ampage Tube Amps / Music Electronics	For current discussions, please visit Music Electronics Forum.

ampage archive

Vintage threads from the first ten years

Accaptable ripple voltage

:

1/26/2005 11:47 PM

Accaptable ripple voltage

Hi.
I'm building my first valve amp, Doug Hammond's Firefly. Thanks also to Paul Marossy for his excellent layout and info for this amp.

My question is; my calculation of the ripple voltage is that it is very small (due to the small current drain). Much less than a volt. I realise that less is better, but how low is good enough. The caps are a significant cost in this amp. 10uF would cost a lot less and take up much less space than the 47uFs specified.

Any suggestions??
thanks

1/27/2005 6:43 AM

Brett, there's more to the story than just ripple.

And as for the amount of ripple, it's more important how much is in the output rather than how much you measure at the B+ pin. There are circuitry ways that are more or less susceptible to ripple.

In your case, the filters are doing at least two other things than simply reducing ripple. One is decoupling the B+ string. This means providing a low ESR path for the B- and signal returns - the filter cap is like the negative lead that grounds your battery in your car. If that lead were to get a poor high resistance connection then everything gets screwed up. The filter value must be high enough to pass everything from all stages connected at that point easily to ground, plus a fudge factor. Otherwise you'll get squeals and motorboat sounds.

The other is providing a reservoir of energy for when you're cranking the amp and drawing big "gulps" of plate current. If the "hot water tank" is too small then after even washing the dishes there's no reserve left for your shower. The effect with a filter cap is that the B+ sags in deep "nosedives" and highly compresses the signal. Not in a nice way either but excessively so.

You've got the same problem every designer faces - how much is good enough?

If the Firefly is a very simple and low power design with few stages sharing a common B+ node then you might get away with it. Unfortunately I know of no other way to find out than "trial and terror". Designers fudge on the heavy side to keep out of trouble. It's a lot harder to push the envelope the other way...

You can try looking at similar classic circuits. Fender Champs got away with 8 mfd filters. Bassmans needed much more.

One positive note - some caps are smaller than others. Illinois filters are much smaller than vintage Sprague Atoms. Might help to look around.

HTH

---Wild Bill

1/27/2005 4:59 PM
Brett	Thanks Wild Bill I'm surprised that the Champ only used 8uF power caps. The Firefly I'm building is only a 1W design (12AU7 output tube!), and the current draw is tiny (about 25mA max), so it's not going to challenge the power supply much. Anyway, I suppose I'll stay safe and stick with the 47uFs. A few extra $$$ for security, I suppose. thanks

1/28/2005 7:20 AM

OK, my 2 cents

47uF is overdoing it IMO. You didn't mention your power supply voltage, but I'm guessing the main B+ is under 450v, no? Order some cheap 450V Xicon radials from Mouser. 10uF - $.49, 22uF -$.62, 33uF - $1.16. You won't need or want anything higher than 33uF and I doubt very much if you'll want that. So for a grand total of $2.27 you can find out just what you want/need in there. Replace later with a higher grade Nichicon or something-con if desired.

This is what I plan on doing with my project and the very reason I'm trying to keep my main B+ below 450v. My filtering needs will be different than yours since it will be a P-P, UL design with a C-L-C input filter for two 6L6/EL34 tubes. Once you get over 450v, the cost is somewhere between 3 and 4 times as much, depending on what method you use. Makes for experimenting with different uF values expensive and a large pain. HTH

John

1/28/2005 8:27 AM

Re: Accaptable ripple voltage

Power supply ripple voltage falls under the 10% rule so often used in electronics.
So, if the ripple voltage is 10% or less than the maximum power supply dc value, you’re ok.
An easy calculation for Vr:

Vr = Idc/fC

In this case,
Idc = 25mA (your number)
F = 120 hz (Full Wave rectifier assumed , 60 if it’s a half wave)
C = 10 uFd

Then Vr = 20.83 volts

(20.83/265) x 100% = 7.86% (I assumed the 265V after quick scan of the Firefly PSU)

So, 10uFd caps will work fine.
You might like a bigger one for the first stage – your call.
47uFd seems like overkill.

1/28/2005 9:43 AM
anon	My ears can easily hear 10% ripple. If you don't mind lots of hum then go with the 10% rule. If you want to hear no hum then stick with the original values. And the argument that the larger caps are way too expensive doesn't make sense. The cost of 47ufd caps is trivial compared to your time. Do it right and use 47ufd as originally designed for. The designers knew a thing or two.

1/28/2005 2:53 PM

The 10% rule is standard engineering practice.
It is standard practice because you can't hear 10% ripple.
If you hear hum, it is probably because you have way more than 10% ripple, or you are getting filament voltage impressed on a grid somewhere in the front end.
The high voltage on vacuum tube plates is the main reason they have such outstanding hum rejection.
47 uFd is large(ish) for any PSU. Check out any selection of common circuits – the Fender Field Guide for example. You will see higher voltage and current demands (than the Firefly) and less capacitance.
The Firefly has rather undemanding current and voltage requirements by comparison and it displays rather common design elements.
I suspect that the PSU cap selection has much more to do with component availability than design driven selection.
Additionally, PSU caps chosen larger than required can cause inrush current issues that can run from nuisance fuse blowing to catastrophic failures.
It is generally a good idea to be right-sized.

Page 1 of 1