[Dozer]

If there are two amps, both have 2000 VA transformers, and one has 120,000 uf capactiance, and the other has 48,000 uf of capacitance... how is this likely to affect audible performance?  

I'm considering a 200 watt integrated amp with a very good reputation, but it only has 48,000 uf capacitance - which seems low compared to a lot of the popular amps.   I know even the stock Stratos has 60,000 and virtually everyone buys another 60,000 as a $200 upgrade.

The question is - what do those extra capacitors give you in performance, generally speaking?    Which is a better predictor of performance, the amount of capacitance or the size of the power supply?

Thanks.

[Sean Parque]

I don't build or sell amps so I may be wrong...but I would imagine they would sound identical up to the point in which the extra capacitance is needed.  Hopefully someone else can chime in with a more detailed answer...

Sean

[audioengr]

The additional capacitance helps to keep the DC voltage to the power transistors from sagging when high-energy bass transients occur in the music.  It is like a series of batteries, with the big D cells being closest to the power transformer and smaller and smaller batteries as you get closer to the transistors.  What this does is to insure that current-draw at all frequencies is possible without allowing the voltage to drop at the power transistors.  The Big caps provide the energy storage for the bass and the tiniest caps, which should be located right at the output transistors provide the energy storage for the high-frequencies.

The reason that the little caps must be close to the transistors is that even small inductance (wiring inside the amp) can cause them to be inneffective.  For the low frequencies, this inductance is less important, so the big caps can have wires of a few inches without any problems.

So you see, the size of the big caps is only a small part of the whole equation.  Case in point is the Adcom amps.  They have great bass response, but dissapointing high-frequency transient response.  This is why I mod them to fix this.

[Dozer]

Hey audioengineer... that's pretty good.   That's basically what the builder, Doug Dale, told me on the phone this morning.  I asked "what audible price will I pay for this, and he said "most likely none, but theoretically a long deep bass transient could sag the voltage" - exactly what you said.

However I have just put this amp (Continuum Unison Integrated) to it's first test and have "floored it" on all kinds of CDs.  It gave all my ears could take, and it's going away the best piece of gear I've ever had (vs Parasound HCA2200II, McCormack DNA 0.5, LeAmp stock and modded, Sherbourn).

I guess the microfarads past 40,000 probably do have a purpose and value - but can be lived without depending on the rest of the design.   In this case, they added a high quality passive line stage with a top of the line volume attenuation chip, which took room formerly dedicated to the other 40,000 uF worth of caps....  as it happens a reasonable swap.  
Anyhow, deal is done.  Thanks for the replies.

http://www.6moons.com/audioreviews/coda/unison.html

[JohnR]

Bigger caps give you more bragging rights

Actually, I think it depends somewhat on the amp, and as Steve says on the implementation. I believe the AKSA FAQ says somewhere that larger caps don't improve that amp and can make it sound worse. Or is that the transformers? Anyway, sounds like you're happy

[mb]

Yes, larger power supply capacitance is something the marketeers love -- "ours has X times the capacitance of brand Y!". As always, it depends, I think.

Some designs will slow slower with too much capacitance... others sound lifeless with less power supply capacitance. There are so many other factors such as local bypass, type of caps, etc, that will vary the result. Yes, in lower-end designs, I believe that there usually is a budgetary or space constraint, and power supply caps are often smaller than optimal. For amps with 2000VA (!!) transformers, it would be silly for the designer to skimp on capacitance, or provide too much and spoil the sound. I'd just go by my ears. In any case the difference in sound is much more likely to be due to many other factors other than PS capacitance alone.

[Dozer]

I wish I could have y'all over to hear this thing (not broken in yet)!   Isn't it great when you score a piece of gear that makes you think that?  I couldn't be more happy.

[Guan]

Yes, larger power supply capacitance is something the marketeers love -- "ours has X times the capacitance of brand Y!". As always, it depends, I think.

Well, my 250 watt stereo amp has 640,000 uF of capacitance...so there! I just shudder if and when I have to change the caps ...

Audioengr makes a very interesting point about the size of caps and how they affect the sound of the amp. I've also been told that using big power supply caps is one possible factor that can cause an amp to sound subjectively slower. My Jeff Rowland Model 8T amp has just 4 huge caps and although it doesn't sound slow, I've certainly heard more agile amps  

I suppose that is one reason why some designs use smaller value caps in a cascade fashion, so they can charge and discharge faster.

In the case of the Rowland 8T, I've read that early models started out with 'only' 320,000 uF of capacitance and used choke filtering. But hum problems in countries using 240V AC caused JRDG to take out the chokes and double the capacitance to achieve the same filtering effect.

[audioengr]

I guess the microfarads past 40,000 probably do have a purpose and value - but can be lived without depending on the rest of the design

I would have to agree. depending on the power rating and speaker impedance.  If you have a true 8 ohm speaker, 10K UFd is probably overkill.  If you have a 2 ohm panel speaker, 100K UFd may not be enough.  If the amp has poor internal wiring, higher capacitance may actually help.  It's a whole lot cheaper to do the wiring right than put in more $100 a piece capacitors.

The Adcom GFA 585 that I just modded only has 36K UFd per channel and it now outperforms the Parasound JC-1 monoblocks, into my 3 ohm speakers anyway.  The Adcom is rated at 400W/chnl into 4 ohms.

[audioengr]

I suppose that is one reason why some designs use smaller value caps in a cascade fashion, so they can charge and discharge faster.

Well, sort of.  The real reson is that the ESR or Equivalent Series Resistance and the inductance of the capacitors in parallel is smaller than a single big cap.  It also helps in some designs because it puts the connections to the caps closer to some of the power transistors, also reducing inductance.  It's a topology thing.

[audioengr]

Dozer - BTW, I know the sound of the continuum gear.  It is designed by the same guy as the CODA and the Innersound amps.  I have three CODA amps that are very nice.  The thing in common between them all is that they do not use any global negative feedback.  This makes them very sweet and nice in the high-frequencies.  

However, this is at the expense of output impedance and therefore the ability to drive very low impedance speakers and also speakers with wildly varying impedances.  You must match your speaker very carefully to these amps.  I found them to be lacking in the tight bass with my 3 ohm KEF speakers.

BTW - the CODA is DC-coupled and I believe the Innersound and Continuum gear is AC-coupled.

[Dozer]

Ah synergy.  My speakers are ACI Jaguars from a kit, 8 ohms nominal, 6.5 ohms minimum.   They're made of the same set of drivers as the original nOrh 7.0.     So I guess I have a lucky combination        I think the Jag is high end enough to let me really hear all of my amp upgrade, but not a complicated load so that sure is lucky.    The system is sounding really good - it's best ever, and no sub needed for music the bass is so rich and punchy.   The bass is sort of "up in the air" rather than just felt through the floor, a feature that has never been present in a system of mine before.    

I bet the Coda guys would be surprised to hear you say they don't drive low impedances well, as they are the OEM for a Innersound which is supposedly great on electrostats....  but I'm starting to talk shit that is over my head here          I can't back that up.... and I don't have speakers which present a tough load so I can't speak from experience.

From the Website description of the Unison Integrated amp:
"Fully discrete FET differential input stage DC coupled to an ultra wideband bipolar output stage"...   So I guess they're all DC coupled - if that is what you meant... I don't know amp topology and I'm not an EE.

Anyway, I urge anyone following this thread to check out Coda-Continuum...  I was at their factory yesterday, and it was pretty exciting.  They have a new person in charge of marketing (no it's NOT me although I do have a lot of enthusiasm for this maker) and they may get some more professional reviews done soon including another Six Moons review of a new dressed up version of the Unison (which I got to see and it is real nice).

Anyway, I have chased down a lot of threads on this subject from every site, and its seems the prevailing wisdom is that performance can't be predicted too reliably on specs - you gotta hear it.

[DVV]

If there are two amps, both have 2000 VA transformers, and one has 120,000 uf capactiance, and the other has 48,000 uf of capacitance... how is this likely to affect audible performance?  

I'm considering a 200 watt integrated amp with a very good reputation, but it only has 48,000 uf capacitance - which seems low compared to a lot of the popular amps.   I know even the stock Stratos has 60,000 and virtually everyone buys another 60,000 as a $200 upgrade.

The question is - what do those extra capacit ...

This is almost impossible to answer. There are several variables to consider here.

More capacitance will make the power supply lines "stiffer", i.e. they will sag (drop in value) less the more capacitance you have. However, for this to be just so, one also needs to have well proportioned power transformers and bridge rectifiers feeding those caps.

The second variable is what do you really need? If your actual power delivery is relatively small, adding more capacitance will have no appreciable effect simply because what was there initially was already an overkill.

The third variable are the filter caps themselves. Obviously, the better quality the caps, the less of them you will need, but the more expensive they will be. Unfortunately, far too many manufacturers do not quote some highly relevant performance parameters, such as voltage (in V/uS) and current (in amps/uS) slew rates. I much prefer one 10,000uF cap per supply line of high quality and speed (e.g. Siemens Sikorel) to several in parallel of lesser quality, because qunatity does not equate to quality. Slow caps, no matter how many there are, will slow down the amp.

The last is how well do the amp electronics actually use those capacitors. If you have a well designed amp, capable of delivering high instantaneous currents on demand, increasing capacitance will push performance further, even if only in extreme cases, when the amp is nearing its design limits, or when it runs into a highly reactive speaker load. For example, even mid-priced audio, as from Harman/Kardon, NAD and Rotel, to name a few, can and do use filter capacitors to best effect, but many others don't. The point is, current limiting occurs in the electronics before the caps become a problem in most cases (but not always, and not by default), becaose puny power devices were used, and there aren't enough of them.

For more on the topic of power supplies, do look over my texts on the subject at http://www.tnt-audio.com and http://www.zero-distortion.com .

Cheers,
DVV

[Ears]

My Legacy [Coda] amp has ac and dc inputs, and Audio engineer...this is the amp that i was going to change the ps caps to blackgates in another thread.
The Piltron 1700 va ...says, custom made for Coda on the side.

Yes theLegacy [Coda] has a sweet top end....a little too sweet for my tastes and this is why i am trying to brighten it up a bit.

After hearing how good the continuum sounds...maybe i will get a passive pre for 2 channel listening, and see if things brighten up.