The best power supply

As far as most things in the world of audio go, there is no real "best". I think power supplies though are an exception where there really is such thing as a perfect power supply, or at the very least one that can be defined as an ideal. To me this means a constant DC voltage with no noise or ripple no matter what load.

So what is the closest to this ideal? Is it a bank of batteries? A lab grade power supply?

I ask this because I want a reference power supply that I can use to test my amp design and be confident that the power is the least significant contribution to what is influencing anything I might be hearing.

 

The answer to that requires a question - what are you trying to power? Where is your powered circuit most sensitive to noise? A couple of years back I dove full into battery power to feed the critical pins of a 9018 DAC I was building. The lessoned learned, was that battery power was not the best sounding. Understand - I took great care, only a certain type of battery, passive charging circuit, etc. It sounded good. It did not sound the best.

On 'lesson' in searching for the best PS for an application is, after you pay attention to the essentials (high PSRR, mains isolation, grounding) - different 'good' power supplies sound different. Why is a bit beyond me (some of it I understand, but certainly not all of it) - but they do. For an analog circuit, I can wire up a Salas Shunt, an AMB alpha 22, and a well designed chip regulator bipolar PS; they'll all sound different. Take just one of them, say the Salas - swap out the smoothing cap for say, an organic polymer - sounds different. Change the diodes - to Cree, or some other very fast switching, low loss - changes the sound.

The last 5 years of DIY have taught me, If you want to improve your sonics, be very sure you've paid attention to your PS; is it the best circuit for what you're powering? Can you improve your components?

Sorry - soapboxing To your question. I don't think it's possible to select/build a PS that completely takes the question off the table - "can it be improved for the circuit / component I'm powering", but you really can start with a solid, high performing PS. For that, I'd recommend the AMB Alpha 11 or 22 of you need bipolar. The shunts are out, as for best performance they really need to be tuned for current draw, etc. More specifically, I mean a 'Salas' circuit. There are other shunts available, but in my experience do not achieve the same potential performance. For analog circuits, the 11/22 are unique in that they are not current limited. They have good PSRR, are fast, and can perform very well when built with the right components.

A lab-grade supply would meet your needs, but as a rule, they're not designed / built for optimum sonics.

My 2 cents.

 

Get a used bench power supply with the features you need for prototyping. Leave the audiophile masturbation for later in the process.

 

@MisterRogers - Well for my purposes its for a high voltage B+. I was curious if batteries would give the best sound result regardless of how impractical it is. Bottlehead uses batteries for their low voltage tubes in their Quickie preamp so it has some validity. For tubes I have also seen a Maida high voltage regulator and a tube shunt high voltage regulator.

 

The choice of power supply depends on the circuit it has to feed.
What's optimal for circuit A may be far from optimal for circuit B.

 

I can see that for unregulated power. However I fail to see how a regulated power source wouldn't work for two different circuits that need identical voltage requirements.

 

It would all depend on how immune the circuit it must feed is to power supply fluctuations (caused by different reasons).
I see no need to go overboard with elaborate circuits if the audio/digital circuit itself isn't affected.

But sure, you can use close-to-ideal power supplies for every audio circuit and just 'know' that part is O.K.

 

Well first the close-to-ideal power supply is for the initial prototyping phase so that I can experiment with the circuit. After I get something I like then I'll work on the power supply for something a bit more sensible for the circuit I am working with but still comparing with my initial power supply for reference.

The other thing is that since this is DIY and not a production model, I wouldn't mind over-engineering the power supply to satisfy my own curiosity/obsession.

 

If you want a single universal option then what you can do is find a relatively high secondary xfrmr rated for plenty of current. Based on whatever you're trying to test/prototype, get a handful of higher power resistors and caps to use as an CRCRC filter. Every R drops voltage based on the current drawn by the load, so you have to do a bit of math but you plug things in. But you'll end up with a fairly ripple free supply. Keep in mind this is still unregulated. A regulated high-voltage supply is a whole project in itself.

Still easier to find a bench supply that can do higher B+ values you need for prototyping, although sonic performance might not be the bestest.

 

I have a CLCLC filter in my 300B amp and it is very very quiet. I use a film cap for the second C (47uF 800V). I would also like to use film for the first cap, but at the time, I didn't have the right cap.

 

I tend to favor lab supplies for experimentation, such as linears designed by HP engineers long ago. Those EEs had a good grasp of universal supply design. HP became Agilent and now the Power Supply division is Keysight. Not cheap, but demonstrably better (can measure and hear differences.)

 

I have just a limited experience with this in a few designs. If you know the noise characteristics of the power supply, you can design filtering on your board to clean it up. That being said, I've found Batteries to be very clean for most noise sensitive applications.

If you're looking for someone to guarantee a low level of noise, they'll make you pay for that guarantee. Batteries are cheap!

Alternatively you can either design a regulator on your board or buy one from a third party.