I used the attached circuit to test some 4700μF capacitors that I pulled off of a project I screwed up. I just wanted to make sure the values looked reasonable after the soldering and desoldering so that I can put them on the replacement board and not buy new ones.

My concern is that they all measured almost exactly the same, about 4300-4400μF, so I wanted to make sure I didn't mistakenly do something that would give me false readings. The consistent values seem suspicious. I measured 10 total from two types: one a 50V high ripple current model, the other a 35V lower ripple current from a different manufacturer. 8 of these caps I pulled from the board (which was only briefly powered up before a regulator blew because of a design mistake), and the other 2 were spares that had not been used at all. I used the attached circuit (using a different 4.7V zener) with 1mA and 10mA test currents, using my oscilloscope's voltage measurement to get the dV/dt.

Is this just chance or am I missing something?

 

Your circuit has a few problems.

The current source should have an NPN follower, not PNP.
Switch M1 will stop conducting as the capacitor voltage increases, as the voltage on the source approaches the gate voltage.

As is, it's not going to perform as you intend.

 

It performed exactly as I intended, it’s just the consistency of the measurements I question. M1 continues to conduct because the high gate voltage is 10V and in a 1 second period the capacitor voltage doesn’t go up enough to cause it to shut off. However, M1 is probably unnecessary anyway.

 

Getting cap readings within 10% of their stated value seems pretty good. What tolerances are they supposed to have?

If you're unsure that your test rig is working, why not try it on some different caps with different values? Do you have any caps with especially tight tolerances you could try to measure with your rig in order to get a better sense of your rig's accuracy?

 

One problem you have that might be giving you a systematic error is that your 4.7 V reference is relative to ground, but your current source feedback voltage is relative to the positive rail. The result is very poor power supply rejection.

Why not refer the zener to the positive rail and then use a precision 4.7 kohm resistor instead of a non-standard 5.3 kohm resistor?

 

Since I put this circuit together just to check these capacitors and not for long-term use, I employed trim pots to get exactly the values I wanted and checked the zener and supply voltages to make sure they matched my simulation exactly. Before testing the capacitors I used a resistor load to verify the function, and it was within a reasonable margin of error for something I didn’t put much thought into.

I wasn’t shooting for precision measurement, but the consistent measurements across 10 capacitors that have 20% tolerance and are from different manufacturers seemed suspect to me. Since I have a tendency to overlook things, I thought I’d ask here.

 

How are you making the actual measurement. Over what time frame are you measuring dV/dt?

 

1 second. I set a function generator to 0.5 Hz for the FET gates.

 

One thing that comes to mind -- are you sure your scope is in a calibrated mode and that its calibrations are still good?