0
\$\begingroup\$

Background: The company I work for makes a variety of special-purpose cable assemblies to go with equipment we make. The people who build these cables aren't doing a 100% reliable job, and nobody really likes sitting there with a multimeter testing them, so I've been asked to make some cable test fixtures to streamline the process of getting these items verified and shipped out the door.

Ok, no problem - for our simplest cables, a power source and a LED or two will make an adequate tester. For somewhat fancier cables, I'm fond of making fixtures that are basically just connectors, a bunch of resistors, and two terminals - if the cable is correctly wired, all of the resistors are in series with the terminals, which can be verified with a single multimeter reading. And for the fanciest ones (like this assembly with 30 pins spread over 5 different connectors), I'm working on microcontroller-based testers with a text LCD to indicate exactly what's wrong with the cable.

However, all of the above is merely verifying electrical continuity, which is hardly the only requirement for a "correct" cable. I have two specific areas of concern (which I believe are closely related enough to belong in a single question) that I am looking for suggestions on how to incorporate into a tester:

  1. Some of our cables incorporate one or more shielded wires. I would like to verify that these have been used correctly - that nobody has swapped the conductor and shield at both ends of the cable, for example.
  2. Some of our cables incorporate an Ethernet data channel, using twisted-pair wiring. I would like to verify that the correct pairs are twisted together. This is a particularly subtle problem - split pairs are not likely to make the cable fail, it will just be slower than it should be due to packet retransmissions.

The microcontrollers I'm using for the fancier testers are PICs (that's what I have on hand, and am familiar with), which often include some number of channels of capacitance measurement (intended for touch switches). That seems like it might be a useful feature here, although I'm not seeing how I could distinguish a center conductor from its shield that way.

\$\endgroup\$
4
  • \$\begingroup\$ Shouldn't you instead measure the characteristic impedance? \$\endgroup\$
    – DKNguyen
    Sep 15 '21 at 1:49
  • \$\begingroup\$ For the last two requirements, consider Time-domain reflectometry. Wiring the wrong pair would reveals itself as slight changes in the pulses on the display. \$\endgroup\$ Sep 15 '21 at 1:55
  • 1
    \$\begingroup\$ The problem lies here: The people who build these cables aren't doing a 100% reliable job - this needs fixing. \$\endgroup\$
    – Andy aka
    Sep 15 '21 at 7:23
  • \$\begingroup\$ @Andyaka every process makes mistakes when you run it enough times. Nobody should complain about extra QA. Of course it might be the problem... \$\endgroup\$
    – user253751
    Sep 15 '21 at 9:15
1
\$\begingroup\$

You should find that the capacitance between two twisted pair cores is higher than the capacitance between random conductors of a multi-core cable.

For screened cables, the best option I can think of is a test rig. The cable could be laid into a slot next to a test wire. Measure the capacitance between each contact on the cable and the test wire.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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