8

When an electrical circuit is not working properly does it make sense to use a multimeter to measure/test the individual components in-circuit (without soldering them out), e.g. measure transistor continuity test/ fall of voltages?

I am asking myself an in-circuit measure disturbs the result and may lead to false-positive (or false-negatives). What are your practical experiences of finding damaged parts/repairing of an electrical component?

I am asking this as an hobbyist with only basic theoretical and practical knowledge - I am trying to repair some devices at my own and like to get more knowledge regarding electrical engineering.

MRalwasser
  • 739
  • 1
  • 6
  • 10

5 Answers5

12

I always start with a good visual inspection for blown components and bad solder joints. For solder joints give special attention to inductors, transformers and connectors that may see physical stress. Next I start measuring in circuit without power applied. Especially identifying shorted transistors, diodes or blown fuses is easy. When a component looks bad, I go on and if it still looks bad I desolder it to check it again. Only when that fails to find the problem I either scrap the board (depending on how many I have of them) or I desolder more components to check them. Especially electrolytic caps are often bad; when they refuse to charge in circuit and measure couple MΩ out of the circuit it is fit for replacement. Also when they look deformed, replace them.

I personally don't like powering up the circuit for measuring, but that is mainly due to the fact that the most circuits I repair are mains power supplies and those tend to bite when I'm uncareful.

There is no right or wrong in troubleshooting a circuit as long as you are aware of what you may be measuring and as long as you don't blow any other components (or zap/hurt yourself) while you're on it.

jippie
  • 33,033
  • 16
  • 93
  • 160
6

Generally, trying to measure components in-circuit is of little use. If you know the circuit surrounding something, that can help interpret results. Usually for diagnosing a complete circuit it is better to run it, perhaps with some deliberate stimulus, and look at the voltages at various points.

There are some exceptions to the above. For example, resistors in circuit should always read at or below their correct value. If a 1 kΩ resistor is reading 10 kΩ, then you know something is wrong. However, if it reads 120 Ω, then there is no conclusion you can draw. Most of the time you will get the latter, so this method will usually be a waste of time.

Another problem is with any stored energy in the circuit. That can distort readings and possibly cause harm.

Olin Lathrop
  • 310,974
  • 36
  • 428
  • 915
  • Thank you. 1. What about in-circuit transistor continuity test? 2. Does this mean in order to find most component problems you have to solder them out? – MRalwasser Apr 02 '13 at 15:01
  • @MRal: Usually you infer a bad component by looking at voltages in the operating circuit. Sometimes you get a special case where you can measure that a component is broken in the circuit, but most of the time you can't tell for sure without measuring the component by itself not in the circuit. Again, try to diagnose by looking at signals. That usually gives you good clues as to which component is not going its job. Sometimes that is not definite either, so you have to remove components or open connections. – Olin Lathrop Apr 02 '13 at 15:08
  • 7
    Measuring components in-circuit can be useful if one has a known-good board in addition to the board one is trying to diagnose or validate. If a meter across a resistor reports a certain value on a known-good board and a much smaller value on some other board, it's likely something about the two boards is different. Perhaps it has a 1K where a 10K should be, or perhaps there's some other issue, but checking a few spots for obvious major differences can be a useful first step when trouble-shooting. – supercat Apr 02 '13 at 15:09
  • 3
    Remember, you only need to unsolder one terminal of a two-terminal device, or two terminals of a three-terminal device in order to measure it "out of circuit". – Dave Tweed Apr 02 '13 at 16:53
5

If the design is new then testing components can make sense.

If it's a faulty PCB / unit from a proven / working device then I test with the components in circuit. Study of the circuit diagram tells you how the circuit should behave.

Unless the device has been abused in some way the components are rarely faulty if your on the end of a production line. Visual inspection, as the others have said, is your friend here.

Checks I make to start with are:

Check the rails

Any voltage references.(Zenner diodes and IC's)

Is the circuit receiving what is expected?

If op-amp based circuits are used I check the virtual earth's of the inverting circuits. They should be close to 0V or the mid rail if it's a mid rail based circuit. Any that are not I check the outputs are against the rails (If the virtual earth is not 0V / mid rail then the output should be against a rail unless limited by something else). Then I check the input is where it's expected.

Then it's feed in changing signals and see what in the signal path is not working.

The most common faults are...

Bad solder joints.

Solder splash.

Connectors not making contact. (Include connects wired wrong here and not just broken wires)

I once had some jumper pins that were open circuit because the flux had got on the pins and was stopping the jumper from making a physical connection. Both the Test Engineer and I were absolutely stunned. It took a coffee break and ruling out other faults to realise this. Just because it's never happened before doesn't mean it can't happen.

Spoon
  • 1,292
  • 1
  • 8
  • 8
4

I also start with an optical inspection:

  • are there missing parts
  • are there burnt parts or traces
  • are there any obvious mechanical issues (heatsinks, evidence of physical abuse, etc.)

After this, I usually probe around as-is (no parts removed) and look for obvious signs of trouble:

  • diodes that don't act like diodes
  • things that aren't low resistance that appear as low resistance (like actual resistors or capacitors, power supply rails)
  • MOSFET body diodes that aren't there

I also try doing selective power-ups of the various power rails using a current-limited external power supply, to see if a rail is drawing excessive current or no current at all (usually both are "bad signs").

This sort of probing does require knowledge and judgement, as Olin indicated, and as your question stated. It's much harder blindly troubleshooting something without a schematic vs. troubleshooting a familiar or well-understood circuit.

I find that doing some work before pulling parts out often helps narrow down the candidates for removal, saving time and unnecessary wear-and-tear.

Adam Lawrence
  • 32,921
  • 3
  • 58
  • 110
2

I'll answer the direct question as posted: Can you test components (implying component values) in circuit?

The TLDR version: sort of mostly.

There is a technique called signature analysis which applies stimulus to a board while unpowered and then it compares the V,I vs. frequency response (amongst other parameters) against a known good board for evaluation.

One device that does this is a Huntron Tracker which is just one device I'm aware of, I am sure there are others out there.

If your circuit is simple enough then you could apply similar techniques if you have a voltage/Current source (waveform generator) and a Oscilloscope.

placeholder
  • 29,982
  • 10
  • 63
  • 110