For some reason the power supply filter architecture has really perplexed me. I had always assumed that totem pole filter caps were simply used to solve the unavailability/expense of caps with high enough working voltages (which we can all agree is not the problem it would have been in the '60s).
The fact that HAD switched gears the opposite direction in later amps seemed backwards, although from the posts I take him for a scrounger, so I wasn't too alarmed. Reading posts I found some builders that found that using totem poles had some effect on the amp performance other than cap longevity. I try to partially disengage my classically trained engineering brain on tube amps because I know that things that EEs might normally write off as trivial turn out to be absolutely critical.
Well, alas, I was unsuccessful on this one and put pencil to paper and I think I can put this one to rest, hopefully permanently. I've attached a picture showing two networks that we might choose for PS filter stages.
The network on the left uses a totem pole config and the one on the right does not. I set the variables representing the capacitance and bleeder resistance so that they would be equivalent between networks. I then solved the complex impedance of both (s is the Laplace transform variable used to analyze complex impedances and for simplicity sake is equal to 2*Pi*freq). You don't need to be an engineer to see that solving the two equations yields the same answer for both networks (i.e. they are equivalent).
What does all this mumbo jumbo mean? You can replace a totem pole cap section with a single cap and parallel resistor. If the resistor does indeed effect the transient response, add it. It won't hurt anything and will have the added benefit of draining the caps when the power is shut off. So, if 220k is standard for current totem pole designs, use 440k on single cap stages and you should see the benefit, if there is one to be seen.
If we think about why totem pole stages ever had the equalizing bleeder resistors in the first place we need to look at an electrolytic cap at DC. It should ideally be infinitely resistive, but it is of course not. This parasitic leakage resistance varies wildly, so without a much smaller parallel resistance to dominate, the DC voltage sharing of the caps would not be equal and we could over stress one of the caps and cause a catastrophic failure. Avoiding the discussion over whether or not the UL would agree that this is a good practice in 2007, it seems that we can put this one to rest....I hope
So in summary...Use totem poles if you like or use single cap stages. If you add 220k on totem pole and 440k bleeders on single cap stages they should give you the same result.
Thoughts, comments, pleads for brevity
Thanks,
Dave
Dallas, TX
