0
\$\begingroup\$

Is there a standard method of schematic design to allow measuring individual resistor values on a PCB? Useful for general debugging, validating a new PCB shop, etc.

Because, if part of your circuit looks something like this:

schematic

simulate this circuit – Schematic created using CircuitLab

And you try to measure R1 on the PCB with a DMM, you are actually measuring R1 in parallel with (R2 + R3 + (R6 + R5 + R4 || R9 + R8 + R7)). You could remove the resistor from the PCB, measure, and resolder but this is time-intensive and not scaleable.

My 2 ideas:

  1. Add a removable jumper on each branch; maybe applicable on a small scale, not so much on a large scale, so the resulting circuit looks like:

schematic

simulate this circuit

Now by disconnecting JMP123, you can measure R1, R2 or R3... Disconnect JMP456 for R4, R5, R6 etc.

  1. Add a sense resistor in line, perform calculations on live circuit using voltage measurements:

schematic

simulate this circuit

If, somehow, you are able to know with certainty the resistance of RS123, RS456, RS789, you can measure the voltage across them, divide by their resistance to get the individual branch current, then measure the voltage across your desired resistor and divide by the current... However this assumes you know + trust the values given for the sense resistors (RS***)... Alternatively, you could carefully design the pcb trace for a given resistance and measure those 2 points to act as the sense resistor instead.

Interested to see others ideas for accomplishing this.

Related, tangentially: How to measure resistance of resistor in PCB?

\$\endgroup\$
  • 3
    \$\begingroup\$ Can you describe a real-life use case where the ability to individually measure resistors on a PCB would be useful? \$\endgroup\$ – Dan Laks Jan 19 '18 at 4:27
  • \$\begingroup\$ It took me far too long to realize that the "DFT" in your title probably means "design for testing" or something of the like, and not "discrete Fourier transform" :) which leads me to the question of in which case your failure model says that it's more likely that a specific resistor will fail, which is pretty much the most elemental of components, than that your specific changes introduced to measure that would introduce additional problems? I think the point of designed for testing is that you consider first what specific type of failure you want to detect and account for that. \$\endgroup\$ – Marcus Müller Jan 19 '18 at 9:23
  • \$\begingroup\$ Having testability for everything is not really a design choice you can make - there simply is too many things in this world. You don't test for e.g. radioactiveness of your PCB. Because that's a) a very unlikely problem and b) the cost/benefit ratio says it doesn't make sense. Now, you say you want to test resistors for the sake of evaluation of new board houses. But that sounds strange - the resistors aren't made by those, and the soldering can better be evaluated with optical inspection. \$\endgroup\$ – Marcus Müller Jan 19 '18 at 9:29
  • 1
    \$\begingroup\$ @MarcusMüller I have actually had to answer the "Does your design use radioactive materials" question (From defense procurement of all people!). It is actually a bit tricky to answer because actually if you use a sufficiently sensitive instrument, most things are somewhat radioactive... We considered answering with the truth ("Yes, but no more so then anything else"), considered the paperwork storm that would result and answered "No". \$\endgroup\$ – Dan Mills Jan 19 '18 at 12:20
  • \$\begingroup\$ @DanMills :D yeah, I can see that! It's like "do you have any explosive material in your check-in luggage" at the airport, and you're supposed to answer "well, I do have alcohol-based deodorant in a pump flask, and frankly, consider the energy stored in a laptop battery! By the way, most flammable materials do explode if brought into a form with much surface and mixed with oxygen at the right ratio..." \$\endgroup\$ – Marcus Müller Jan 19 '18 at 12:23
3
\$\begingroup\$

Most folk add test points on each node and use "golden boards" as references for what the correct design implementation is. If during a test of a real production board the combination of this resistor and that resistor in parallel with another series resistor looks beyond what the tolerances suggest, then one of those resistors is probably wrong and testing another two nodes might pin-point the wrong resistor.

Be practical.

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

Yes, it is possible to measure resistors on a PCB individually, even when other components are connected across them, without needed to open switches, subject to certain restrictions. It's easier if you can physically isolate them, but still possible even if you can't.

This answer discusses it fully. The only extra hardware you need are one virtual ground op-amp, a third test lead, and the circuit diagram.

The restrictions are that a single component directly in parallel with another component can't be disambiguated. In this case, only some sort of switch to disconnect the other component would work.

\$\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.