Ceramic vs. Polystyrene Capacitor Distortion Analysis

[Mark (Basspig) Weiss]

This is some very interesting results that I obtained while testing capacitors in interstage op-amp circuits.
I breadboarded a quad of inverting op-amp stages and intercoupled them with capacitors for this experiment.

The first trace with the green callouts is four polystyrene capacitors intercoupling. The second is four ceramic disc caps in place of the poly caps. No other changes. I performed this test yesterday at 1KHz and found small differences that only my Tektronix AA501 Distortion Analyzer could detect reliably, but today I used 100Hz as the test frequency and the differences amplified 100-fold.

The ceramic caps produced .216% THD while the poly caps produced .0031% THD at 100Hz.

I also did testing at 10Hz and this produced enough distortion that I could see it on an oscilloscope and see what exactly was happening. The waveform with the ceramic cap became almost triangular with a notch at zero crossing. The poly produced a clean sine wave.

Here's a superimposed snapshot of the 100Hz spectrum analysis. It proves that I was not measuring noise, but actual harmonics. Note that the ceramic has a dominant THIRD harmonic here:

http://i16.photobucket.com/albums/b28/mweiss/CeramicvsPoly100Hz.jpg

So this tends to suggest that the grade of interstage coupling capacitors used in audio amplifiers, particularly at bass frequencies, is rather important.

 

[fas42]

There are big differences in behaviours between the different types of capacitors: in ceramics, there are big differences between the type of ceramic material !! Everything matters, the fussier you are, the more obvious it is ...

Except, are you positive these are polystyrene, not polyester? There are a number of "poly"s around, and they are all different from each other. Polystyrene is getting pretty rare these days ...

Frank

 

[Mark (Basspig) Weiss]

These capacitors were from a bulk stock I acquired in 1984 and are 'new old stock' (NOS). I got four bags of about 100 each in different values up to .47µF. They may be rare today, but were not so much so in the early 80s.

I plugged in a polarized solid electrolytic just to see what would happen at 10Hz. It had less distortion than the ceramic at first, but after a few seconds, the negative peaks were soft-clipping and distortion went way up. Interestingly, it did not show this behavior at 100Hz and above.

I can imagine a multi-stage amplifier or preamp, where this distortion adds up to audible numbers, due to multiples of these capacitors. One also could think about this in terms of what might possibly go wrong with a capacitor, causing it to produce distortion due to a defect or malfunction.

 

[fas42]

Then you're sitting on gold with those cap's! That's the good stuff, forget about the Black Gate carry on, etc. And probably most of the other ultra expensive "audiophile" devices. The dielectric properties of polystyrene are quite substantially superior to anything else, apart from Teflon, so if the units are well made then the superior audio qualities will show through: certainly will trounce any ceramic stuff ...

Frank

 

[Mark (Basspig) Weiss]

Normally, ceramic are used for shunting hash in the power busses to ground, but I've seen some tube gear with them in the actual signal path.

 

[DonH50]

Interesting! Do you have a network analyzer? Be interesting to characterize the caps a bit more...

 

[Mark (Basspig) Weiss]

Interesting! Do you have a network analyzer? Be interesting to characterize the caps a bit more...

Not at this time. I am VERY curious as to what the properties of the ceramics are that cause this phenomenon.

 

[c1ferrari]

Mark,

Thanks for the graph pic and illuminating explanation

 

[andy_c]

There are many different types of ceramic capacitors, and lumping them all together for the purpose of characterizing distortion is a big mistake.

In particular, if you look at the C0G capacitors (which used to be called "NP0", and which have very low temperature coefficient), they will not show this type of distortion, and are in fact among the lowest-distortion capacitor types available.

Cyril Bateman did a comprehensive set of articles for Electronics World in which he measured distortion of many different types of capacitors. He goes into all sorts of detail about this issue. What's going on here is the nonlinear Q-V curve of the inferior ceramic capacitor types.

If you're interested in coupling cap applications, I'm not aware of any competent designer using ceramic caps in this way. Generally, the values aren't large enough. And although polystyrene caps are among the best for distortion measurements, they are rare now, and even when they weren't, they were not available in values suitable for coupling applications either. Anybody who thinks they can find polystyrene caps for typical coupling applications is seriously misinformed.

 

[fas42]

Further to that, and only just now skimmed over it, a seemingly decent overview of such matters is Capacitor Distortion Mechanisms: http://stephan.win31.de/capdist.htm. Plenty of further links therein to continue researching ...

Frank

 

[andy_c]

Bateman's old geocities site is no longer up, but the file he made available for download, containing his Electronics World articles on capacitor distortion can be found here https://rapidshare.com/files/2934180817/cap_dist.zip.

 

[microstrip]

Mark,

You can find detailed studies about ceramic, tantalum, film and some other types of audio capacitors in the reference article PICKING CAPACITORS by Walter G. Jung and Richard Marsh, Audio Magazine, February and March, 1980

http://www.reliablecapacitors.com/pickcap.htm

 

[FrantzM]

Very interesting thread .. Thanks Mark fo starting it. I would be interested to read something about independent and objective studies on some audiophile darlings such as the Duelang capacitors. is there anything special about these? And in what applications?

 

[microstrip]

Very interesting thread .. Thanks Mark fo starting it. I would be interested to read something about independent and objective studies on some audiophile darlings such as the Duelang capacitors. is there anything special about these? And in what applications?

As far as I know, other than a few studies showing that some capacitors (such as tantalum, electrolytic or ceramic capacitors) introduce measurable distortion, you will find no objective and independent studies on the effects of capacitor types in sound quality - illusions (in the audiophile sense of the word) are not simple to prove by scientific methods.

My personal view - there is not such thing as the sound of a component, but there is the sound of component in a defined position of a defined circuit. In other words, there is so single magic, only group magic.

 

[FrantzM]

Micro

Let's not go there... it is clear that the type of capacitors has an impact on the sound and the impact can be measured .. Heck! it is measured in the papers you presented.,, It is as if as if the design of an electrical circuit is magic .. Come on! man! If the Dueland sound and repeatably, there has to be a change of measurable characteristics...

As for topology, It is a truism , Different capacitors have different place in circuitry .. that has been known for a long time. But, given a topology there could be differences between capacitors of the same material and some capacitors may present better results in a given application than others .. Call it be whatever name you will .. Component "Sound" is as good an expression as another...

 

[garylkoh]

Mark,

You can find detailed studies about ceramic, tantalum, film and some other types of audio capacitors in the reference article PICKING CAPACITORS by Walter G. Jung and Richard Marsh, Audio Magazine, February and March, 1980

http://www.reliablecapacitors.com/pickcap.htm

While RelCaps are not "sexy" and the flavor of the day like Duelund, Mundorf, etc. they are consistently reliable and produce among the best capacitors available. They can design and make polystyrene caps in any number and value you like.It is not true that you cannot find polystyrene caps for typical coupling applications. If you are a manufacturer, just call Bas Lim and tell him Gary sent you. For DIY, you may be able to find some of Rel's Multicap series at Parts Connexion. They are not cheap, but they have some of the lowest distortion measurements of any coupling capacitor.

Also, +1 to andy_c's comment on C0G ceramic caps.

 

[andromedaaudio]

I am a fan of duelund /mundorf in the crossover as serial caps , i would love to try the CAT caps though , because i think they are very good after listening to the pre and i know ken stevens makes his own
I havent tried rel yet.
A good circuit is of first and foremost importance

 

[treitz3]

You can find detailed studies about ceramic, tantalum, film and some other types of audio capacitors in the reference article PICKING CAPACITORS by Walter G. Jung and Richard Marsh, Audio Magazine, February and March, 1980

http://www.reliablecapacitors.com/pickcap.htm

Nice link! That said, given the advances within the past 32 years? It would be great to possibly see an update.

 

[andromedaaudio]

http://www.humblehomemadehifi.com/Cap.html This guy makes speakers and has tested a lot of caps

 

[microstrip]

I remembered reading about this project some years ago - it was carried in collaboration with an UK University. Unhappily, as usual, secrecy rules:

UK audio researchers have discovered two of what they suspect will be several subtle links between loudspeaker cross-over capacitors and the audio they produce.

One is the effect of equivalent series resistance (ESR), even when apparently swamped by other resistances. The other is phase change due to mechanical resonance.
“Resonance tends to be between 10kHz and 20kHz, within the range of a tweeter,” said Dodds. “At resonance you get a phase change which affects the tweeter, and sound from the capacitor.”

(...)

ClarityCap has developed an undisclosed technique to measure this phase-change behaviour. “We can get frequency and time parameters,” said Dodds

Continued listening tests have also begun to link audio language to these parameters.

“There is a definite bias to certain use of audio language: like ‘harshness’ and ‘brightness’. It is very, very tentative at the moment,” he said.

Conventional test gear is not detecting the phase changes Dodds has looked at. “It sounds like black magic, but it really isn’t,” he said. “It is just that the ear is better than £5,000 or even £10,000 test gear.”

http://www.electronicsweekly.com/Articles/09/11/2006/40117/Scientists-listen-to-loudspeaker-capacitors.htm