PP power supply design questions

[SY]

It still sags by the same amount

torrence got it- slow recovery means slow sag, so that for a given overload event, the supply will sag less.

 

[jon_010101]

torrence got it- slow recovery means slow sag, so that for a given overload event, the supply will sag less.

For a CLC supply with solid state diodes, the steady-state sagged B+ voltage is the same with an X uF vs. 10XuF. Unless the event is less than a few hundred ms or so, or unless the supply has thousands of uF, the minimum voltage is still likely to be achieved. It just depends how long you want to wait for it ramping down and up when transitioning from small-signal to clipping and back, and how much you trust the 1000uF caps. I'm not suggesting an undersized supply, but one tuned for sag ~1-2% of B+, that recovers fully between typical transients. Approaches may vary, depending on bias points and expected applications, desired chokes, or use of tube diodes

 

[SY]

The devil is in "steady state." Most overload events are much more transient than that.

 

[torrence]

Note also that the characteristic response time is also not independent from the Thevenin output impedance of the supply. To get the stiffest possible supply you want the lowest possible output impedance. In AB operation (much more so than pure class A) the dynamic load seen by the supply can change by large amounts. Since there is always some quiescent load at zero AC input, the quiescent B+ will always be loaded down from the no-load (Thevenin equivelent) voltage. Minimizing that loading is equivalent to minimizing output impedance, and will always reduce transient distortion (unless you have some fortuitous cancellation with tube non-linearity). Typical loading is much greater than a few percent. A stereo PP load is around 5k, which would need a supply output impedance of order 50 ohms. That would be a beefy supply approaching op-amp performance.

 

[jon_010101]

Supply impedance at 1% of load is rarely achieved in the classic amps, but it's a great DIY target for serious AB(use). My suggestion of aiming for a just-over-damped-of-critical transient response obviously demands very low supply impedance, otherwise the supply and amp will communicate too much. A Citation II with big caps is in this territory, with only a -5V drop from 300-400mA. My monoblock project amps have ~twice the supply impedance, but they are admittedly Citation-inspired with ~50% downsized iron.

 

[SY]

There's a reason I actively regulate.

 

[ray_moth]

SY, I also think active regulation is a good idea. If we're talking true pentode PP Class AB, I would regulate the screens (and the voltage amp stages, through being decoupled off the screen supply), and also the fixed bias supply. but not the OP tubes plate supply. I assume you'd agree?

 

[SY]

Well, since that's exactly how my published pentode amp is set up, that's a safe assumption.

 

[ray_moth]

Trouble is, thinking more about it, I suspect that would tend to rule out 'partial feedback' and any other approach wherein the driver B+ is derived from the OP tube plates, because of sag. The alternatives, I suppose, would be either to stick to Class A or to regulate the whole lot. And what of triode and UL operations - should the whole supply be regulated for them too?

 

[SY]

For triode and UL, the main B+ should ideally be regulated. True, it removes some of the tube colorations that some people like, but the output better replicates the input (for better or for worse).

 

[SpreadSpectrum]

Inspired by other threads, I built a Maida regulator for my power tube supply for 200-500mA. It was surprisingly easy. I haven't really hooked up the test equipment yet (need to build a dummy load first), but it definitely sounds better than the earlier version of the supply with a capacitor input filter and Hammond 193Q (10H 500mA).

It also got rid of a lot of other annoying features that the choke brings (voltage overshoots, LF oscillations as I was biasing up the tubes).

I'll never buy another choke. Regulators are much cheaper and work much better.

 

[SY]

Delighted it worked for you. Chokes were fine in 1949, but we can do better these days.