Cap Porn

[boots]

I have been obsessed with coupling caps lately.

Here are a few old Astrons I salvaged from an old Pro Reverb (black face, brown tolex). I have used some of these in my builds, and they sound great to me.

The other day I got around to checking some of them for leakage under about 170 VDC, and most of them leak about 1 to 3 volts. I found a couple that leaked only about 500 mV, and I put them in a 5E3 clone. They sound pretty good, but I am a little worried about the leakage.

For comparison, a brand new orange drop only leaks about 140 mV. How much leakage is too much? I'm not hearing any weird noises from the amp. I saw in some thread somewhere that the old Astrons are known for being leaky, and that most of the ones out there are probably bad.

[xtian]

How do you test for DC leakage?

"ASS-TRON" caps are the best kind for capacitor p0rn.

[boots]

I tested for leakage by borrowing a voltage from an amp and connecting one end of the cap to it - about 170 volts or so from the plate supply of V1. The cap was rated at 400 VDC.

Then I measured the voltage between the other end of the cap to ground. After the initial charge-up on the cap, the voltage should settle at a very low value, if the cap is indeed "blocking" voltage as it should.

I read somewhere that seeing ANY voltage at all indicates a leaky cap. But several brand new caps still show a couple hundred millivolts, so I take that to be an acceptable level.

Anything over a volt probably indicates a leaky cap, and is probably enough to affect the bias on the following stage.

At least that's my story, and I'm sticking to it until somebody educates me otherwise.

[Structo]

From what I have read, leakage should be less than 100ma or so.

[Aurora]

Measuring like that tells you nothing - the open end can float at any arbitrary voltage.
The only way is to connect to ground via a suitable resistor, say 10k. Any voltage developed across the resistor will indicata leakage.

[vibratoking]

Measuring like that tells you nothing - the open end can float at any arbitrary voltage. ...

I would prefer to complete the circuit with a resistor, as mentioned, but the DMM in DC voltage mode is technically completing the circuit. Leakage is not measured in Volts, but Amperes. So divide your DC measurement by the resistance of the meter and you can calculate the leakage. Again, I would just complete the circuit...Either way works.

[boots]

I figured that the DMM completes the circuit as Vibratoking mentioned.

The free end should not float at an arbitray voltage as long as the circuit is completed through the meter?

[katopan]

But quoting voltages as Ok / not Ok is misleading because different meters have vastly different input resistance values.

[boots]

Well, OK, you guys make sense, dammit. I agree that completing the circuit through a resistor is the more proper way to do it.

I re-measured the voltages across all the caps with a 100k resistor in series and came up with very different readings.

The worst-case cap had 25 mV across it, which corresponds to about 0.2 microamps of current. This makes my stash of Astrons look pretty good. The brand new orange drop had essentially no voltage across it at all - maybe a tenth of a mV.

Am I doing this right? I value all your opinions.

Again, the amp sounds pretty sweet with the Astrons in it, so that leads me to believe they are in good condition....empirical evidence.

[Aurora]

A modern DMM has an input restance of typically 5 , or 2-10Mohms, which is way too high to act as a reliable load for such a test.

25mV across 100k gives .25 uamps, which should be OK......

Film caps, in general terms, should show no leakage, even after many years.
Very old PIOs and such could be a bit more dubious....

[boots]

That was going to be my next question.

I have a few PIO caps that I have salvaged from old equipment, and I'm wondering what is old PIO's reputation for leaking? Some of them are the metal can variety with the Micamold label. Others are tubular plastic, made in Japan, and are labelled as paper in oil.

When I get a chance I will test them for leakage.

[boots]

I think these are PIO's?

They are not marked as such, but they have that look to them. I tested these and they do not leak at all. I will pop one into an amp for comparison when I get a chance.

[Cliff Schecht]

Yeah those are PIO for sure, the good stuff. No need to pop one open!

[boots]

Cliff, I'm curious to hear about your experience with PIO's.

I just subbed the Micamold 0.1 into an amp and found it to be very dark and bassy sounding compared to a yellow Astron. I have no way to know how close the value it to 0.1 - I suppose it could have drifted from its intended value.

I need to do some more comparisons when I can crank it up a bit, but so far, I think I prefer the sound of the yellow Astron over the Micamold.

I have a Japanese PIO that I will try tomorrow. It came out of an old Sony reel to reel.

[Cliff Schecht]

First and foremost, my experience with Japanese PIO's is horrible. Every single one I've measured was leaky. Toss these or set them aside, they're probably no good.

My usage of older PIO caps in amps is well documented on these forums. I prefer them specifically because they introduce very little distortion or tonal "flavoring". I rely on the circuit I am building to determine the voicing of my amps.

So with that said, I've used PIO in almost every amp I've built so far as well as polystyrene capacitors because I like how little these technologies "distort". For their time periods, PIO and polystyrene caps were the go-to technologies for the absolute best performance. All of the old USA made equipment I've salvaged (Tek, HP, old military stuff, etc) uses PIO caps and for the critical positions. With polystyrene caps, circuits from the 60's and 70's used these often for timing circuits in RF oscillators. They have the lowest tempco (at the time) of any technology and thus were very stable over a wide range of frequencies.

I do like to use Astrons, blue molded and similar older technology caps in my old amp clones (i.e. tweed/brown/BF Fender stuff) but typically these amps are built with whatever I have on hand that fit this description. Most of my Fender builds are a hodgepodge of whatever I think is cool enough and old enough to fit the bill. They sound like the real versions of the amps I've put them up against so I'm satisfied (although another big part of this is getting the voltages and bias conditions spot on, that's a whole 'nother post!).