[dsavitsk]

My preamp has a buffer that is powered by +/-15V. Each channel has its own tle2426 and about 880uF total capacitance (440 on +, 440 on -). To my ear, while it sounds good, it might need a slight bit more bass. Since I have not tested the buffer with a different PS, and nor have I tested the PS with a diifferent buffer, I am not quite sure if this sound is inherent ot the circuit, or if it is sufferent from not enough power.

So, the question, is there a standard way to figure out how much capacitance is optimal, or is it trial and error that I should add a bit more and see what happens? Also, is there an upper limit (and a way to determine it) such that something could go wrong if I add too much? The 440uF on each rail is largly a result of what was in the capacitor drawer when I built it, so there is no real theory there, either. I happen to have 4 extra 1000uF caps, and though I might replace what is there, or just add them as I have the space.

For reference, the rails look like this. Also, while i have the space to add the 1000uF's, that space is after the 0.01uF's if that makes a difference.



(ignore the dots, the are there to get things to line up)

+................-
|-------|(-------| <--220uF
|-------TLE------|
|---|(---|--|(---| <--220uF (both)
|---|(---|--|(---| <--220uF
|---||---|--||---| <--.22uF
|---||---|--||---| <--.01uF
|........|.......|
+.......V........-

 

[mono]

You might need to capture this with a scope to get a better idea of the situation. Personally I'd try to have at least 6-22uF of relatively fast (film or tantalum) caps unless your 220uF are themselves VERY fast types like ultra-low impedance OS-Cons, other soild hybrids or similar. To that extent, using 1000uF may help from (their typically lower) impedance standpoint rather than higher capacitance.

Another issue is where the minimal voltage is for the circuit. Supposing you have a lot of margin below +-15V, a vDrop of a couple V may not be such an impact. I'm leaning towards it being the circuit rather than the caps but I'd do as described above, trying more intermediate 6-22uF fast caps.

 

[dsavitsk]

All the bigger caps are FC's. The .22's are some blue box ones from DigiKey that didn't get used in a Pimeta as input caps.

 

[tangent]

Quote:

Originally Posted by dsavitsk is there a standard way to figure out how much capacitance is optimal

Not really, no.

There is engineering math you can employ to learn things like how much load-modulated ripple your application will produce, and that can be somewhat helpful, but in the end you still have to do the trial-and-error thing because it's fundamentally a subjective task you've set yourself.

The only numerical recipe I can offer is that you should double or halve things like this to get noticeable results. Replacing your two 220s with a single 470 would be silly, for instance. Go for two 470s, or one 1000.

Quote:

is there an upper limit (and a way to determine it) such that something could go wrong if I add too much?

Yes, there are limits.

Some people have reported that the circuit just gets "sloppier" when there's too much capacitance.

More critically, more capacitance means greater energy required to charge things up, and more energy that has to be gotten rid of when the amp powers down. These facts give different effects in different circuits. Some circuits will just lock up if there's too much capacitance. Others will have increasingly large turn-on and -off thumps, which can be dangerous to your cans' health. If there is no output protection on some of your circuits, more capacitance means more current for that circuit to play with during a short, so greater chance of blowing something up.

Is there a way to calculate all of these? Theoretically, yes. But all three of the bad things I allude to above have different math behind them, and some of them you'd have to specify very closely to get any reasonable result. My practice is just to use common sense: if doubling the capacitance doesn't give you an audible result, back it off. If the turn-on and -off thumps get too loud, back it off. If the circuit locks up or starts oscillating...well, you guess.

Quote:

'd try to have at least 6-22uF of relatively fast caps

yes, good plan: parallel some fast caps in there. I'd widen the range to 1-22uF, though.

 

[dsavitsk]

I removed the two 220uF caps form each rail and added a 1000uF FC and a 1uF Wima. There is tons more bass, but so far the sound is not impressive. All of the delicacy and detail and musicality is gone. I have only given them a few hours to break-in, so maybe things will improve. So far, however, I am not impressed. I, of course, did the thing you should never do which is that I changed two things at the same time. I would guess that the change is due to the big caps, but I don't really know. Any suggestions on where to go from here? More small caps, less big caps, both, neither?

-d

 

[mono]

Is it possible the added bass is simply overshadowing the rest now? I could make a similar statement about many devices where I simply turned down/off a bass boost control. You might consider removing the 1000uF entirely, NOT putting back the 220uF caps yet and then checking the sound, proceeding from there to decide what to do next.

 

[dsavitsk]

Quote:

Originally Posted by mono Is it possible the added bass is simply overshadowing the rest now?

I have nbo doubt that is what is going on.

Quote:

You might consider removing the 1000uF entirely, NOT putting back the 220uF caps yet and then checking the sound, proceeding from there to decide what to do next.

If it doesn't settle down in a day or so, I'll give that a shot.

 

[Whit]

Try removing the 1uf cap and leave the 1000uf where it is.

 

[mono]

I still feel it may be useful to have a film or tantalum in a higher-than-1uF value, but that the 1000uf AND the 220uF caps may not be needed at all. I'm suggesting it's possible that a total capacitance of 22uF or less total, per rail, may be all you need.

 

[pbalcer]

Quote:

Originally Posted by dsavitsk I removed the two 220uF caps form each rail and added a 1000uF FC and a 1uF Wima. There is tons more bass, but so far the sound is not impressive. All of the delicacy and detail and musicality is gone. I have only given them a few hours to break-in, so maybe things will improve. So far, however, I am not impressed. I, of course, did the thing you should never do which is that I changed two things at the same time. I would guess that the change is due to the big caps, but I don't really know. Any suggestions on where to go from here? More small caps, less big caps, both, neither? -d

To me it looks like right now you have better rail decoupling than before. Maybe what you refer to as "delicacy and detail and musicality" was a result of the distortions caused by inadequate rail decoupling? The distortions produced by the amp define its sonic signature. The amp without distortions would just simply sound transparent/clinical/whatever-you-might-call-it (some people love transparent sound). Maybe you just liked the type of distortions that were present before?

It looks like your circuit can lack bass - which you don't like, and distort signal in a way that you like, or can have improved bass response - which you like and reduced distortions - which sounds less enjoyable to you right now.

Maybe you should try another (different) amp circuit to see if you can "eat the cake and still have it" - good bass and slight sound coloration that you might like...

 

[dsavitsk]

No, at least so far, it is not that it is more transparent. It really just got muddy and sloppy.

My personal sound preference is a tubey sound, which was not how this circuit was. This circuit was increadibly detailed and transparent -- very good solid state. It was not my cup of tea, but it was undoubtedly clear -- just a little bass shy. Now it is bleah.

-d

 

[pbalcer]

Quote:

Originally Posted by dsavitsk No, at least so far, it is not that it is more transparent. It really just got muddy and sloppy. My personal sound preference is a tubey sound, which was not how this circuit was. This circuit was increadibly detailed and transparent -- very good solid state. It was not my cup of tea, but it was undoubtedly clear -- just a little bass shy. Now it is bleah. -d

Maybe your 1000uF caps are not as fast as the caps you were using before?
Good power decoupling should use low ESR caps (it is amazing how many designs use good audio-quality caps in the signal path and lousy caps in the power supply... - just a digression).

But it also depends on the circuit as well.

The amp I built for myself is a single-ended class A design that doesn't require very low ESR because it has current source between the signal and power rail (current source has very high impedance for AC).

Most symmetrical designs work in push-pull mode so the AC is always being decoupled by power rail caps - that's why you need good audio-quality caps on the power rails to provide optimal performance.

I also prefer tube-like sound. In case you would like to see my circuit, I put the info on my amp on the web site:
http://www.geocities.com/bpiotrek/ha1_e.html

 

[dsavitsk]

Thanks for the link. This circuit is also discrete class A. It is only a buffer, however, fwiw. Also, I am using it as a preamp, not to power phones.

Anyhow, the sound is settling down quite a bit, so I am going to leave it for now. I have the sense that with the quality of parts I am using (FCs), I am just trading bass and clarity and that it is a matter of finding the spot that I prefer. Probably better quality caps would help.

-d