0
\$\begingroup\$

I have a faulty PCB which is showing an unexpectedly low resistance between certain pairs of bus address lines. I'm measuring a little over 4k ohms between these bus lines, but according to the schematic I should be seeing 44k.

I'd like to try to isolate the fault down to a particular IC or area of the PCB. None of the ICs are socketed, so I'd like to have some idea of where the fault is before I start lifting pins.

I've heard of using a milliohm meter to find shorts in PCBs, measuring resistance between various points and looking for a minimum. But unfortunately as I mentioned above, this fault is about 4k. I'd be looking for a tens of milliohms change in R relative to 4k (roughly 1 part in 100000), and I don't have a high-enough resolution meter to detect that (all I have is a 3 1/2 digit meter).

What would be the most promising direction to take for isolating a fault such as this? I'm thinking something along the lines of a wheatstone bridge + instrumentation amp, but maybe there's a better technique?

\$\endgroup\$
8
  • 2
    \$\begingroup\$ Are you sure the resistance you're measuring isn't from the ICs on the bus? \$\endgroup\$ Jul 25, 2018 at 2:10
  • 2
    \$\begingroup\$ Check the 44k's aren't 4.4k \$\endgroup\$ Jul 25, 2018 at 2:18
  • 2
    \$\begingroup\$ Before starting to lift pins, wash the board thoroughly in clean isopropyl alcohol or in flux removing solvent, and dry/bake it. \$\endgroup\$ Jul 25, 2018 at 2:38
  • \$\begingroup\$ Yes, to re-iterate what Tony said, this could very well be a misload. Presumably the 44k you are expecting is caused by a 44k resistor, or a few resistors in series. If the 44k or several series resistors are the wrong value (off by a factor of 10 is easy to do!) then you may be seeing 4.4k instead of 44k. \$\endgroup\$
    – user57037
    Jul 25, 2018 at 3:20
  • \$\begingroup\$ Does the board work? Do you have more than one board, and are they all like this? Resistances can be unpredictable for a lot of reasons. It is best to compare with a known good board if possible. \$\endgroup\$
    – user57037
    Jul 25, 2018 at 3:22

1 Answer 1

0
\$\begingroup\$

The way I do it is remove the pins attached to the bus one by one (or all at once) until the offending part is found (and in your case the resistance would go back up to 44k). This can be done by unsoldering a pin and then lifting it up carefully. SMT parts will have to be lifted completely. Try to rule out the easiest parts first.

Another way I have done this in the past to find power problems (a short is similar to yours) is to cut a trace (which may not work so well for high speed signals( 10Mhz+), but works great for power an low speed signals. I usually cut the trace to cut the traces attached to the net in half, and then test the two halves. Once the offending part is found it can be replaced.

A hi res IR camera might be sensitive enough to detect a temperature change of the offending part if you have a board to compare this to.

\$\endgroup\$
6
  • \$\begingroup\$ I am afraid I may have to start doing exactly what you suggested (after checking the resistor values and washing the board). The reason I'm hesitant to lift pins is that my faulty board is a "classic" COSMAC VIP microcomputer from the 70s. I'd like to lift as few pins as possible, so I was hoping to somehow "ohm out" the general location of the fault first. \$\endgroup\$
    – CForbin
    Jul 25, 2018 at 4:00
  • 1
    \$\begingroup\$ Any thoughts on the utility of a wheatstone bridge + instrumentation amp? It seems that the difficulty is: instead of a short somewhere, I've got a stray 4.19k. Looking for changes of resistance on the order of 1 part in 100000 might be tricky, even with some sort of bridge/amp circuit. \$\endgroup\$
    – CForbin
    Jul 25, 2018 at 4:07
  • \$\begingroup\$ The time it would take to do this would be more than that of pulling pins, you have to replace a chip anyway, that is the way I would look at it. \$\endgroup\$
    – Voltage Spike
    Jul 25, 2018 at 4:42
  • \$\begingroup\$ UPDATE: PROBLEM RESOLVED! After a long hiatus from this project, I finally found the time to dig into it again and have solved the problem. The board is an old RCA COSMAC VIP 711 microcomputer. It's a 4K system, with a total of eight 2114 DRAMs. Six of these are socketed, and two are not. Here's a summary of the steps I took to solve the problem... \$\endgroup\$
    – CForbin
    Mar 27, 2019 at 1:31
  • \$\begingroup\$ 1. Performed visual inspect of all bus traces. Only suspicious thing I found was some gummy flux. 2. Washed the board with isopropyl alcohol, removing the flux. No improvement. 3. Removed the six socketed DRAMs. No improvement. 4. Removed the first unsocketed DRAM (destructive). No improvement. 5. Removed the second unsocketed DRAM (destructive). STRAY RESISTANCE RESOLVED! \$\endgroup\$
    – CForbin
    Mar 27, 2019 at 1:32

Your Answer

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

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