I've been researching about signal caps, and their impact in sonic quality when used on signal paths.

Its funny that when I headed for my first headphone tube amp, I thought tubes will be the more expensive part... heh, I was completely wrong, caps are probably the most, at least on the amp part...

Threads like http://headwize...?fnum=3&tid=288 and other were very illustrative). Thanks also to Tomo, and pmillett for their advice.

Here's what I think I've learned plus some questions around every "fact".

* General: Do couple (parallel) high capacitance, "low cost" bypass caps with another at 1/1000 or 1/100 the value to enhance frequency response an overall sound quality (enhance dynamics, speed?).

This way, one achieves a cost effective compromise: the smaller, high quality one (usually metallized paper, polipropilene, paper in oil, etc) will be cheaper that this same cap on the higher capacitante value: I freaked when saw a 100uF, 450VDC SCR (Solen) for 48.68 Euros... same brand, 0.01uf/1000V costs 2.98EU; 1uF/630VDC co 4.39EU, all at a "good price" vendor...

Why this proportion? It seems this avoids "spectral overlapping". Is funny that the exact value depends on the source: at TRT, (http://members.aol.com/trtwonder/ , makers of Infinicaps) is no more than 1/100; from most other sources (waarde amp and others reflect this) it seems to be 1/1000.

It also recall my attention this TRT statement: "Note that, in electronics, you're pretty free to vary capacitance values used for coupling, power supplies, etc." any opinions on this? I'm having a hard time to find good quality electrolytics at 220uF for C1/C2 on the Waarde amp. Could this value be eg, the more usual 100uF?

* Electrolytic caps: these in signal paths can (if poor quality) completely ruin the sound, specially as output caps. Good brands for this seems to be Sprague "Atom", Tesla, LCR, Nichicon "Muse" series (KZ,FA,FG,FX, FS polarized, and the ES non polarized, premium KG), Black Gate, Elna "Cerafines". I don't know which "sonic" signature a good electrolytic cap produce, only that a bad choosen one removes dynamics and muffles the sound?

* Good quality, what I call "delicatessen" caps, metallized polypropilene, paper in oil, etc. The one used when coupling with a cheaper, higher capacitance, bypass cap. Examples of this are Jensen/NOS oil caps, SCR (Solen in USA/Canada) MKP, Hovland "Musicaps", TRT "Infinicaps", etc. On the "cheaper" side are Sprague "Orange Drops", WIMA, Azon, etc.

Question: In the Waarde amp, and given my headphones impedance (120 Ohms) I can use pretty much any value from 47uF (cutoff at 28Hz) to 330uF for the output caps. I guess it will be wiser (and convenient, because I have a not very expensive, close source to get them, while electrolytics like Nichicon are hard to get here) to just only use a SCR/Solen MKP at 47uF/630V so costs are in-between; also I'd avoid electrolytics, whose response changes with age and are more prone to fail in the future.

Question into question I read using caps rated at higher voltages than spec'ed could deteriorate them. Does this apply only to electrolytics? Do bypass (film) caps suffer from the same? AFAIK when paraleling caps voltage ratings sums up; If I couple eg a 630V (eg, a 47uF SCR MKP) with a 0.047uF Hovland "Musicap" 1000VDC one, when it was originally spec'ed a couple with voltage ratings of 100V + 250V, will I compromise their life? Note that it seems higher voltage rated ones seems to usually have the best "sonics", even for the same brand/series caps... (unluckily, they're also the more expensive ones

I'm also very interested on hearing the tonal quality these "delicatessen" caps give to sound from people who have actually used them. My goal in this research isn't to end up using the more expensive -while affordable for me- cap, but to suit my choose to my listening likings, in short, to be able to determine how one of these will change the sound.

Thanks for reading until here!

For those interested, I found Michael Percy Audio catalogue (http://www.percyaudio.com/Catalog.pdf) to be very illustrative. For those speaking spanish, Eufonia (http://www.eufonia.com/catalogo.pdf) is a worth look at also.

[Edited by pukka on 01-06-2002 at 07:15 PM.]

This link is much more direct...
http://www.capac...cap/pickcap.htm

[Edited by Gariver on 01-06-2002 at 07:07 PM.]

Yes, it will take you a while to digest all that info. BTW, the article that started the big capacitor debate is: "Picking Capacitors."

In my opinion, the very best capacitors are made of Teflon. However, those babies are extremely expensive. Polypropylene runs a close second, and it is much more affordable. Of course, paper in oil capacitors are highly regarded for tube-amps.

[Edited by Gariver on 01-07-2002 at 12:36 AM.]

I usually try and avoid debates about which (cap, resistor, opamp, 12AU7, wire, chassis, knob) sounds the best. But I'll throw in my 2 cents about caps...

There is certainly a difference - though often subtle - in the way different caps sound. Steve Bench's article gives some insight, as does an application note written by Wima pushing their MKP10 caps (unfortunately, I can't find this on their web site, but they did a nice paper about why polypropylene caps of lowest DF were best for audio).

Technically, if you look at the things that make a real capacitor derviate from the "ideal" cap (things like ESR, dielelectric absorption, dissipation factor [DF], leakage, etc.) you come up with almost the same results as if you rank caps according to their perceived sonic quality - for example, film caps are good, electrolytic and ceramic caps are bad. But there are exceptions; paper-in-oil, an antiquated technology, has some positive sonic characteristics (to some), while technically inferior to some of the plasitc film caps. [Personal opinion: PIO caps sound great in some tube circuits, but (like everything else in audio) it's dependent on the circuit.] Teflon seems to be rated the highest, but I haven't listened to them, and I'm not sure if the hype is real or not. It's certainly one of the most expensive, and techically a very good dielectric.

Anyway, to directly address some of your points and questions... I also believe in paralleling caps, at least when it comes to electrolytics, and also to high-frequency bypass caps. The idea is to make sure that the high-impedance point of the "bad" cap is overlapped by a low impedance from a second cap. I routinely bypass electrolytics with a film cap between .1uF and 1uF - I don't hear any difference between those values, at least with good electrolytics. For HF bypassing, I will often parallel a ceramic, film, and electrolytic (or tantalum). This is true for not only audio, but also high-speed digital stuff.

I have read that some people disagree with this technique, claimning it sounds worse than one single cap.

In terms of sizing coupling caps - there certainly is lots of lattitude available, but remember two things - bigger caps are usually not as good (more leakage, higher DF) as smaller ones, and they will cost you more money. There's not much to be gained by making the cap bigger than you need to have reasonable freq response at ~20Hz. I usually shoot for -1dB to -3dB at 20Hz.

I really don't know if there is black magic in "audiophile" electrolytics or not. It would be interesting to compare a black gate to a really good, low-Z electrolytic designed for a switching power supply application. I know black gate makes some claims about noise and other things. But it is really true that el-cheapo electrolytics are not good caps - just try and use one on the output of a switching regulator! Personally, for coupling I use Nichicon Muse (fairly cheap) and some Elna Cerafine caps, and usually parallel with a polypro cap. For power supplies, I usually use good low-Z electrolytics, always paralleled by film.

I've got a bit of experience when it comes to headphone output coupling caps :)... The best I've found is to use a polypropylene cap like the Solen/SCR. You're exactly right that 47uF is about the minimum value for normal headphones. A 47uF 630V Solen will set you back $20.00 or so, but it works very well. The other cap I've used is Nichicon Muse or Elna Cerafine, paralleled by polypro. To get a high enough voltage, you may have to connect two electrolytics in series, and parallel the whole thing with a polypro cap. Not as good as pure plastic caps, but still pretty good - better than a single "power supply" electrolytic (though the big black gates should be fine). For the electyrolytics on a tube amp, I've used either 220uF or 330uF caps, with two in series, for a total of 100uF - 160uF.

Audiophile capacitors are like wine - the price vs. (perceived!) quality goes up exponentialy! The goal is finding the spot on the curve that youre comfortable with. Personally, I shoot for the midpoint, and for film caps use lots of orange drops, Wima's, Roedersteins, and Solen's, and the occasional Vitamin-Q PIO cap, along with Muse, Cerafine, and plain 'ole Panasonic SMPS caps for electrolytics.

Hope my rambling is useful.

Pete

P.S: BTW, you just missed the Solen's price by $.20! (22EUR)

Pukka: Are you from Spain by any chance? I ask because you mentioned a site in Barcelona, and you just quoted prices in Euro.

Gariver
gariver911@yahoo.com
San Juan, Puerto Rico

Regarding Euros: yes, we're just using them since 2002, right now both peseta and euro are valid currency... what a mess!

Won't you know any reputable place in Madrid to buy caps/resistors? I know places in Barcelona (Eufonia, Amptek) and Bilbao (Iga), but a friend comes this friday from Madrid and that way I could avoid yet another shipment; right now I have spent about 340EU in parts and although a lot of spares will remain, I would like to minimize costs...

These capacitors are however not recommended for signal coupling as their chemistry doesn't like that, so you're stuck with regular electrolytics, and bypassing them is a common practice.

No, I'm not familiar with Spain's audio-component sellers. However, I'm linking Welborne Labs for you. Reasons? They have NO minimum order for Spain, and the shipping costs are based on weight alone. That's very good for you. On the other hand, Mouser's minimum international order is $100. That's way too high!

www.welbornelabs.com/parts.htm

Buena suerte con tu amplificador.

[Edited by Gariver on 01-07-2002 at 04:32 PM.]

I disagree with the opinion of some that the human ear is the most sensitive instrument available. The ear may be very sensitive, but the brain F's it all up. Listening is soooo subjective in any case, but especially when you just spent many bucks on something and want it to work better than your old thing.

Most audio folks listen to their old thing, then plug in the new thing and tell us about the differences they hear. You just can't do it that way and give a definitive answer. It's too dependent on your brains interpretation of what you hear, and all the baggage that goes along with that.

There may be a difference between the sound of caps, but without proper studies, no one knows for sure.

J

I dunno if those kind of orders are taxed here, but wouldn't like to discover that the local adouane have taxed 20% foreign + 16% VAT to my order... like it happened to me with some books in the past.

And, after all, most manufacturers are closer to me than to the States! there must be somewhere to buy near!

[Edited by pukka on 01-07-2002 at 08:16 PM.]

Unlike high-frequency SMPS use (which OS-CON's were primarilly designed for) ESR is not the only consideration for decoupling... ESL comes into play at higher frequencies, causing the impedance to rise at higher frequency.

For very high speed digital switching, it is best to parallel ceramic caps of different values, and sometimes even different dielectrics (X7R in parallel with Z5U, etc.) to get the impedance minima spread around so that you don't have resonances. On CPU boards using 32 bits of 300MHz SSRAM, I've used three different types: .1uF Z5U, .01uF X7R, and 1000pF NPO's, all scattered around. This worked well.

-------

As far as the "sound" of different caps goes - it's the same 'ole "engineer vs. audiophile" arguements that go on ad infinitum. I try and not take sides here, since I claim to be both an engineer and an audiophile... however...

Can I hear all the things that some people claim to hear? Nope. Do I really believe that a power cord changes the sound of a preamp? Nope. Do I think silver wire sounds better than copper? Nope.

But I won't dismiss everything that others claim to hear... my ears (and brain) aren't perfect, any more so than the guys that buy $1000 power cords (well, maybe they are a little more wacky than me...). I still think the bottom line is, do whatever makes you happy. Sometimes I think the folks who are totally tone-deaf have it the best - they aren't compelled to go out and spend money on fancy audio gear!

Pete

Interesting results! Well, I don't mean that they cannot reject the null hypothesis: PP film caps are sonically equal to ceramic caps. Instead I am surprised that, based on the null hypothesis, the test showed SIGNIFICANT (p-value < 0.005) perceivable sonic differences when a HP coupling circuit's 3-dB cut-off frequency increases from somewhere up to 40 Hz.

For me, I am sad to admit that I can bearly hear sinusoidal signals below 30 Hz in my system, which has been tested to have less than 1-dB attenuation down to 10 Hz. And I use Senn HD580, which possesses good low-frequency response. Even at 40 Hz, I just can know there is really something playing, not my auditory hallucination! :)

The test would be of more reference value, in addition to that already recognized, if the background of listeners and the music used for test are posted. If the same result could be obtained by playing solo violin when testing, it would be much more amusing!

Once you get in the sub-30Hz range, it's so much more what you feel by vibration and that's something that headphones have a hard time reproducing.

b

And I would guess that ringtheorist's opinion supports my original point: the test environment and conditions are not fully specified to make its results robust. I do not know if some sub-woofers are used in this test setting, but in another test posted in the same web page, AKG 280 headphones are used.

As I know, few tests done by audiophiles have substantial credibility as they intended to appear to have. They either lack solid statistical modeling or present data without providing crucial side information. In fact, they are in no way comparable to the conscientious evaluation process used in audio industry when developing audio compression or noise reduction techniques like MP3 or Dolby NR.

Probably, these tests could be another sorta, quote Pete's, do-whatever-makes-you-happy things.

[Edited by pukka on 01-09-2002 at 07:46 AM.]

Well, yeah. Why do you ask questions about what you cannot change? We are not interested in recording side of the story. You cannot get better than the quality of recording. All I can do is to get ultimate recordings. (And, I know where to get them too!)

But that is beside the fact. What is important to "us" is how to faithfully playback the recording or how to make it sound good. It has nothing to do with how the music is recorded. (If the recording is bad, it will always sound bad. That's as far as we get to the recording side.)

No, you cannot change how the recording is done. But you can change how you playback. That is where component choices comes into play. But you are right about the fact that because of this definition, each of our personal taste reflect each of our ideas of "good sound."

For examples, I prefer DC coupled right now, but I don't mind cap coupled AC amps. I can craft sound with caps. Pretty neat trick, if you ask me. On the other hand, some of HeadWizers are exclusively DC coupled.

However, there are some fundamental things that we all agree.

1. No big bad electrolytics coupling caps.
2. No low power supply bypass.
3. Power Supply (Either regulation/precision or power capacity. You don't need both, but you can have them. Low impedance!)

The rest is all up for style and taste. 1 is obvious. 2 is pretty clear too. Bass and high will thing out if low. 3 is also important; it is from the definition of "amplifier." So picking good bypass caps are important. OSCON, Tantalum are good choices. But you can take advantage of paralleling to get lower impedance as well.

AC coupling isn't all that bad if done correctly. Things to worry about is interstage coupling capacitor. This can change sound. I would put some "tasty" caps there. Teflon ... I dunno. I am thinking Jensen.

(Voltage/temperature/availability requirement sometimes chooses what dielectric is good. Polystyrene and Polycarbonate run for lower voltage so using them in a tube amp is not really good idea. Polystyrenes are very weak to heat so using them in tight quarter of Class-A amp is a bad idea. OSCON is probably best for bypassing, but you cannot get them for high voltage. So you cannot use them in your PassZen amp. (It only goes up to 20~30V) etc etc etc ... )

Tomo