Splitting Power Supply Filter Caps For Smoothing and Ripple Reduction

[TomCarlos]

Forum Members:

I am studying the Fender Deluxe AB 763 Schematic and looking at the Power Supply Caps - so let me see if I have this right...

I know some amps use a technique of splitting caps by using series or parallel combinations in order to increase voltage handling capabilities. For example, the Deluxe uses two 16uf caps (450 volts) in parallel that yields 32uf. (This is shown on the attached schematic). An alternative to this design (Dan Torres, "Inside Tube Amps" and others) is to use two 80uf caps, 450 each, in series. Each caps has a 220K resistor across it (to act as a voltage divider). This yields 40uf but with a voltage rating of 900 volts! This is well above the typical 420 volts that come off the Rectifier. So the idea is achieve higher voltage handling capabilities. But would this setup really help reduce ripple (to either significant or audibly detectable levels)?

And what about doing the same thing for the caps at points B,C, and D? Would it make sense to do the same thing?

I know caps have vastly improved over time. But I was wondering if anyone has used the above method in any of your rebuilds or if you think the original setup is sufficient.

Thanks, Tom

[eschertron]

...use two 80uf caps, 450 each, in series. Each caps has a 220K resistor across it (to act as a voltage divider). This yields 40uf but with a voltage rating of 900 volts! This is well above the typical 420 volts that come off the Rectifier. So the idea is achieve higher voltage handling capabilities. But would this setup really help reduce ripple (to either significant or audibly detectable levels)?

Thanks, Tom

Tom,
The series 'totem pole' arrangement of caps is purely for the life of the caps, protection from over voltage. There is no sonic advantage. Caps today are not rated for higher voltages, or they get $$$ real fast as voltage rating goes up. Compare 2ea 350v caps (double the capacitance, as you observed) to 1ea 600v cap. Choose the method that makes the most economic sense. Do you need a cap to handle voltage that might peak beyond 600vdc? I don't think you'll have an option then.

But, yeah, two 450v caps for 900v? My guess is that's what they stocked, so that's what they designed around.

[R.G.]

Eshertron is right. The splitting then stacking of filter caps is purely to get higher voltage ability, less voltage stress on the caps.

Well, maybe not purely. Power filter caps have a thermal life too. The continual surging in and out of current makes heat in the small, but present resistances inside the cap. That generates heat, and the heat has to get out. This is done by the caps having a temperature rise, and the outside getting hotter, radiating away the heat. It is also how caps can fail - if the heat boils the electrolyte, the cap vents the vapor and dies. It is >>possible<< that two smaller section caps in series would have more surface area to radiate heat than one big one. But this is a tiny, minor effect and a designer would have to be really reaching to use this as an excuse.

Mostly it's done for voltage ability.

I actually thought you were going a different place with this question. For ripple reduction, the equations are the equations. If you have a load current I, a fixed power frequency (i.e. 50 or 60 Hz), meaning a fixed time between charging pulses (dt) and a fixed capacitance. The equation is:

I = C dv/dt

so you can pick any dv (the ripple voltage) by knowing I (the load current) and the capacitance.

Rearranging, dv = I*dt/C

So you can get ripple (dv) as low as you want by knowing the load current I and putting as big a capacitor C as you want.

So you can make your B+ as ripple free as you like by just putting in more capacitance.

But not if you use tube rectifiers. The bigger the capacitors, the bigger the pulse currents charging them. And tube rectifiers will be degraded and die quickly if the pulse currents are too big. So every tube rectifier puts a limit of how big a first capacitor you can use for ripple smoothing.

I had a dodge to get you past the limitation on first capacitors for tube rectifiers - but that wasn't your question at all.

[TomCarlos]

Eschertron- thank you.

R.G. - yes I was also asking about ripple- I think you covered that too.

One other thing. I am looking at the specs of the GZ34. It shows the Capacitor Input at 60uf. Does that refer to the FIRST cap that the rectifier sees (before the OT, before the Choke) or does that refer to all the caps in the Power Supply chain?

[Malcolm Irving]

One other thing. I am looking at the specs of the GZ34. It shows the Capacitor Input at 60uf. Does that refer to the FIRST cap that the rectifier sees (before the OT, before the Choke) or does that refer to all the caps in the Power Supply chain?

Just the first cap.

[R.G.]

The trick is to put a first capacitor there that is less than the maximum first capacitor value, then isolate that from a second filter cap with a resistor or inductor that keeps the second cap from being "seen" by the rectifiers. The amount of resistor is small. You lose a little of your voltage, but the value of the second cap can now be much, much larger than that first filter cap.

Whether this is worth while or not is up to the person using the amp. A big cap for low ripple may be perceived as too "strict" - or not. Depends.

Another way to mess with it is to just regulate the preamp stages with simple follower regulators. This cuts the ripple in the early stages hugely.

The other way to get around the "first cap maximum" is to change to solid state rectifiers, perhaps with tube-rectifier-faking parts attached, to sidestep the maximum current limitations. Solid state diodes commonly have surge current limits in the tens of amperes, sometimes hundreds of amps. With SS diodes, you can largely use any amount of capacitance for the first cap that you can afford.

[stokes]

As to the recommended input capacitance,I have done extensive testing with various rectifiers and you can go way over the recommended values.One example is I have a 5E3 clone with a 5v4 rectifier,the book says 10uf.I put an 80uf as the first cap.My son has been gigging with this amp for almost 10 yrs.I have another amp my other son has been using with a 5R4,book says 4uf,I have 40uf in that one,again,using it for years with no problems.

[R.G.]

Interesting.

I wonder if:
> the old admonitions no longer apply; perhaps rectifier tubes are just much, much better today than in the Golden Age
> the old advice written into tube handbooks of the day were just wrong, or just way too conservative; that would have let them sell fewer rectifier tubes
> modern caps don't "suck" as hard for some reason; maybe those two amps's filter caps are higher ESR or have drifted to higher ESR
> there is something about those amps that let them sidestep the Golden Age advice; maybe higher PT resistance or something
> those two amps have super good rectifier tubes somehow

It would be really interesting to find out whether the technical advice is/was wrong or if those two amps are exceptional. And whether we can generalize from those two amps to all amps with tube rectifiers.

[uneumann]

One example is I have a 5E3 clone with a 5v4 rectifier... I have another amp my other son has been using with a 5R4...

The 5E3 doesn't usually have a standby sw. Does the 5R4 amp have a standby sw?