Power supply design

[jelle]

Hi everyone,
There have been some questions concerning power supplies. I spent some time on this subject and learned a few things which I’d like to share with you. Please let me know if I’m wrong. I know its a long piece to read but to me it’s a complicated subject, so here goes:

Concerning power supply design, this is the way I understand it:
The 220k-470k resistors across the two caps have several functions:
1) protect the caps against transient peaks in voltage, thus stabilising the PS,
2) 2) divide the voltage equally across the two caps,
3) 3) lower the voltage ( this might be good or bad…)
I prefer to lower the voltage here and not with the 470 ohm – 22K resistors between the caps. These resistors, besides the voltage, also regulate the frequency emphasis of the stage. This was on the ampage a few weeks ago and somebody mentioned the 27k and 56k (IRRC) values in the normal and bass channels in the 6G6-A bassman and that Leo really knew what he was doing. This was kinda new to me and I didn’t understand the technical explanation. Maybe someone here does?
Example:Yesterday I changed the screen to pi resistor from 8.2k to 20k in a jtm 100 (560V screens) and this added some darkness to the amp. Great result! (I know this also dropped the voltage a bit)

So a high value will give lower freq emphasis.

Now the voltage:
Headroom (voltage) is easy explained with the old datasheets for the 12ax7 in which can be seen that the amount of distortion decreases when the voltage increases. Higher voltage will give higher gain and lower distortion.
I spend some time tweaking my superlead an my 97 clone in the PS area. Really made a big difference.

I like to use 2x 220 or 2x 300uF with 270k’s across for the plates. This way the plates are stable and will give great headroom. This will not make the amp stiff. The stiffness comes from the caps for the screens. Just compare a ’70 superlead (100uF screens) to a ’67 superlead (16uF). One of them is stiff and and the other one isn’t. I like to destabilise the screens a bit by keeping the caps small here. I use 47 II 47uF with 470k’s across them for el34 and 6L6. This keeps the amp from being stiff but the plates have tremendous headroom.
I learned this from Grimwood’s site. It’s basically for hi-fi but good info can be found here and he has a nice view on building amps:

http://www.webace.com.au/~electron/tubes/ look for screen grids.

The resistor instead of the choke does create a lower freq emphasis. I noticed a lack of highs with a high value. Perhaps this is EVH’s brown sound…….
It also destabilises the screens and de-stiffs the amp. I agree that the screen caps values have to be increased with the resistor in place to prevent the amp to become muddy.


For the more complex amps that use reverb/dumbleator I’d use single caps for the preamp supplies (no 470k resistors) to keep the B+ high enough.
HAD does this also ( 150W dumble schem in Aspen Pittman’s Book) and the voltage might be the reason.

These are my experiences. I hope this is of any help,

Jelle
The netherlands

[ayan]

Hi everyone,
There have been some questions concerning power supplies. I spent some time on this subject and learned a few things which I’d like to share with you. Please let me know if I’m wrong. I know its a long piece to read but to me it’s a complicated subject, so here goes:

Concerning power supply design, this is the way I understand it:
The 220k-470k resistors across the two caps have several functions:
1) protect the caps against transient peaks in voltage, thus stabilising the PS,
2) 2) divide the voltage equally across the two caps,
3) 3) lower the voltage ( this might be good or bad…)
I prefer to lower the voltage here and not with the 470 ohm – 22K resistors between the caps. These resistors, besides the voltage, also regulate the frequency emphasis of the stage.

{snip}

Jelle
The netherlands

As you wrote later in your post, the value of the resistors on the totem-poled 220uF caps is 270K, not 470K. Not that it really matters per se, IMHO, but the time constant of the 470K with 220uF is almost twice as long as with the 270K resistors, which means your amp will drain slower after you turn it off... As far as I am concerned, the main purpose of those resistors is to ensure the volatge is divided evenly between the two caps, otherwise chances are they'd do they own thing and one of them could exceed its voltage rating; secondly, as already sort of mentioned, it helps drain the caps when the power is shut down.

I didn't realize anyone used these resistors so as to lower the voltage of the B+? For one, I wouldn't think this method would lower the voltages all that much, and for another, I think there are other methods to lower the B+ in a controlled fashion (zener diodes).

The dropping resistors between the nodes in the power supply are not there just to drop the voltage, they form 1st order low pass filters with each subsequent capacitor stage and thus control (or not, depending on the design) the intermodulation levels between the various stages in the amp. Some people have stated that Ken Fisher toyed with that such that he would get very musical feedback between the various stages through the power supply. One way to get around this, i.e., to minimize the interstage modulation, is to regulate the B+ voltages in the preamp. As it were, I understand Fuchs amps do just that and I will defer any comments about the musicality of that feature to folks who have pulled the DC regulators out of their Fuchs amps to find out what they did to the sound.

It's interesting to see how Fisher did this on the one hand, and on the other hand Kevin O'Connor made a point of noting that his company, London Power, was the only one to decouple of every single stage in their preamps [Reference: The Ultimate Tone]. This goes to show the ying and the yang in all of this -- although I think we all know whose amps people drop huge sums of money to get.

Cheers,

Gil

[ElectronAvalanche]

Hi Jelle,

I have to agree with Gil. I always thought that the resistors were merely there to divide the voltage more evenly when putting electrolytic caps in series.

I have used 330k dividing resistors with two 100uF in my newest amp. If there is indeed such a major influence on the bass sound character of the amp, I will have to see if I can tweak with that. Maybe I will try to parallel some 330k with the existing ones and report back on my findings.

I am really undecided if these resistors have such a vast influence.

Cheers,

Dominik

[ElectronAvalanche]

Hi there,

I lowered the divider resistor values on my amp (2*100uF) and could not really hear a difference. Be aware though, that I played with the Master very low so as not to disturb the neighbors.

I paralleled a 330k resistor each to the already existing 330k dividers getting some 150k total. The difference was not audiable for me.

There was a slight fluctuation in volume when taking the extra resistors off (I used alligator clips).

I tried the same on the other totem caps in my amp (2*47uF caps with 470k dividers across them) and I personally could not hear much of a difference.

The phenomenon probably sets in under heavy load, when the caps do not have the chance to fully recharge.

My 2 Euro cents worth.

Cheers,
Dominik