a system that calls for a 120 ohm cable?

C. Typically, 120 ohm cable requires (2) 120 ohm terminal resistors. If the (2) terminals equate to 60 ohms, why would it matter that the impedance of the cable be 120 ohms?

Examples:

  1. Peak Systems; IPEK-0030001 (PCAN-Cable 2) which uses
    LAPP's UNITRONIC® BUS LD Cable (2170203)***

  2. Grid Connect Inc. GC-CAN-CAB-2MT-GC

***LAPP's Data Sheet lists 100 - 120 ohm, but every site lists it as 100 ohm.

\$\endgroup\$
5
  • 1
    \$\begingroup\$ (C) doesn't read well. Please spell it out what you mean or draw a diagram. \$\endgroup\$
    – Andy aka
    May 31 at 15:13
  • \$\begingroup\$ Show us an example i.e. put a link to the datasheets of the cables you mention or at least give mfg part numbers. \$\endgroup\$
    – Rohat Kılıç
    May 31 at 15:18
  • \$\begingroup\$ @RohatKılıç updated with links. \$\endgroup\$
    – joel cox
    May 31 at 19:59
  • \$\begingroup\$ What is the purpose of those cables? Not to build a whole CAN bus, but only connect over short distances (the standard allows for short unterminated taps of a certain number and length)? \$\endgroup\$
    – Peter Mortensen
    Jun 1 at 2:50
  • \$\begingroup\$ @PeterMortensen to my knowledge, there is a recommended maximum un-terminated stub length of 0.3 m... not sure of the qty. \$\endgroup\$
    – joel cox
    Jun 1 at 11:19

2 Answers 2

Reset to default
4
\$\begingroup\$

  1. The mismatch may not have any other purpose than to use a cheap or non-standard cable with 100 ohm impedance such as CAT5 that is easily available, rather than correct cable with correct 120 ohm impedance. For example, fixed DMX512 installations allow to use 100-ohm CAT5 infrastructure cabling and termination still with standard 120 ohm like with RS-485 cabling.

  2. It may not work so well, may allow shorter buses, etc.

  3. Signals don't have infinite speeds. Think signals on a bus like waves of water in a pool, and how they take time to propagate and they reflect off from hard pool walls. When cable is driven from one end with a signal, the voltage signal propagates in the cable with current signal that matches the characteristic impedance. In some languages this is called the wave impedance, as that's the impedance the wave of signal sees when it propagates along the cabling at the wave velocity, about 0.66 times the speed of light. So if the cable is infinitely long, it would look like a 120 ohm resistor if connected to a battery. And since impedance mismatches cause reflections, cable end is where the impedance changes from 120 ohms to infinity if nothing is connected, and the wave reflects back. Only if the cable is terminated with 120 ohms, to a signal propagating in the cable, it looks like the cable is infinite and it does not reflect back as there is no impedance discontinuity. Extend that to a cable that is infinite to both directions, the two infinite cables are then a 60 ohm load. And that's why both ends of a bus must be terminated with 120 ohms so that neither bus end cause reflections that keep ping-ponging between the cable ends, and indeed the total DC load will be 60 ohms when cable is terminated with 120 ohms on both ends.

\$\endgroup\$
2
\$\begingroup\$

To expand on B (@Justme already provided detailed answers for A and C), the cable and termination mismatch is usually worse than slightly off-specs impedance.

The resistors affect performance of the drivers and maximum number of receivers on a bus. Some transceivers can have Rin as low as 10k. Connecting 200 nodes to the bus will then be equivalent of adding 50 Ohm resistor in parallel to the termination. The transceivers have to be able to drive this load to the dominant voltage differential. In short, off-specs impedance will affect the number of nodes and/or the maximum possible bus length.

The mismatch, on the other hand, results in reflections which can make your bus unusable or at least limit the throughput.

\$\endgroup\$
2
  • \$\begingroup\$ There's also another way to look at it. Bus being lower impedance is only a short transient overload on bus as if it had more capacitance, so it's more important to have the rated termination DC impedance. I also recall reading somewhere, that if you have to choose between terminating with lower or higher impedance than cable characteristic impedance, it is usually better to terminate with higher impedance. I can't say by how much, but in any case the reflections will be adding to amplitude, compared to terminating with lower impedance where reflections will be subtracting from amplitude. \$\endgroup\$
    – Justme
    May 31 at 22:37
  • \$\begingroup\$ @Justme while "adding to amplitude" sounds like it may even be a good thing, if we consider multiple time-shifted reflections going from both bus ends and mixing along the cable I think the off-specs but matched termination should look more appealing. But if we absolutely have to use mismatched termination then yes, higher value seems to be better. And it is even better as far as driver load is concerned. \$\endgroup\$
    – Maple
    May 31 at 23:04

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.