usability of an old transformer

[RocketScientist]

Originally posted by star882


A well designed regulator is neither expensive nor inefficient. A simple buck converter will work fine. It will allow the beginner to learn about power supply design without having to deal with a direct connection to the mains.

You might want to read what some of the audio amp "gurus" like Douglas Self, Randy Slone, etc. have to say about regulated power supplies and audio amps that put out any real power (i.e. over 15 watts or so). Switching regulators, like a buck regulator, present their own problems--not the least of which is lots of high frequency "hash" from their high current switching pulses. And if the audio amp tries to draw more peak power than the buck regulator allows, most buck regulator chips will shut down or otherwise seriously glitch the supply voltage causing nasty noises in your speakers. So the regulator has to be sized for seriously high currents to prevent it from going into protection. As does the switching inductor to prevent it from saturating. It starts to get expensive even if you can solve the noise problems.

In Class-D amps (not what we're talking about here) switching power supplies are more popular because the Class-D amp already needs filtering to get rid of the high frequency garbage from the power supply. But in a Class-B amp, like a gainclone, the noise on the power supply from a buck regulator would likely be a problem.

It's also tricky to design high current, high voltage buck regulators. Most regulator chips don't meet the requirements. And the extremely wide range of load currents presents challenges with respect to sizing the inductor, minimum duty cycle, etc.

[star882]

Originally posted by RocketScientist
You might want to read what some of the audio amp "gurus" like Douglas Self, Randy Slone, etc. have to say about regulated power supplies and audio amps that put out any real power (i.e. over 15 watts or so). Switching regulators, like a buck regulator, present their own problems--not the least of which is lots of high frequency "hash" from their high current switching pulses. And if the audio amp tries to draw more peak power than the buck regulator allows, most buck regulator chips will shut down or otherwise seriously glitch the supply voltage causing nasty noises in your speakers. So the regulator has to be sized for seriously high currents to prevent it from going into protection. As does the switching inductor to prevent it from saturating. It starts to get expensive even if you can solve the noise problems.

In Class-D amps (not what we're talking about here) switching power supplies are more popular because the Class-D amp already needs filtering to get rid of the high frequency garbage from the power supply. But in a Class-B amp, like a gainclone, the noise on the power supply from a buck regulator would likely be a problem.

It's also tricky to design high current, high voltage buck regulators. Most regulator chips don't meet the requirements. And the extremely wide range of load currents presents challenges with respect to sizing the inductor, minimum duty cycle, etc.

Put the bulk capacitors on the output of the regulator and only use small bypass capacitors on the input. Essentially an active PFC but with a buck instead of a boost converter. The operating frequency is generally well above 20kHz, nicely out of band. There should not be any in band noise unless the feedback loop is unstable. (If you really want engineering challenges, design a power supply for DSL equipment. Most switchers have the operating frequency and their harmonics right inside the bandwidth you're trying to work with.)

[RocketScientist]

Originally posted by star882

Put the bulk capacitors on the output of the regulator and only use small bypass capacitors on the input. Essentially an active PFC but with a buck instead of a boost converter. The operating frequency is generally well above 20kHz, nicely out of band. There should not be any in band noise unless the feedback loop is unstable. (If you really want engineering challenges, design a power supply for DSL equipment. Most switchers have the operating frequency and their harmonics right inside the bandwidth you're trying to work with.)

That only addresses one of many problems. You also need two regulators (one each for the positive and negative rails), and unless they're running from a common clock, you're likely to get beat (difference) frequencies that will be audible. Even running from a common clock, the current demands for each rail will be constantly different, hence the duty cycles will vary with respect to each other and likely create audible artifacts. High current fast rise time PWM waveforms create all sorts of nasty broadband energy that likes to find its way into the audible band--especially when you have two such generators running in close proximity to high gain and relatively high impedance audio circuitry.

And many have demonstrated that even out of band HF energy can degrade the sound in various ways. And that still leaves cost, protection circuitry interfering, inductor saturation, lack of chips that can handle the voltages, etc.

Switching regulators are the way to go for digital power supplies of 5 volts or less. But they really are *NOT* the way to go for a class-B audio power amp. There's fairly clear consensus on that. Do what you like with your own designs, but I'm trying to help zvir avoid a potential nightmare.

[star882]

The easiest way is to center tap full wave rectify the transformer to get a single 60v output, then run a BTL from it.