Geeky E-Cap Measurements

[David Root]

I just got five E-Caps from RJ. Two new current production, three NOS from the '70s and '80s. I measured actual capacitance (Fluke 187) and ESR (Anatek Blue ESR Meter). Here are the results.

1). Tech Cap 47uF/500V 51.8 uF 0.76 ohm

2). Mallory TC78 40 uF/450V 50.8 uF 0.46 ohm

3). 1985 CDE type WBR 40 uF/450V 58.4 uF 0.31 ohm

4). 1976 Sprague TVL 1723 Can 40 uF/450V 46.5 uF 1.4 ohms

5) 1973 Mallory Black cardboard 20 uF/450V 33.0 uF 0.44 ohm

The NOS were not reformed, just as I got them. According to my meter, anything below 1.0 ohm is good on a 500V cap, and anything below 8.5 on a 450V cap.

So the Trainwreck guys that say use only 40 uF, not 47, should maybe take another look?

[diagrammatiks]

it actually looks like all the nos ones should be reformed.

only the tech-cap is close enough within tolerance for use.

On the sprague tvl 1.4ohms is horrendous for a 40uf cap.

[David Root]

I will be reforming the old ones. Would you expect their capacitance to drop closer to nominal after reforming?

#2 was a bit of a surprise to me. I don't think I have seen a new current production e-cap measure 25% above nominal, although the specs for this one are +50% -10%, so it is within mfr's spec.

What do you consider is acceptable ESR for a 40 uF 450V cap, and why?

Most of the very large value e-caps (200 uF and greater, 450 or 500V rated) I have bought were actually 10-15% lower than nominal. Again, all new, current production. Any comment on that?

[Structo]

I wonder why the tolerance on E caps is so bad?

[diagrammatiks]

I will be reforming the old ones. Would you expect their capacitance to drop closer to nominal after reforming?

#2 was a bit of a surprise to me. I don't think I have seen a new current production e-cap measure 25% above nominal, although the specs for this one are +50% -10%, so it is within mfr's spec.

What do you consider is acceptable ESR for a 40 uF 450V cap, and why?

Most of the very large value e-caps (200 uF and greater, 450 or 500V rated) I have bought were actually 10-15% lower than nominal. Again, all new, current production. Any comment on that?

when caps leak they'll go way above nominal and the esr will increase as well.

in terms of esr for a 40uf 450v cap...

how does your meter measure the esr?

It's also important to check for leakage as well or else all the numbers will be off.

[jelle]

Agreed with Structo...

What is the leakage?

If you plot the ESR from 0 - 20kHz, what does it look like?

[XgamerGt03]

I did some testing like that and got interesting readings on the Sprague Atoms. Most of the 40uF caps I had were around 47uF as it was. My guess is that they just sort out caps from a line to put into the Atoms.

The ESRs of the caps were all around .5ohms too.

The way I checked ESR was to use a 1kHz pulse wave with a 5 microsecond on time. I could see the voltage drop across the cap and compute the approximate ESR since I wasn't charging the cap up at all. I'm fairly certain a lot of ESR meters function similar to this.

[David Root]

My meter's manual says:

(1). Resistance Measurement: as stated previously, this meter is really an AC ohmmeter with an equivalent test frequency of about 100kHz and capable of measuring non-inductive resistances from 0.01Ω to 99Ω.

(2). Basic Signal Generator: the meter’s test signal is a 500mV P-P (open circuit) burst of 8 usec pulses at a 2kHz rate, repeated several times per second.

I can't test DC leakage of a not in circuit cap, so I don't know. All three NOS caps look normal, no bulges, physical leaks etc.

I do have a signal generator so I could run different AC sinewaves through them I suppose to see what the ESR curve looks like. Any established procedure I should use? Don't want to reinvent the wheel here.

[diagrammatiks]

i've been looking at a lot of data sheets, it seems that in order to match up numbers...the test needs to be

a 100hz test at 20 degrees celsius.

a 100khz wave at 20 degree celsius is used for impedance.