Power supplies of all vintage tube amplifiers from 50-60x look so miserable with such small capacitors. It is especially true for stereo amplifiers when PS is used for both channels in parallel without any separation. I can say for sure such a PS will not work for a SET amplifier without feedback even if it is just 2 watt power.
How does it work in all these vintage amplifiers? Does feedback make them less sensitive to PS or the reason is the push-pull topology?
@alexberger Feedback improves power supply rejection and crosstalk; ideally anything that isn't the signal. For example, in an LP mastering situation, the feedback winding on the cutter head is essential to allow channel separation. The stereo LP would not exist without it.
Such a power supply as you described above will work fine in any SET that is running class A1 on the power tube, since at any signal level current draw is constant; IOW from idle right to full power. Consequently this statement:
The issue with small capacitors is not the 120Hz noise from AC , but voltage stability during playing of load and complex music.
-is false.
Boosted capacitance does help with a stereo amp or even a mono amp if the amp is push pull and operating class A2 or any form of class AB (IOW where current draw varies with power), so as to reduce intermodulation.
In all cases, the proof of the pudding is to put an oscilloscope probe on the power supply rail and run the amp up to full power at a variety of frequencies to see what sort of noise is present in the power supply. Since the output circuit isn't linear, power supply noise can intermodulate with the audio signal in the output section.
@hilde45 The article you quoted was true back in the 1980s but is not true now. The real issue isn't that we can't measure it, its knowing what the measurements are telling us. That knowledge is in short supply! I've been explaining exactly why SETs sound the way they do. Again, in a nutshell: They shine when not presented with bass or asked to make much power. I've already explained why they sound good when they do so.
Harley's mistake was simply not asking amplifier designers why SETs sound the way they do. But if he had done so he might not have had an article to write 
I was online pointing this out when this article originally published a few years back.
I posit at least part of this quandary is a function of low parts count in SET. SET doesn't have to negotiate multiples of caps, resistors, transistors, what have you, and if point to point wired no circuit board. I can hear sound quality differences with a changing out a single resistor, cap or hook up wire, this goes to show how transparent these amps are. I used to have to change out multiples of these items in other amps to exact a change in sound quality. This suggests a signal passing through a less circuitous route is less changed/affected/contaminated, in other words more pure to the source.
@sns This statement comes to a false conclusion. The reason you hear differences due to components is not because SETs are particularly transparent (due to their having the highest distortion of any kind of amp made, and because distortion tends to obscure detail, its arguable they are the least detailed amps made), its because they lack feedback.
Feedback helps the amp reject that which is not the signal. Artifacts from wire, resistors, capacitors and transformers are not the signal, as well as noise from poor grounding, layout problems, line Voltage and so on. So when there is no feedback, everything (and I mean everything) makes a difference.
The problem of course is whether or not the feedback is applied properly and we've all heard amps where it wasn't. I've no doubt that many might prefer an SET over such amps! But its usually not due to detail- its (IME, IMO) due to amps with poorly applied feedback not sounding particularly musical on account of harshness or the like.
