It can be a matter of sound quality and availability in stock ;-).
Paralleling litle caps with less internal parasitic inductances gives a different sound, compared to a sigle big cap of the same overallvalue. Balance between basses and trebles, on a subjective point of view. Chose your poison.
For the overall value, i do the same, experimental, increasing the value up to the point where more does not seems to improve anything anymore ;-)
A margin can seem overkill, but, as lytics have their values reduced with age and wear...
Oh ! a shunt regulator made of a single zener on a power amp ? No way without a serial resistance before-it to limit the current, .. of too big value for an impedance low enough.
Agreed, that ain't gunna work.
BUT, as a reference, a higher power zener might be a good thing....lower internal impedance/lower intrinsic noise.
Dan.
Esperado, it's a little more complicated than that.
For example. if you use two 10,000 uF caps instead of one 22,000 uF, interesting things can happen when you use two caps but from different manufacturers (eg. 1 from say Fisher & Tausche, and 2 from say Elna).
Sometimes, you can run into troube as well. On occasion, parallelled caps can form a resonant circuit.
That's probably why it's fun finding out.
For example. if you use two 10,000 uF caps instead of one 22,000 uF, interesting things can happen when you use two caps but from different manufacturers (eg. 1 from say Fisher & Tausche, and 2 from say Elna).
Sometimes, you can run into troube as well. On occasion, parallelled caps can form a resonant circuit.
That's probably why it's fun finding out.
The circuit was the Xono a single ended quad Fet input followed by basically a discrete op amp. The input is not something I would be happy with today especially the inverter but it was well regarded and sold just fine. It was my first phono stage. The schematic is hanging around the net.
The supply that people found hard was a regulator before the regulators and the cap multiplier was the better sounding. At that point I stopped but would like to investigate further when I return from the coast.(oh wait CES
) I should measure supplies for noise and sidebands but usually measure outputs. Something like the XP 30 is under 40 uV max gain on the output.
Back to my power supply example. I happened on this combination with the Levinson JC-2 phono stage, 40+ years ago. It is 'primitive' compared with the CTC Blowtorch, but it represents the important areas that I want to discuss. It is important not to miss the FOREST for the TREES. I mean this seriously. Don't concentrate on details, before fully discussing the general areas that contribute to the optimum power supply performance for audio.
For example: What happens if we remove the output cap multiplier? Noise, transients, etc? What are the compromises with a cap multiplier? lousy xtalk, added distortion, etc.
What about a cap multiplier alone? Lousy DC voltage regulation, lousy hum regulation, etc.
What about different active regulators? transformers? cap values? diode bridges?
For example: What happens if we remove the output cap multiplier? Noise, transients, etc? What are the compromises with a cap multiplier? lousy xtalk, added distortion, etc.
What about a cap multiplier alone? Lousy DC voltage regulation, lousy hum regulation, etc.
What about different active regulators? transformers? cap values? diode bridges?
Last edited:
For pedagogic purpose, it would be good to remember that some countries exists outside of US. In fact only 43 countries uses 60Hz and all the rest of the world 50
Anecdote: I just found out that Japan uses 50Hz in one part of the country and 60Hz in another part.
Syncing is apparently so hard, that for one of the high speed trains that start in the 50Hz area, they carry the power feeds along the tracks all the way until the end station in the 60Hz area...
Jan
I feel John is right regarding his warning. All too often, I see circuits made like art for art's sake, other times I see strange looking circuits which work very well nevertheless.
For not too precise regulation, I find a zener assisted cap multiplier work very well indeed. But it is sensitive to transients in contex of required output, meaning it needs to be oversized right off the bat if you don't relish solving exotic problems. For trouble free operation, I would suggest oversize by a factor of two.
For low level circuits, like preamps, a nice shunt will do well both in terms of noise and regulation, and by its nature it is VERY fast.
For complex ciruits, I like to use text book examples of a feedback amp. Effectively, the circuit is exactly like a power amp, input stage a diff pair, VAS a single transistor, a driver and the output device. Fussy, relatively large, but can be expanded at will and even used for the entire power amp.
I'll post the pictures tomorrow, so we all see what we're talking about.
Christoph Esperado is laying low a bit, go to his site and you'll see a regulator for a MOSFET power amp project. Actually, Christoph, I would very much like to hear your discussion on that. I know Nige would be delighted, he's 200% into regulators just now.
For not too precise regulation, I find a zener assisted cap multiplier work very well indeed. But it is sensitive to transients in contex of required output, meaning it needs to be oversized right off the bat if you don't relish solving exotic problems. For trouble free operation, I would suggest oversize by a factor of two.
For low level circuits, like preamps, a nice shunt will do well both in terms of noise and regulation, and by its nature it is VERY fast.
For complex ciruits, I like to use text book examples of a feedback amp. Effectively, the circuit is exactly like a power amp, input stage a diff pair, VAS a single transistor, a driver and the output device. Fussy, relatively large, but can be expanded at will and even used for the entire power amp.
I'll post the pictures tomorrow, so we all see what we're talking about.
Christoph Esperado is laying low a bit, go to his site and you'll see a regulator for a MOSFET power amp project. Actually, Christoph, I would very much like to hear your discussion on that. I know Nige would be delighted, he's 200% into regulators just now.
Yes, that works well for me too.
No decisions until I try at least several variants of the same thing.
"Overkill" is a slippery word. Much depends on your personal view and of course on what you expect to come out of the amp, and in what quality. I arguably overkilled my own Marantz 170 DC power amp, when I took out his original dual concentric 2*12,000uF/56V Elna caps and dropped two BC Components 22,000 uF/63V. That's +83.3% more.
However, I did that only after examining the bridge rectifier and found that it had been nicely oversized right in the factory. I got what I wanted, I could play VERY loud transients with zero variation in sound quality, it just came across louder. Accordigly, in my view, I did not overkill, I did the right thing.
A simple C multiplier using a transistor that's not any slower than necessary, can have better RF PSRR than the vast majority of chip regulators, as I understand. That was one of the ideas behind my Kmultipliers. A single-transistor C-multiplier has the best RF PSRR, but often at this point it's better just to use a low-capacitance inductor, say 5uH or so.
Karl, what people will pick up to some degree is that that recording will deliver a biiiiig sound, if the playback system is up to it. On a cold, very ordinary PC I obviously don't get it happening, but I've been there hundreds of times on other systems in the past - a huge, full, rich soundscape will open up, as big as the sky,Pavel, are you are becoming a subjectivist ?
. This is because back then they weren't into fancy close mic'ing of every tiny, sound producing element in the room, to be be dragged off and endlessly fiddled with, via a million and one enhancing doodahs - the recording picked up the big picture of what was going on musically, and if this is reproduced with decent integrity the result is very, very impressive - it's a completely immersive event ...
Pretty darn simple - minimal audible distortion is the answer. There's a campaign around,
, that distortion isn't as important as other stuff, but IME it's the killer factor - get distortion right, and everything else just evaporates as being subjectively important ...Distortion? What distortion?? You don't call it such, the 'correct term' in these circles is the "sound of the system",
. A system has an expensive sound, a cheap sound, a SS sound, a tube sound, a digital sound, an analogue sound, a horn sound, a full range sound, a ... ... . These are just cute ways of describing the audible signatures of various types of distortion - the only solution, IME, is to remove all these various listening artifacts, and just be aware of the "sound of the recording"; when all is in order this normally always sounds the same, because it is relatively fixed ...
Think you're right. I have many power resistors but I think they will create too high an impedance to sound good.
A Zener is actually not too bad noise wise.
One of the members here measured a whole bunch of them and in the 5-10 V range you are in low single digit uV wide and noise. Follow that with a cap x and you have a very quiet PSU.
For my part, I rectify, regulate and then follow with a cap x and only then combine the two halves. I have not measured the wide band noise on the PSU, but reckon its very, very low. Operate everything heavily into class A, and then any concerns about load reg are also resolved. There's a write up on my blog 'Symphony' preamp about the technique - I call it 'relaxed' regulation.
I think layout and especially local decoupling (damped supply rails) and decoupling ground returns are also critical.
One of the members here measured a whole bunch of them and in the 5-10 V range you are in low single digit uV wide and noise. Follow that with a cap x and you have a very quiet PSU.
For my part, I rectify, regulate and then follow with a cap x and only then combine the two halves. I have not measured the wide band noise on the PSU, but reckon its very, very low. Operate everything heavily into class A, and then any concerns about load reg are also resolved. There's a write up on my blog 'Symphony' preamp about the technique - I call it 'relaxed' regulation.
I think layout and especially local decoupling (damped supply rails) and decoupling ground returns are also critical.
Last edited:
- Status
- Not open for further replies.