Zen output capacitors

I've been considering building a Zen (revisited) amp recently, and an idea just hit me. In 'http://capacitors.com/pickcap/pickcap.htm', it says that, in general, higher voltage electrolytic capacitors have less distortion than lower voltage electrolytic capacitors. Considering this, what would be the effect of replacing the two 2200uF/35V output capacitors with eight 560uF/400V output capacitors? Has anybody tried this?

In addition to lowering the created distortion, I think this change would also increase the output current capabilities, as it should lower the ESR and ESL, and raise the safe ripple currents. Unfortunately, it would increase the amount of PCB space used, and it would increase cost.

Thanks

Thoth

[Edited by thoth on 07-27-2001 at 06:11 PM]
 
High and low voltage electrolytics

I don't advocate using electrolytics "directly" in the signal path, but for the record:

Capacitance of electrolytics is a result of the area of the films, the distance between them AND the radicalness of the etch. Low voltage units have short distance between films and thus for a given voltage high forces. They also have flimsy etched out films. It would be safe to assume that they are more easily modulated by audio signal (very bad) than higher voltage units.

Petter
 
Petter said:
I don't advocate using electrolytics "directly" in the signal path, ...

Petter,

Unfortunately, I don't have the option, unless you know where I can find ~4500uF of film capacitor cheap.

The Zen is a single-ended, single-stage, unbalanced, single-rail amplifier. Because of the single-rail, it is necessary to use a capacitor to couple the output. Redesigning the amp to elimininate the output capacitor (use dual voltage rails) is, according to Nelson Pass, not practical.

The good news is that both the output and input capacitors are enclosed within the feedback loop. However, I'd rather minimize the creation of distortion, then try to correct it with feedback. There are four things I can think of that might improve this one component. They are:

1. Improve bypass capacitance. The Zen Revisited has a 10uF film capacitor being used to bypass the electrolytics. This could be increased (multiple devices, not larger ones).

2. Use higher voltage capacitors. This was described in the original post. I think you agreed that it probably would help.

3. Use more capacitors to attain the given capacitance. This was also described in the original post.

4. Finally, increase the value of the effective output capacitor. This could be done using more capacitors, or larger value capacitors. I think there's little question that this would improve the sound, and, probably, lower the distortion.

Of course, there are many places in the circuit where adding capacitors might improve the sound. These include: Filter capacitor on the AC line, Snubber capacitors around the recrifiers, and Bypass capacitors on the filter capacitors. Also, the input capacitor could be changed to film-only.

This project is not at the top of my list, but I'd like to complete it by the end of the year.
 
You could always check out Unilytic(TM) capacitors http://www.ecicaps.com/. They are high-value film, available from Michael Percy http://www.bainbridge.net/percyaudio and are available up tp 1600 uF. Also check out aerogels http://www.cooperet.com/products_supercapacitors.asp -- low voltage, very low impedance, 10F cells, ultra high current ....

Beware of the 1600uF 300V Unilytic cell due to low self-resonance -- the 600uF 300V units seem the best choice. Smaller devices are OK from that perspective, but I have not tested them myself for sound quality.

You could also consider using "novel" devices such as OS-CON's which are available in the audio grade up to 30V. These devices are polarized. More importantly, they use solid electrolyte which should be good for mechanical stability and they are physically small so a stack of them would not build much volume. Try them!

Petter

[Edited by Petter on 07-28-2001 at 10:09 AM]
 
In a similar vein, I've heard several reports that higher voltage film caps sound better than their lower voltage brethern (same manufacturer, design, etc.). I have yet to see a mechanism postulated, although I've argued this back and forth with myself and can give reasons both pro and con...none conclusive. I have not tried it myself, and so have no opinion. As noted elsewhere, I prefer having a rationale that I can sink my teeth into, but don't let that get in the way of my ears if I'm hearing something good. Since film caps aren't etched, it's going to have to boil down to relative thickness of film and/or plates if there's anything to this.

Grey
 
Petter said:
You could always check out Unilytic(TM) capacitors http://www.ecicaps.com/. They are high-value film, available from Michael Percy http://www.bainbridge.net/percyaudio and are available up tp 1600 uF. Also check out aerogels http://www.cooperet.com/products_supercapacitors.asp -- low voltage, very low impedance, 10F cells, ultra high current ....

Beware of the 1600uF 300V Unilytic cell due to low self-resonance -- the 600uF 300V units seem the best choice. Smaller devices are OK from that perspective, but I have not tested them myself for sound quality.

Michael Percy sells the 260uF/600V Unicap for $78.95. Even if the larger, lower voltage units had a similar price (I'd expect much higher), this is far too expensive.

As for the SuperCaps, the highest voltage they're rated for is 5.5V. They won't work.


You could also consider using "novel" devices such as OS-CON's which are available in the audio grade up to 30V. These devices are polarized. More importantly, they use solid electrolyte which should be good for mechanical stability and they are physically small so a stack of them would not build much volume. Try them!

Unfortunately, I can't, as the suggested voltage rating is 35V. With a 34V unregulated supply, I'm suprised the suggested rating isn't 50V.

I think I'm stuck with electrolytics. My goal is to minimize the distortion, while not spending so much that I could have bought one of the Pass X-series amplifiers.

From the comments you and Grey posted, it looks like the high-voltage caps win. Too bad nobody's tried this. That would have been useful information.

Thanks.
 
Zen Capacitor

I was faced with a very similar problem when building the original JLH se Class A amplifier some years ago and went for Elna Cerafine capacitors. They have the capciaty & voltage ratings you require.

i understand these are no longer being manufactured (though i don't know how true this is).

Another possible choice (though more expensive) is to use black gates - these are supposed to be comparable to film caps though i have no personal experience of them (yet).
 
Aerogels and voltage ratings

With Aerogels, you would need to put a stack together. I think they are each about $10 or $20. This is how they do it in electric cars. I am not saying it is great, but hey ...

Looking at the Zen circuit, I would probably not loose too much sleep over running 30V caps in there, at least for testing purposes (and with a fused output). Another way to get around this is to use a lower voltage rated device and stack them up in series -- and paralell a bunch of these hybrid devices.

Another alternative if you find low voltage bipolar devices, is to run split rails and work against ground.

The final alternative would be to run balanced Zen's, no output capacitor needed (and no input ones either if you planned it right). You can accomplish this by duplicating the lower FET + support circuitry and possibly add small source resistors + a small trimmer. NP did it with resistors, but it should work with current source as well and probably perform better. (long-tailed pair on steroids, but watch device polarity! Now you could also feed this with single ended input without problems). I suspect someone has tried this already.

Petter
 
Input caps

If you make the power supply floating, you don't need input caps, especially if you are using current sourced long tailed pair.

That means that you CANNOT hook up the secondary side to utility power ground. I think there is some mention of this in the Pass X600 user manual. Your reference "ground" now becomes the balanced ground you supply from the source component. You can also set up a "phantom" ground to whatever voltage you like using a potential divider and bypass caps (film!). You could also use center tap ...

Now you CAN ground the chassis to power ground if you so wish ...

Petter
 
Replacing non-polarized...

A traditional method of avoiding expencive nonpolarized caps in small signal circuits have been to use two electrolytics and connecting them together + to + , possibly also with a high value resistor ( 1oK ++ ) to the appropriate supply to keep the polarization... There is no logical reason for not connecting the negative poles and biasing to ground in a SE circuit....I have not been working on SE power amps since my very young days, and have not tested this on a modern SE power amp...might be worth trying, as one of the main arguments against ellytics in signal paths have been the lack of proper polarization....the idea is free for testing, I have no operational circuit readily available to test it myself...????
 
From the price they must be metallised polypropylene, but for what it is worth try http://www.maplin.co.uk/mainframe.asp

Product code search for TX84F.

Cheers, Jake

[whoops, didn't see before posting that the 4500uf mentioned was the total value of two coupling caps, nothing there for 2200uf. Only a 4700uf cap at the url. In fact I'm wondering if they aren't really electrolytics in the wrong section of the catalogue]

[Edited by Jakeh on 07-30-2001 at 12:44 PM]