"*kg Date:3/15/2002 1:48PM Something to try...
Bypass your cathode to the B+ node of the stage instead of ground. you will need to heed the voltage requirements of the cap, obviously.
this increases dynamics quite a bit. as the tube turns on, the B+ node gets pulled down and the cathode voltage is decreased, which increases plate current a bit more. it's essentially positive feedback from the power supply.  Can be used on preamp as well as power amp stages. ken

*Dan Z 3/15/2002 7:14PM ken - A couple of questions.
Do you leave the cathode resistor connected to ground, and have only the cap go to B+?  Could you make this connection "above" the plate resistor? Seems like a convenient place with a short wire run. I'm picturing a SF Fender. Thanks, Dan Z
*kg 3/15/2002 8:23PM - yes, rk continues to be connected to ground.
yes, the connection of the bypass cap goes "above" the plate resistor.
in a perfect power supply, the voltage of that b+ node "above" the plate resistor wouldn't change, but we don't have perfect power supplies in our guitar amps.
watch the polarity on the cap too--it would be reversed from what normally connects to the cathode.
in fact, you can actually use such a connection to the cathode to decrease power supply noise (ripple). it was described in tubecad journal. ken
*PaulC 3/16/2002 1:00AM - I'm not seeing something here (like that's any thing new...). Ck shunts the AC component to ground keeping the bias voltage from being modulated by the input signal. The bias voltage is coming from the current through Rk. If Ck is connected to AC ground at the B+ node how does this differ from earth? The signal is bypassed in each case - I'm not seeing how this could effect anything. Is it something like boot strapping? A little bit of the ripple on the B+ could modulate the shunting of Ck? This one's messing with me!! 
*kg 3/16/2002 2:58PM - paul, the key is that the b+ node is not a perfect ac ground (certainly not the ac ground like earth is), therefore it WILL have some signal on it. ken 
*Jay Doyle 3/15/2002 7:50PM - Ken, This is interesting, I'm more of an FX guy so do you think that this would work similarly in a transistor stage?  Could you explain how this is different from grounding the cap? I don't see how the B+ supply would sag.
Thanks for the tip! I've got to try and experement with this. Jay Doyle
*MBSetzer 3/16/2002 1:40AM - The B+ does not only sag, clip your scope on there and there will probably be a node where you can see a reasonable facimile of a wide-bandwidth sample of your guitar while you are jamming.
This will be millivolts or more of responsive ripple riding on the B+, if B+ is not smooth enough it might be harder to see with the low Hz surf rolling in. Of course the better filtering the less there will be. Plus you might need a high voltage scope (tube input) to go without a 10X probe.
Whether you can measure this very well depends a lot on the particular combination of gear. Mike
*J Epstein Date:3/18/2002 5:15PM - 
Bob Danielak and I have discussed this a fair amount and I agree with him as far as what I have heard. I feel safe in being able to present his ideas on this subject (mine will be mixed in too, of course!)
Here's our best guess:
A typical two-stage amp has an additional pi-filter stage between the power tube B+ connection and the driver tube B+ connection. This introduces a frequency-dependent lead/lag between the two power supply connections. (Now, it also does good things like reduce ripple to the driver, so removing this pi filter typically requires adding more filtering before the single B+ node in the "modified" circuit.)
In the "modified" circuit, you draw B+ from the same spot for both amplifier stages and any transient demands on the power supply will be partially cancelled between the two out-of phase stages FOR ANY FREQUENCY. No, it's not perfect cancellation, but it is frequency-independent and it is cancellation in perfect phase. Maybe it's the best we're gonna do here. 
Neither Bob nor I have done serious comparison with completely separate supplies for the driver stage, this might be the next frontier.
Note the amps we are usually discussing are very low parts-count single-ended hi fi amps and in such amps the power supply provides a LOT of the mojo.
As far as two stereo channels being "vaguely correlated" I couldn't disagree more - let's face it, all the best music was recorded in mono anyway so those records are ~100% correlated (as you can tell, I place a great deal of importance in how a stereo setup sounds with a mono source!) Also with a typical stereo mix, where is the lead instrument, lead vocal, bass guitar with its dynamic power demands on the power supply?
"Smack dab in the middle, let me rock'n'roll to satisfy my soul."
So I think you are going to be better off with a dedicated supply BY STAGE rather than a dedicated supply BY CHANNEL, and a single supply might outperform either one, that's not been tried yet (by me or Bob D.)
With three-stage amps and greater, this analysis does not apply, as you will most likely get motorboating if you try using one node for all three stages. -j 
*PAULH 3/18/2002 9:20PM - Thanks for the interesting tip Ken. Would this work the same on transformer coupled circut? Aside from voltage requirement, how big of a cap we talking about. The same value as the bypass cap?
*kg 3/18/2002 10:19PM - yes, it will work on a xformer coupled circuit as well.
from what i've picked up on the net, it will also work in a PP output stage--instead of bypassing shared cathode resistor to ground, do it to the center tap instead.  The capacitance value should remain what it was before. the voltage rating will go up, and the polarity will be reversed. ken


 

©