"AC power is supplied to the DUT via a Variac set to approx. 10-15 volts AC output, but if a sine wave generator or similar is available with this level of output, this would be fine too. (Generally use lower voltages for larger caps, to avoid too much loading of the generator and DUT, but very small caps may need higher voltages to give a proper reading)
Regrettably, most audio oscillators I have seen don't give a high enough voltage output.
Readings are obtained by attaching a 'sensing' coil to the probe of an oscilloscope and I have found that a miniature 12 volt relay with a coil of 700 ohms impedance (which I just had to hand) was ideal, but this is in no way critical. I would *guess* that almost any small multiturn coil of wire will indicate some kind of reading when one end (only) is attached to a 'scope probe.
Set scope Probe to X1, Time/Div set at 5mS, and Sensitivity set to 20mV. Again, these settings can be varied, but this is a good starting point to see some readings on an 'average' cap.
The coil should be rested at the side of the DUT, but precise location is unimportant since the two 'opposite' readings should be quite positive. Don't deliberately move the coil position between the two readings, though, as this may distort the results.
At rest (i.e. with no signal applied to the DUT) a clipped sinewave should be seen on the scope, and this waveform is the 'datum'.
Connect the AC generator (using both leads) to the cap, and one of two things will happen to the 'scope trace.
1) The 'scope waveform will increase by a significant factor.
2) The 'scope waveform will stay pretty much the same (possible marginal increase).
If the *difference* between the two waveforms is not very well defined, try to reorientate the cap and/or move the coil about relative to the cap until a very positive difference is seen.
When the waveform is as in 1) above (i.e. larger), the 'live' lead of the generator is attached to the 'outer' cap foil.
In reality, it is a lot easier to do than perhaps the above might suggest, but if those instructions are followed carefully, it will give a satisfactory result. There is nothing critical about the procedure, and having achieved a satisfactory reading once, it will seem like 'riding a bicycle' thereafter, and it merely takes seconds for each cap.
All that is being attempted here is to inject a signal, which is then
picked up by the coil through inductance, to either the outer or inner foil. When the signal is sent to the outer foil, a much larger signal will be picked up by the 'proximity' coil than if the signal is sent to the inner foil, which is partially shielded by the then grounded outer foil. "
link