What determines your B+ filter capacitor size?

[Steveo]

Hello, does anyone know what determines the different sizes of filter capacitors in an amp? Im talking about the filter caps right after the bridge rectifier. My rivera has 2 330uf caps in series for a 165uf total, my framus had 2 220uf caps for 110uf and a 1959 marshall seems to have 2 dual run 50/50 caps in series for 50uf total I think, is that right?
Anyways, these are all 100 watt amps but with quite different values of capacitance after the bridge rectifier. Why so different?

[xtian]

The basics are, more filtering gets you lower AC ripple (heard as hum) and more stable DC supply for the tubes (so, less sag).

(One sentence and out. Mic drop. Bye. Gotta make dinner!)

[Steveo]

I understand how they smooth out the ripple and can prevent sag but does the overall size have something do with the transformers? For example ive changed out my riveras to 390uf from 330uf. It was great for riffs but felt too tight for solos. Then i tried 220uf it was great for solos but sagged too much for riffs. But that still is alot more capacitance than most marshalls but i dont think say a jcm800 with 100ufs sag as much as my rivera did with 220uf. The output transformer on my rivera is really big though and the amp has tons of bass even with the cathode resistors changed out to more marshall like values (2k7/.680). It leads me to believe the bigger bass response from more iron in the OT needs more filtering to keep it tight, maybe? I imagine theres some formula for this.

[xtian]

You are not going to find formulae that describe the reservoir cap's effect on bass response because the filter caps are not in the signal path.

Also, I would bet cash money that you were letting your preconceptions guide your ears when you were playing 220/330/390uF reservoirs, and hearing any difference in sag. Let me point out amps I've worked on that are known for sag: Fender's Deluxe (5E3) or Champ (5F1). Both have tube rectifiers followed immediately by a 16uF reservoirs that feed the plates. When turned up loud and you hit a power chord, you can easily notice the sag. But it has to be loud in order for the current demand to drain the reservoir cap.

The "size" of the power transformer, that is, its ability to deliver current, is also part of the equation. Both fender amps have very small iron. And we know that bigger, over-spec'd iron can improve bass response.

I don't know which Rivera you have, but looking at the Knucklehead, I see it has solid state rectifier with 330/2=165uF reservoir. That's stiff! I can't believe you're hearing sag, unless you're using an attenuator, and you've got the amp cranked way up loud.

That said, you're the one in the room with the amp, so let me know what you think.

[wpaulvogel]

What you are going to find is that the capacitance curve is exponential, similar to wattage/dB. I find that the sag is more contributed to the current capacity of the power transformer than to the capacitor. A really strong power transformer will keep up even with small capacitance but the ripple will be greater. A weak power transformer teamed with huge capacitance will still fall on it’s face with any sustained demand. I use very beefy power transformers in all my custom circuit builds.

[Steveo]

The 220uf caps really sagged alot when playing loud. I didnt notice at small volumes so i thought great this feels really good im going to keep these but when i cranked it and played some lower tuned riffs loud it was very noticable. The 390uf werent fun to play too stiff, 330uf stock seems right. Your right they dont sag. It is a knucklehead 3, it has a very big low end, great amp imo. I mainly ask this question because i want to make my own amp soon and i haven't seen much discussed about this.

[Roe]

sag is desirable for blues but not so much for metal. it really depends on the application. you can go very low on the mains filtering of a push pull amp if the power amp is perfectly balanced

[roberto]

Larry Novosibir used (still?) to install a Rheostat (and then a capacito) from the main power supply to supply to the screens, in order to adapt the sag of the screens to what was played with his amp at the moment.
It gives a bit of "blanket" effect IMHO, but could be what you need.

[pdf64]

I suspect that if you can detect more than negligible difference in the amp’s performance between 390uF and 330uF series reservoir cap pairs, then there’s probably something else at play. eg the 330uF caps are aging, or otherwise have higher esr than is normal.

The older an amp’s design, generally the lower value its HT caps will be; as higher values would have been unacceptably bulky / expensive / not commonly available, back then.

Note that tube rectifiers require some degree of resistance in series with their plates, in order to limit current, if used with cap input filters. Back when tube rectifiers were the only or a common option, PT design ‘rules of thumb’ incorporated a suitable degree of resistance into the HT winding, to assist equipment designers, saving them the need to add discrete resistors. Hence PTs designed in that era for this application tended to have poorer regulation (saggier) HTs (and so run hotter) than is now feasible / typical.

[Steveo]

Thanks for all your help here. I might have to do some experiments in the future with this. I wish there was a way to have lower frequencies use higher filtering and higher frequencies lower filtering within the same amp somehow. Thanks again.

[roberto]

I suspect that if you can detect more than negligible difference in the amp’s performance between 390uF and 330uF series reservoir cap pairs, then there’s probably something else at play. eg the 330uF caps are aging, or otherwise have higher esr than is normal.

What I thought when I saw his statement is that most probably he tried a 330 uF -20% = 260 uF and something like a 390 uF +20% = 470 uF, so something very different but still within tolerances. This can make a clear difference, but not two real 330 vs 390 uF.

[Steveo]

Your right there wasn't much difference between the 330s and 390s. It just felt like high notes weren't sustaining much at all but yeah it could have been other things. I guess my main question is like why does a marshall 100 watter need only 50uf and my rivera has 165uf. I'm thinking its all up to the what we need out of the amp, no universal formula for it.

[Phil_S]

If I'm not mistaken, the Marshall 100W is tube rectified. It is the rectifier spec that limits capacitance to 50uf. Check the tube data sheet and you'll see. Solid state rectification generally frees you from such limitations.

[roberto]

Phil,
JTM45 is tube rectified, all Marshall 100 W are diode rectified.

Steveo,
What Rivera do you have?

[wpaulvogel]

I think that with the 100 watt Marshall they were still in the experimental stages of development as a company and they had 4 cans lined up there for the plates and screens and everyone was going bonkers over the amp. When they did the 200 watt amp the ultralinear transformer took away the screen filtering and allowed bigger caps for the plates. The 50uF thing was probably the best they could come up with at the time and if it ain’t broke, don’t fix it.