choosing caps in power supplies

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

Mbira

Well-known member
Joined
Jun 4, 2004
Messages
2,422
Location
Austin, TX
I didn't see this in the Power Supply meta. How is the capacitance determined in the caps of a power supplies. I'm looking at a lot of guitar amps these days and trying to figure it out. All the early fenders seemed to use either 8uf, 16uf, or 20uf caps. The voltage of the PT doesn't seem to be a factor in this descision. What is? I'd like to order a quantity of one value that I could use in several tube based projects if possible.


thanks!
Joel


:guinness:
 
You choose the cap that achieves the ripple reduction spec you're shooting for according to the RC time constant....

the caps basically shunt high frequencies (ripple) to ground, the bigger the cap, the closer to 0HZ you get, but they cost more.
 
Guitar amps can be picky if you are looking for the best tone.
 
As a rough guide, use 1,000uFd per Amp of load.

Kev's rule is similar but gives twice the value, often a good idea.

> The voltage of the PT doesn't seem to be a factor in this decision.

Actually it is. The above rule gives about 1 volt ripple. That seems small on a 500V supply, may be large on a 5V supply. Nevertheless, it often works well.

> All the early fenders seemed to use either 8uf, 16uf, or 20uf caps.

Through the 1940s, 16uFd was "large". Anything bigger was very expensive, or assembled from 16uFd boxes (yes, early electrolytics were in waxed cardboard boxes). 40uFd became common in the 1950s. By then you could also get enormous 250uFd 450V "Computer Grade" caps but they cost a week's pay.

In push-pull guitar amps, total power resistance and capacitance affects the peak/sustain ratio. Single-ended amps don't show the same effect.
 
In amps with a tube rectifier, there is a maximum capacitance that the rectifier tube can feed into. The fact that silicon rectifiers are a "newer" development (relative to tube amp design) is probably another reason that early power supply capacitors didn't have as much - uh - capacity.
 
OT: This is kind of funny. I pulled a really old lytic from a chassis the other day and was very suprised at the ratings:

16 uf
600 volts dc!!!

What the.....:? They didn't have that technology back then!
Here is the cap. An Aerovox "Dandee":


dandee_1.jpg


When I opened it up, Surprise!:

dandee_2.jpg
 
THis is great, thanks guys.

I am looking thru the tube data sheets for rectifiers and seeing:

5AR4/GZ34's can handle up to 60uf.
5Y3's can handle 10uf.
6X4's can handle 10uf.

One question-the fender champ schematics I have seen have 16uf caps, but using a 5Y3. What's up? here
is the schematic. My question was probably answered as the curent draw of the champ circuit may not be so great as to need that full capacitance-is that correct?

Joel
 
That max. capacitor value is really just for the first cap in a pi filter, which is the one drawing most of the current at startup. If you use too big a cap, the rectifier conducts too much until the whole supply is up to full power, and it's life will be shorter. If I understand it correctly, the danger is only during startup.

I built several foolish power supplies when I was first starting to build stuff, and one of them had a 500! µF cap immediately after the 6X4 rectifier. Nothing bad ever happened, but I certainly wouldn't recommend it.
 
Joel,

Leo Fender was a master of running tubes outside of their ratings and getting away with it most of the time. Study the schematic for the blackface or silverface Deluxe sometime; that's a circuit that separates the men from the boys as far as 6V6 tubes go. I remember about ten years ago there was a real problem for Deluxe owners because Sovtek 6V6GTs (the only 6V6 type in production at the time) would POP if you put them in a Deluxe. That's 'cause the Sovtek wasn't a real 6V6GT, but rather was a Russian type that was similar to a 6V6GT within its maximum ratings (which were around those of a 6AQ5, 250V max plate voltage). The situation may have improved since then; I don't know because I've been out of the biz for a while.
 
That 10 uf rating on the 5Y3 is just a "CYA" spec. (cover your ass). Since the manufactures back then didn't know all the details of the circuits used, they pick a safe limit for a worst case scenario. Some of the circuits back then were transformerless, so they didn't have the secondary impedance of a transformer to limit things. Also, they did not know what RL was going to be, (total load of the circuit) .
In a guitar amp, there is much more current required. But in a mic pre, eq, limiter, there is substantially less current draw, so you could get away with much more capacitance.
A series resistor, say 10 ohms, is not a bad idea if your worried. But 5y3's are dirt cheap and plentiful, so don't sweat it. I think I have about 25 of them which I will surely never lived to see depleted.

I have run 100 uf on the 5Y3 with no problems.

The GZ has an indirectly heated cathode wich provides a slower warmup, and therefore it can handle more capacitance then the directly heated 5Y3.
This is why they are highly desired. I paid 100 bucks for a Mullard GZ34 (actually labeled RCA) for my Deluxe Reverb, but it sounds great and will last forever.

A champ is a single V6, maybe 30 ma max plus the preamp tubes, so tou won't need a ton of capacitance. Suprisingly, sometimes you can make a muscical amp sound worse by adding more capacitance. Especially guitar amps. So I would stick with the stock Fender value.
 
> 5Y3's can handle 10uf.

Oh, bosh.

On the older rectifiers, anything over 8uFd was unlikely; they cost too much when the rectifiers were first spec-sheeted.

Also any rectifier REALLY loves choke-input, hates cap-input, and for a long time there was a strong prejudice against cap-input designs. And cap-input surge currents are hard to calculate. The semi-authoritative analysis of "how bad it could get" wasn't published until the late 1930s, IIRC.

So they just didn't bother to test and spec the tubes for heavy cap-input duty. In fact rectifiers were generally the no-respect end of the tube business. Some of the classics are just shockingly lazy designs.

As we all know, commercial designs have used 2X and 3X the book values and survived the years. 10X over-size as electronaut has done is pushing things: a worst-case rectifier might have died first-time or in the first few days.

> I pulled a really old lytic ... An Aerovox "Dandee"

That's not even old.

> They didn't have that technology back then!

And there isn't any electrolytic technology for more than 450-500V in a single can. You can't build the oxide layer any thicker. That was true of the orginal wax-box electrolytics, and seems to be true today. 450V is a standard part, 500V can be found at high price, I daresay there are 550 and 600 ratings if you pay for extra time and care in the forming bath. But really anything over 400+V needs series electrolytics or you go over to oil-caps.
 
On the older rectifiers, anything over 8uFd was unlikely; they cost too much when the rectifiers were first spec-sheeted

Well, I'm not sure what "old" means, but all the schematics I'm looking at used 8mfd at first, then 16mfd on most of those early fender designs. I would like to get a bulk order of one value of cap to use in several different types of amps. Should I just go with 16mdf caps?

Joel
 
interseting - i never really thought about how the capacitors affecting the tone doesnt matter as much in SE. hence why everyone puts huge caps in micpres and its no thang..

still, there is some difference in sound as your psu is the heart of the circuit.
I paid 100 bucks for a Mullard GZ34 (actually labeled RCA) for my Deluxe Reverb, but it sounds great and will last forever
didnt realize those mullard rectifiers were worth more than 20 or 30bucks.


And there isn't any electrolytic technology for more than 450-500V in a single can. You can't build the oxide layer any thicker

dunno if they are single can but sprague has an atom thats 20uf@600v
ce makes multican caps @ 525v too.
apparently capacitor technology is still movin along!
 

Latest posts

Back
Top