Recommended capacitors for AF circuits

Indeed. While I don't go around promoting myself as an expert, I do usually stick to talking about things I know about, and perhaps that gives me a certain credibility. With respect to capacitors and distortion, I've enjoyed Walter Jung's writings on the subject, and I especially enjoyed Cyril Bateman's series on capacitor distortion in the 2002 issues of Electronics World. But I give them credence not only because of what they write, but because of my own measurements, which I'll get to below.

As I've said many times, it's important to distinguish between the contribution a capacitor makes to an circuit when it has no or little AC voltage across it to one when it has a large voltage. For example, consider a simple R-C lowpass filter. At low frequencies where the filter has no effect, you won't be able to measure distortion no matter what type of capacitor you try. This is the common case for a coupling capacitor.**

Likewise at high frequencies, where the capacitor is shorting out the signal, it won't matter what type of capacitor you use. This is the common case for a bypass capacitor.

Sadly, some of the capacitor-distortion measurements I've seen were taken under one of these conditions, and therefore they can't teach us much about the issue.

The critical region for measuring capacitor distortion is where the AC voltage drop across the capacitor is large. For example, in an R-C rolloff filter is voltage drop should be similar to the drop across the resistor, i.e. near the 3dB -45-degree phase- shift point. One does not often encounter such a condition in audio circuitry, but three places one does are tone controls, speaker crossover, and RIAA phono compensation. Folks who have not taken actual AC measurements on capacitors under the -3dB condition may be expected to have all kinds of opinions about it, but surely actual measurements trump their ideas.

Here're some actual R-C filter measurements, taken at 1kHz with a Krohn-Hite 6200B distortion analyzer, on some random parts.

. R C dielectric dist comments . --- ----- ---------- ----- --------------- . 22k 8.2nF polypro 0.002% Panasonic 2% ECQ-P1H822GZ . 22k 8.2nF film 0.003% Phipps & Bird subst. box . 22k 8.2nF ceramic 0.071% CK05 100V MIL jellybean . 160 1uF tantalum 0.046% 50V gum-drop ECS-F1HE105K . 160 0.1uF tantalum 0.040% 50V radial molded Kemet . 160 1uF electrolytic 0.017% 1uF 25V radial . 160 1uF electrolytic 0.053% at 3 kHz, -15dB

As I cautioned Thomas, the common popular ceramic capacitors, which are fine for bypass, etc, are bad news in audio filtering circuits. This shows with 0.07% distortion values. Moreover, the distortion output of the 6200B that shows what's left after subtracting the fundamental, reveals the distortion products contain a number of evil-looking strange harmonics, which no doubt are as evil-sounding as they look.

Although the 0.07% ceramic capacitor distortion is only a few times higher than a LM1036 makes at low signal levels, it's present at all signal levels, and I'll wager it sounds worse.

Although the electrolytic I tried showed plenty of distortion, it had much less than I expected to see.

The film types had distortion values near the 6200B's limit, and the monitor output showed more analyzer noise than signal. So they're pretty good in my book. We'd have to turn to Cyril Bateman's equipment designs to discover their true distortion.

** This means I do not generally subscribe to the notion that selecting the right kind of coupling capacitor will improve an audio amplifier. An exception may be the use of electrolytics for low-frequency signals, where the coupling cutoff is not far below the signal frequency. Is this why one sees 5Hz rolloffs?

--
 Thanks,
    - Win

I'll take your lab data over a troll's blathering any day. Keep up your usual good work, Win! :-)

--
Service to my country? Been there, Done that, and I\'ve got my DD214 to
prove it.
Member of DAV #85.

Michael A. Terrell
Central Florida

As will I. And yes, PLEASE keep up the excellent work. And thank you very much for sharing your knowledge and wisdom.

Thomas

--

A thoughtful pause, then resumes the prince, "We must Learn from 
Yesterday, Live for Today, and look Forward to Tomorrow, for The Past be 
The Present, and No Less The Future."

Thank You. To clarify, by "early on" I mean years ago.

Thomas

--

A thoughtful pause, then resumes the prince, "We must Learn from 
Yesterday, Live for Today, and look Forward to Tomorrow, for The Past be 
The Present, and No Less The Future."

"Winfield Hill"

** Then why the HELL did you direct the OP and others to notorious snake oil filled articles that do exactly that?

You are one monstrous fake and hypocrite.

...... Phil

This doesn't even go into another serious problem with ceramic capacitors in audio circuits (and in MOST OTHER circuits where there might be the slightest bit of mechanical vibration) is that they are MICROPHONIC. Grab one at random, connect it to the input of an amplifier, and hit it with a pencil to hear thumps come out of the speaker. Or connect it to a DSO, and see the largest peak voltage you can get out of it before it breaks.

I really doubt that's THE reason, though doing so does have the advantage you state, lower distortion at 20Hz. I recall that the reasoning for such low rollofs is that an audio signal, from the microphone to the speaker, will likely go through a chain of MANY such single-pole high-pass filters in various pieces of equipment, and the lower the rolloff frequency, the less drop there will be at 20Hz (frequency response is usually advertised to be FLAT to 20Hz).

"RoyalHeart"

** ROTFL

What a gullible MORON.

One of P T Barnam's finest.

........ Phil

I knew what you meant. :) My point is that you don't want to have to learn a lesson like that a second time! ;-)

--
Service to my country? Been there, Done that, and I\'ve got my DD214 to
prove it.
Member of DAV #85.

Michael A. Terrell
Central Florida

Gullible? He's got you pegged for the id10t that everyone knows you are.

--
Service to my country? Been there, Done that, and I\'ve got my DD214 to
prove it.
Member of DAV #85.

Michael A. Terrell
Central Florida

I have just re-read the 1980 Audio articles and that's not what they say, SFAICT. In fact, they're quite clear on the test conditions, "when the signal conditions are such that there is appreciable A.C. voltage dropped across it. Or, stated another way, when its reactance becomes appreciable in relation to that of the load." If some careless readers have taken that beyond its context, which for a coupling cap would be right at the low-frequency -3dB point, it can't be the fault of the writers, since they were clear about it.

They have a graph (fig 3) showing distortion from tantalum caps dropping to low levels at high frequencies as expected, but still creating significant distortion at 5x the cutoff frequency. Readers who use this information to draw their own conclusions will probably correctly decide that it can make a difference. (They didn't publish distortion plots for electrolytic capacitors, which are more commonly used, and which were 3x better than tantalum in my measurements.)

They say, speaking of RIAA equalizers, "It is probably ill- advised to use a ceramic unit for equalization if quality results are to be expected." That's right on target, since the equalizer's high capacitor voltages and distortion occur well within the audio-frequency range. (They could have also mentioned tone controls, potentially a worse problem, given where in the audio spectrum these are used. But haven't most manufacturers been long aware of this, and don't they use film capacitors in such applications?)

The only beef one could have is where, after 17 pages of detailed measurements and good advice about understanding real capacitor distortion and imperfections, all on target, they finish by patting themselves on the back, "the total degree of improvement was greater than any other improvement ever employed." Sheesh, what an over-the-top subjective remark after a mostly quantitative article.** However, I seriously doubt their ending remarks outweigh the detailed preceding 17 pages in the reader's minds. The little self- congratulatory slip they allow themselves after the hard work that went into researching their article does *not* rise to the level of condemning the article as snake oil.

You certainly love to sling gratuitous insults at the drop of a hat. Let's review this conversation, to explore for a monstrous fake. First Thomas asks us about selecting a capacitor for a tone control, one of the few places in an audio system where the capacitor will be operating in the dangerous region outlined by Walter Jung and Richard Marsh. I answer, pointing out the issue, and suggest some types of capacitors to avoid. You insult me and state this is "long ago discredited audiophool rot about 'capacitor sound'" (BTW, references please?). "The real crime is that it is still being cited as if it is 'fact' 26 years later." I respond with detailed measurements showing your errors. You answer, snipping my measurements without addressing them, and assert I am a "monstrous fake and hypocrite".

** These articles are arguably most useful for scientific instrument designers such as myself, and analog engineers here on s.e.d., compared to the average audiophile reader, who doesn't have a reason to care about dc leakage currents, temperature coefficients, and internal series resistance. The audiophile may legitimately be concerned with dielectric absorption, but the article's information is best applied to sample-hold circuits, etc. Recommended.

--
 Thanks,
    - Win

"Winfield Hill" Phil Allison

** The ENTIRE second article is nothing but SNAKE OIL BULLSHIT !!!

The major thrust revolves around the alleged but NEVER established *bad capacitor sound* arising from DA and DF when used in audio circuits.

It is all the most unmitigated crap imaginable.

The title says exactly what the purpose of the two articles is - to encourage audiophools with a hot soldering iron to swap cap types about furiously to make their audio gear "sound" better.

Only a complete *bloody ass* would fail to see that fact.

A posturing fake calling itself Winfield Hill is exactly that *bloody ass*.

** You are a truly MONSTROUS fake - Win.

A revoltingly pompous, endlessly posturing, conceited, pseudo academic twat.

Worse than merely being just another source of bad information - you actively conspire to silence the truth.

And that is a unforgivable crime.

** ROTFL.

As if "Win the Fake" is going to run the case against himself fairly !!!

How conceited can the smug, pommy prick get ????

** Shame about all that DA and DF crapology - is it invisible to you Win ?

Has Win got highly selective vision as well as highly selective judgement ???

What a fake.

Here is the UNMISTAKABLE reply the OP made to "Win the Fake" after skimming the two snake oil filled articles.

" Okay, after perusing "Picking Capacitors" parts 1 and 2 (and which I'll thoroughly read a bit later, when I'll have more time), and per your suggestion, Win, polypropylene it is. Again, thanks for the suggestion, and yet another order to DigiKey is in the works. "

** The OP fell for the audiofool crapology embedded in the articles - hook, line and sinker.

A bloated, poisonous pommy toad fish put him onto the bait.

What a loathsome twat.

....... Phil

"RoyalHeart"

** Thomas - you are anencephalic s*****ad.

PISS the HELL OFF !!!

....... love, Phil

The only fake I see here is you, Phil.

And so Phil "I Am GOD" Allison demonstrates his incessant need to cover up his obvious inferiority complexes by degrading everyone around him.

And Phil, I'll believe a proven liar before I'll believe YOU.

Thomas

--

A thoughtful pause, then resumes the prince, "We must Learn from 
Yesterday, Live for Today, and look Forward to Tomorrow, for The Past be 
The Present, and No Less The Future."

I have small tantalum caps down to 0.1uF, but I've not seen many aluminum electrolytics that small, where'd you get them, how big are they? Are they SMD parts?

That's the same time the gain is at a maximum. One wonders what your load is - perhaps the load current is creating power-supply drops that the input stage can see. The LM1036 has rather poor supply rejection, and would be especially vulnerable to any load current in any of its bypass-capacitor grounding paths. You can add a follow-on opamp to drive your load.

Also, is there any way your input can see some of the output?

--
 Thanks,
    - Win

"Inferiority complex"? No, Phil suffers from "Delusions of adequacy"

--
Service to my country? Been there, Done that, and I\'ve got my DD214 to
prove it.
Member of DAV #85.

Michael A. Terrell
Central Florida

I rest my case.

Thomas

--

A thoughtful pause, then resumes the prince, "We must Learn from 
Yesterday, Live for Today, and look Forward to Tomorrow, for The Past be 
The Present, and No Less The Future."

That too.

Thomas

--

A thoughtful pause, then resumes the prince, "We must Learn from 
Yesterday, Live for Today, and look Forward to Tomorrow, for The Past be 
The Present, and No Less The Future."

"RoyalHeart"

** Thomas - you are anencephalic s*****ad.

PISS the HELL OFF !!!

....... love, Phil

So a question comes up. If capacitors are bad, what is the minimum number you can get away with in the signal path by direct coupling and balanced supplies ? My most recent projects involved video so you could DC restore the signal to get rid of cumulative offsets.

Is there a way to servo the audio to prevent offsets from 'stacking up'? Or would that then get to the 'opamps are bad' too ? I've always been a fan of the minimalist approach to audio and NO TOOBS for me. Lots of watts and no magnetics or coupling caps. OPA2134s work well.

GG