Power supply filtering

[torcamaniac]

Hi,
do any of you have experience with paralleling small film caps across the big filter caps in amps power supply?
I understand that big caps are used against LF-Ripple and small caps against HF-Ripple.
I've read somewher on the net that many are using this trick in the amps power supply with the rule that the small bypass cap should be at least 1000 times smaller than the big cap.
It should help with noise, improving ESR, reducing sag and as such improving tone???
For example, if I were to put 4,7 - 47nF film across each 47uF power supply filter cap in my Marshall, what would be the effects?
Or is this a bad idea?
What are your thoughts on this subject?

[Leo_Gnardo]

Ran across this suggestion in Audio Amateur magazine, late 1970's. "Helper" caps were intended to make up for the - then - deficiencies in large filter caps to decouple hi frequency signals in hi fi amps. And the ratio was generally 100:1.

I've tried it in a couple instances in guitar amps when I thought the regular filter caps might not be enough. Didn't help much. Some boo-teek amps might have these. If it was such a revolutionary improvement, everybody would be doing it now.

Digital circuitry otoh often does include local bypass caps besides main filter caps. You'll see 0.1 or 0.01 uF caps tacked here & there on some digital circuit boards power supply rails, a cheap way to keep the power supply voltage clean well into the GHz range.

[Enzo]

It wouldn't hurt, but I doubt you would hear anything different. It surely would not affect sag. Its purpose is to remove little high freq transients that the main filter is too slow for.

I would put it on my list of solutions looking for a problem to solve. Do you HAVE a problem?

[torcamaniac]

Hi Enzo,

don't think there's a problem...
Just thought it might improve the amp somewhat.
The only problem if it is one at all is small crosstalk. Its more present when I bias hotter. At loud volume its masked so not a big deal.
I noticed when the gain knob of second channel on JTM 60 is over 12 o clock, the volume on both channels zero and master volume higher you can hear a thin background distorted sound when playing. On both channels

[Enzo]

My outlook is that for improvement, there needs to be some aspect of performance that will be better.

[torcamaniac]

Any thoughts on crosstalk I mentioned before?

[Enzo]

I didn't see mention of crosstalk, but it often occurs in MArshall amps.

[torcamaniac]

I didn't see mention of crosstalk, but it often occurs in MArshall amps.

The only problem if it is one at all is small crosstalk. Its more present when I bias hotter. At loud volume its masked so not a big deal.
I noticed when the gain knob of second channel on JTM 60 is over 12 o clock, the volume on both channels zero and master volume higher you can hear a thin background distorted sound when playing. On both channels

[Chuck H]

Since the problem occurs with more gain I might suspect parasitic oscillation. That can be caused by audio lingering on the HV rail but it's not typical because most amps are designed so shared filter nodes serve out of phase stages.

Nothing to lose by trying the small cap over the big cap. Don't expect much. In fact my humble opinion on the matter is that you really don't want the decoupling to be TOO good sometimes. Little things do happen to the audio signal in the supply rail. See above where I mentioned that most designs share filters on out of phase stages. In this case, if there IS a little NFB going on that is keeping an amp stable then the parallel small value cap would actually do more harm than good! It probably doesn't matter much anywhere after the first amplifier stage, but that first stages gain is amplified many thousand fold in a high gain amp and pretty much ANYTHING can be heard. So small interactions in the decoupling might be responsible for the mojo of certain designs. You wouldn't want to undo that when making an "improvement". Not that I wouldn't use the small cap over the big cap if I thought it was needed or made any audible difference "I" liked. Whatever sounds good IS good. Like Enzo said, and I paraphrase: You should have an actual problem before attempting to fix anything. So try it. If it improves tone or solves the problem, leave the added cap in there. If it detriments anything, take it out.

[trobbins]

A modern electrolytic cap, especially a radial, has low impedance out to pretty much 1 MHz before self inductance starts to be noticed.

A poly film cap will have a particularly notchy impedance, depending on its size, but likely to be around 1MHz (data sheet often shows the plot). So the film cap may have a lower impedance than an electrolytic in the vicinity of its notch.

Impedance notches may not be a good thing, as they can causes quirky behaviour.

So it is pretty much an aesthetic feel good thing to do, based on a mythical expectation that what is good for digital logic PCB operation must be good for lower frequency audio circuitry.

[Old Tele man]

Unfortunately, there's a HUGE difference between digital-signals having sharp/steep rise & fall times and audio-signals which, in comparison, act rather more like modulated DC.

[Chuck H]

...audio-signals which, in comparison, act rather more like modulated DC.

I wonder why they do that?

[Gaz]

You should probably just try it for yourself, but you won't hear a damned difference with the little bypass caps, and it surely isn't the crosstalk problem, which so many multi-channel amps have. Reason is simply bad design, bad construction, or both, whether it be layout, switching circuit, or lead dress. When there's enough gain it bleeds into the other channel and past the channel volumes. It's really common, and I would definitely expect it in a low end amp like the JTM60 (I had one when I was a kid). You're just lucky if yours doesn't have the typical overheating problem those are known for.

[Chuck H]

I've wondered if the little cap over the big cap thing might not help amps like the Fender HR series that have like phase gain stages sharing the same HV filter node. From what I've read here they can develop oscillation problems prematurely and new filters solves that. Not that filters that don't "filter", or decouple shouldn't be replaced anyway, but other amp designs that don't share like phase circuits on nodes seem to keep working happily for a lot longer. The little cap bypass would improve decoupling for the HR amps.

[torcamaniac]

Thanks guys,

I'll leave it for now. Maybe i'll experiment with it next time I open the amp up.
I still want to fine-tune the already modded stuff, but nothing very important...
Amp sounds very good the way it is now anyway.
Gigged it a few times after the mods and the sound amazed everyone...