[Amps] switching power supply for tube amplifier

Manfred Mornhinweg mmornhin at gmx.net
Wed Sep 19 10:35:00 EDT 2007


Hi David,

> Say you have a simple, non-stabilised switcher; then, in a push pull or 
> bridge design, a small guard band is needed between each side 
> energising/de-energising transition to allow for storage effects, to prevent 
> cross-conduction.

Now I see what you mean. Yes, of course. It was a naming issue: I know 
this as "dead time"!

> This guard band is "filled in" with the output capacitance.

Yes, but in any case it's bad practice to use such a simple switcher 
without any inductance. There should be some inductance to limit the 
inrush pulse in each conduction cycle. Of course, this inductance can be 
very small.

 >  This value of capacitance would be a lot smaller than needed
> in a stabilised design.

Not necessarily. In big switchers, it's more economical to do most of 
the output filtering (and energy storage) in the coil, and not the 
capacitor! High power switchers often have a large filter inductor, 
followed by just a few polystyrene filter caps totalling 1uF or so.

I would not build such an unregulated switcher, though. It's very bad 
economy! In such a thing, the output 100 (or 120) Hz ripple is 
essentially the same as appears on the primary filter capacitor, so you 
need a huge primary cap to get decent ripple. And large electrolytics 
are expensive, often becoming by far the most expensive part of such a 
power supply!

If instead we use regulation by pulse width modulation or by phase 
shifting, then we can maintain the output constant, with essentially no 
100 or 120Hz ripple, while using a much smaller and cheaper primary 
filter cap running with 30 or even 40% ripple on it. The small parts 
needed to implement the control of the power devices are dirt cheap, and 
so the whole power supply with the smaller input cap ends up less 
expensive, with less output ripple (essentially none), and there are two 
added bonuses: Tightly regulated output voltage, and an improved input 
power factor caused by the shallower and more stretched current pulses 
taken from the power line! That's without adding any power factor 
control circuitry, which of course can be done too, but increases 
complexity and cost.

For the above reasons, the reduced cost and much better performance of a 
regulated switcher compared to an unregulated one, I truly wonder why 
the power supply commented in this thread is unregulated! I would like 
to know if the designers had any good reason for it, or if it was done 
simply due to lack of expertise.

It has been an extremely long time since I last came across any 
unregulated switching power supply in commercial production! That last 
time was back in a time where pulse width modulation was an esoteric 
technique and had to be implemented using lots of discrete components!

Manfred.



More information about the Amps mailing list