Capacitance in buttons?

True, but it doesn't explain why it would work most of the time with intermittent failure. Failure being fixed by applying a multimeter.

The multimeter would probably have a 9V battery. When in continuity mode, it would be measuring current to determine continuity. Its equivalent circuit could very well use the full 9V with low resistance of around 470 ohm to do the measurement, hence fixing the problem. Most functions that measure voltage would have high impedance (M?).

Intermittent failures are probably just because the pushbutton is being used out of spec. On top of that, by being normally-closed type, its always relying on the spring for contact pressure. All the user can do is temporarily open the contacts. Alternatively, normally open contacts rely on the user to apply pressure, which can be forced to make better contact. All the spring does here is open the contacts.

A pushbutton with gold plated contacts is a better choice, also self-wiping contacts are better. Keeping within a push-button's specifications is really the best choice.

dlloyd:

True, but it doesn't explain why it would work most of the time with intermittent failure. Failure being fixed by applying a multimeter.

The multimeter would probably have a 9V battery. When in continuity mode, it would be measuring current to determine continuity. Its equivalent circuit could very well use the full 9V with low resistance of around 470 ohm to do the measurement, hence fixing the problem. Most functions that measure voltage would have high impedance (M?).

Intermittent failures are probably just because the pushbutton is being used out of spec. On top of that, by being normally-closed type, its always relying on the spring for contact pressure. All the user can do is temporarily open the contacts. Alternatively, normally open contacts rely on the user to apply pressure, which can be forced to make better contact. All the spring does here is open the contacts.

A pushbutton with gold plated contacts is a better choice, also self-wiping contacts are better. Keeping within a push-button's specifications is really the best choice.

Multimeters don't use 9V for resistance measurements in my experience, more
like 1V. Not enough to light an LED even.

True, but I was referring to continuity test (current measurement). The test voltage would depend on the manufacturer but is usually quite high and test impedance is usually quite low.

For example: In this high end Fluke Calibration Manual (page 14) it specifies:

  • Continuity test indication: Continuous audible tone for test resistance <100 OHM
  • Open circuit voltage: <2.9 V
  • Short circuit current: 220 uA typical

This is "mostly" OK, but notice the 2.9V and low impedance. I wouldn't use it to check parts of a circuit requiring 1.8V power like VDDIN, VDDOUT on the SAM3X or some CPLDs and FPGAs. Cheaper multimeters may use higher test voltages for continuity testing, but this would probably not be specified (if they even had a manual). Easy to check this by using a second multimeter to measure the DC voltage across the open circuit test leads of the multimeter under test that's switched to continuity test mode.

Actually, I don't interpret that as low impedance at 2.9V. Rather, open circuit 2.9V, 220uA short circuit current means about 13.2k ohm @ 2.9V.

In any case, I've not read anything yet tried by the OP ruling out anything, including poor connections to the switch. Spade connectors can get loose, and simply pushing on them with meter leads can make a good connection. I'd want to see him simply push on the wires with something nonconductive while it is acting up and see if it goes away. Check that the spade connectors are on solid.

I can see the advantages in using an NC switch. It should be really resistant to interference when not activated, but subject to constant problems if anything goes wrong.

Yes, I interpret it the same way. Note that this is a high end calibrator to test multimeters and it puts 13.2K in the circuit. This is more common in the field - Fluke 12 Multimeter:
"In the continuity/ohms position, the input impedance of the meter is 2 k?, and LoZ is displayed to indicate that the meter is in the low input impedance mode. In the volts position, the input impedance is 5 M? in ac and 10 M? in dc."

That is more typical, I think, of continuity check.