What the ? - Odd Cap Results

[Phrazemaster]

Hi gang, I'm building a U47 clone and decided to be an audiophool and get some high-end caps.

Here is a series of oscilloscope shots from three high-enders. We have 3 caps: a .01uF V-cap TFTF; a .1uF V-cap OIMP, and a 1uF Mundorf Silver/Gold Oil cap.

A pic is attached. Since I can't attach more than one per post, please go on to the next post.

 

[Phrazemaster]

So I wanted to pre-burn these babies in before installing them in the mic I'm building. I read Jim Williams of Audio Upgrades suggested a 10K square wave at 10V into a 100R resistor. Since some of these caps take hundreds of hours to burn in, and since I'm using 50-year old vacuum tubes that I'd rather not put those kind of hours on, I'm pre-burning them in.

So here's the results of an oscilloscope test. The top image is the square wave input. These are actually 3 different outputs so even though they look identical, I'm including them for honesty's sake.

The images below are what the cap/100R resistor combo did to the signal.

Guys, I'm no EE (my dad is), and perhaps I don't know what the f I'm doing. But can anyone explain why in cat's heaven the V-cap OIMP and the Mundorf are destroying the input signal, and the V-cap TFTF (teflon) are not? Could it have anything to do with the value of the caps...same 10v square wave driving each one.

If the OIMP and the Mundorf really are destroying signals like that, could they be damaged caps, or is this what they actually do to a signal? Why would anyone ever use anything but the teflons then...

Thoughts please? Be kind. I'm no electronics genius.

Thanks,

Mike

PS all caps are brand new from the factory.

 

[pucho812]

wait are all the caps the same value? one is a .01uF and the other two are .1uF. different levels of capacitance. not to mention capacitor tolerances. What do the caps really measure out on a DMM vs what is printed on the outside.

 

[dfuruta]

Guys, I'm no EE (my dad is), and perhaps I don't know what the f I'm doing. But can anyone explain why in cat's heaven the V-cap OIMP and the Mundorf are destroying the input signal, and the V-cap TFTF (teflon) are not? Could it have anything to do with the value of the caps...same 10v square wave driving each one.

The higher valued caps are shunting the higher harmonics in the square wave to ground, as they're supposed to do.  Thus, it looks less like a square wave as the cap value gets bigger.  This has nothing to do with the manufacturers of the caps you're using (to confirm this, you might grab some "ordinary", "cheap" caps of the same values, and see how they look on the scope with the same circuit).

EDIT:  is it possible you have your scope set differently in the middle picture?

 

[Andy Peters]

Guys, I'm no EE (my dad is), and perhaps I don't know what the f I'm doing. But can anyone explain why in cat's heaven the V-cap OIMP and the Mundorf are destroying the input signal, and the V-cap TFTF (teflon) are not? Could it have anything to do with the value of the caps...same 10v square wave driving each one.

You have three different-value caps, one is 0.01uF, one is 0.1uF, the last is 1.0uF. The values increase by an order of magnitude. That explains the significant differences you're seeing.

If you're really interested in comparing "audiophile" caps with something more reasonably priced, at least ensure that the values and the dielectrics are the same!

-a

 

[PRR]

> a .01uF V-cap TFTF; a .1uF V-cap OIMP, and a 1uF Mundorf Silver/Gold Oil cap.

0.01u into 100r = 159KHz
0.1u into 100r = 16KHz
1u into 100r = 1.6KHz

Test-tone = 10KHz

It appears you are measuring across the cap.

The 0.01u and 159KHz low-pass hardly rounds a 10KHz square.

The 0.1u and 16KHz low-pass significantly rounds a 10KHz square.

The 1u and 1.6KHz low-pass "integrates" a 10KHz square.

But there's something way wrong about the last, a bit wrong on the first. The rise/fall should be perfect exponentials, shortening to near-linear ramps as F(cutoff) drops below F(test). Instead we see a distinct bend. Also the 1.0u sure isn't attenuating much.

Are the signal generator photos taken *with R-C load attached*? Not all siggens will drive 100 ohms happy, and this becomes more obvious when load is infinite at low freqs but 100r at high freqs. Note that 10V at 100r is 100mA, a huge amount for a signal generator.

The image below shows what *perfect* R-C networks do

Parasitic resistance in the cap or in the test rig WILL give odd zigs and zags.

 

[Phrazemaster]

Thank-you all for the thoughts. It's not that I set this up to compare audiophool caps - I'm really just trying to burn them in and I was curious about what the caps were doing to the signal. Otherwise, of course Andy, it's a silly post. I was just confused at the results - as I said I'm not an EE and I appreciate your insights.

dfuruta, the scope was set the same in the middle pic.

PRR, I took the output of the sig gen into a mic preamp, and the mic pream out. My sig gen won't go much above 7v output anyway; it's a cheapie one from china I think but it works. And yes I was measuring across the caps as you suspected. Is there a different way I should do it?

Thanks all for the thoughts.

Mike

 

[PRR]

Are the signal generator photos taken *with R-C load attached*?

I'd also suggest that most mike-amps are flawed at 10KHz 10Vp-p. That's why this *under load* look is vital.

> I read Jim Williams of Audio Upgrades suggested a 10K square wave at 10V into a 100R resistor.

I can understand Mr Williams not troubling readers' heads with R-C computations.

None of these R-C products "makes sense" in audio. A 0.1u cap would normally drive 1Meg resistance (for 1.6Hz); and so-on up/down for the other values.

I assume the goal is "stress", hence the very low resistance and high frequency.

And that the 100 ohms happens mostly because if any lower, most signal generators would punk-out. (Some will already.)

But there's a 100:1 difference in stress between the 0.01u and the 1.0u. The 0.01u has near-zero voltage across it nearly all of the time. The 1.0u has at least 7V all the time.

> hundreds of hours to burn in

Being dumb, the 0.01u "should take" 100 times longer than the 1.0u because it is loafing, while the 1.0u is passing significant energy and reactive VA per time.

 

[Phrazemaster]

Thanks PRR...as always, educational! I realized I hadn't answered your question; yes the 100r was attached. In series with the cap. Measured across the cap and resistor combo. Should I have measured this in a different way?

Looks like I need to hit the books to understand what values to use to stress the caps. Any thoughts are welcome, and thanks.

Mike

 

[Andy Peters]

Thanks PRR...as always, educational! I realized I hadn't answered your question; yes the 100r was attached. In series with the cap. Measured across the cap and resistor combo. Should I have measured this in a different way?

Looks like I need to hit the books to understand what values to use to stress the caps. Any thoughts are welcome, and thanks.

I think the only stressers would be voltages that exceed the cap ratings, and heat.

Either way the life of the cap is shortened by an unknown amount.

I don't see the point of "breaking in" a capacitor.

-a

 

[dfuruta]

While I respect Mr. Williams's (NB not the same Jim Williams of Nat Semi/LT fame) abilities and knowledge, it's worth recognizing that his business is, largely, founded on expensive capacitors, super-fast opamps and fancy solder & wire;  take it with a grain of salt when he claims radical changes from breaking in capacitors or using silver solder instead of sn/pb.  I've gotten in arguments with him about some of his claims over on the tape-op board.

 

[s2udio]

I don't see the point of "breaking in" a capacitor.

Owww come on a good shot of DC @ rated voltage for 24 hrs lossens the dielectric to allow the capacitor to produce more Air !!
Also Makes for smother transients !

;D