[Nixie]

I asked someone for how to make a regulator for a power amp supply (Aleph-X, balanced single ended), and he sent me a generic schematic, as well as a description which I reproduce below. Besides that the MOSFET is on backwards, I can't figure out what values to use to get this to work in the simulator. I'm unable to contact that person anymore, so I'm looking for help here.

Quote:

Originally Posted by hifizen My first recommendation is to stick with a relative reference voltage, rather than absolute. Add some low Rds(on) MOSFETs as the pass elements instead of BJTs, and you'll be able to minimize the forward losses without too much concern that a sagging line voltage will cause dropout. Build a feedback loop with a bipolar diff pair, some Rs and Cs, and presto, you have a nice little regulator. See attached schematic. This version will give you an output which remains 0.65V-0.7V below the average input value at all times. This assumes that input ripple is already fairly low - you'll need to simulate this to make sure there isn't any dropout under worst-case conditions.

 

[peranders]

This regulator has at least one drawback: Will not regulate particulary well due to lack of open-loop gain. It will also be sensitive for incoming voltage variations. PSRR will be so good.

If you want a fast regulator with extremely good PSRR and extremely low noise you may check out the super regulator concept.

 

[PeterR]

What do you want to use this for? A 0.50$ LM317 won't be that easy to beat for most applications. If you need more current, the LT1083 family goes up to 7.5A. But if you want to go overboard, google for "Sulzer regulator" and "Jung super regulator".

Edit: D'oh, read twice, post once. Aleph X...

 

[Nixie]

I have 12 capacitors that are 36000 uF, so 6 per channel (monoblocks), 3 per rail. I figure CRC-regulator-C. Since the Aleph-X draws constant current, and there's 36000 uF after the regulator (per rail per channel), I do want just a simple one. The main two points are dropout voltage, because of the power levels and cooling needed, and relative reference, as variation of mains voltage can overtake the dropout amount.

Any suggestions for circuit values to get this to work in the simulator? I'm getting at least 2 V drop no matter what. For some reason, even if I just have the MOSFET all by itself, with gate connected to the input, it's still giving me this large voltage drop, no matter if I use a MOSFET model with a very low Ron. What?

 

[PeterR]

Dumb question, as you said "the MOSFET is on backwards", are you using a P channel Mosfet?

 

[peranders]

Nixie, you must ask yourself WHY you what to have a regulator. Then when you know the answer you must draw up the specifications so you will get a design goal. Without a goal you can hardly verify the result.

 

[Nixie]

Peranders, what do you recommend? Superregulator is overkill. I do care about heat since it's for a power amp, and thus low dropout -- and then it does need to be relative, slowly responding to long-term dips so as to prevent the possibility of input dropping below the overvoltage. I understand the capacitor as an integrator, so it lets through only very low frequency variations. In that sense it's a bit like Rod Elliott's capacitance multiplier.

Peter, thanks for the comment, I was using N-channel, but using a P-channel allows very low dropout.

I've got it to work in the simulator, but I need to ask -- what does C2 do? In the simulator, it seems to slow down the response to large variations of the regulator, but I don't quite see why it's there.

And, what's the best way to do the negative side to get best matching (as the amplifier is balanced)? Use reverse devices, or have the pass device on the ground line as in the Blue Hawaii power supply (schematic on Headwize)???

 

[PeterR]

Quote:

Originally Posted by Nixie No, N-channel, as in the diagram.

Should be a P-channel. Try R1,R5=2k; R2,R6,R7=10k; R3=4k; R4=100
Just seat of the pants values, but simulates ok.

Edit: I see you've figured it out in the meantime, got distracted by work while replying

 

[Nixie]

What if I replace the transistors with an opamp? That way I have a bigger range of gain to play with.

 

[tangent]

Quote:

Originally Posted by peranders If you want a fast regulator with extremely good PSRR and extremely low noise you may check out the super regulator concept.

I'm not sure it's clear to all of you that Nixie's project is a power amp.

The Jung 2000 design is only good for about 300mA as designed, and it isn't possible to take it too much higher, due to the limitations of the error amp's ability to sink current. The various related designs probably have similar limits, only it's a matter of sourcing current in those cases.

You can go and add a bipolar in there to help the op-amp to yank harder, but really, super regulators just weren't conceived with power amps in mind.

 

[Nixie]

I like the idea of a capacitance multiplier for smoothing out ripple with little dropout from Rod Elliot. I came across Mr Evil's (from diyaudio) upgraded version here:
http://www.cheese83.freeserve.co.uk/amp/amp4/psu2.htm
I figure it wouldn't be too hard to up the power.

 

[PeterR]

Quote:

Originally Posted by Nixie I like the idea of a capacitance multiplier for smoothing out ripple with little dropout from Rod Elliot. I came across Mr Evil's (from diyaudio) upgraded version here: http://www.cheese83.freeserve.co.uk/amp/amp4/psu2.htm I figure it wouldn't be too hard to up the power.

Basically that's a somewhat more elaborate version of what's in your schematic.

 

[peranders]

Quote:

Originally Posted by tangent I'm not sure it's clear to all of you that Nixie's project is a power amp. The Jung 2000 design is only good for about 300mA as designed, and it isn't possible to take it too much higher, due to the limitations of the error amp's ability to sink current.

I'm afraid you are wrong becasue later Walter Jung added an emitter follower which ends up in uA from the opamp.

I see no reason to scale up the design. Audiocom for instance uses a mosfet as pass transistor.
http://www.diyaudio.com/forums/showt...863#post639863

A super regulator is a basic topology and can be tuned in for every need within it's limits.

Max voltage out is 36 volts but even this can be streched.

 

[Garbz]

But would the Aleph need a regulator at all? Wouldn't a CRC filter with overkill capacitance do the trick?

 

[rickcr42]

take what is out there and make it work for your application !

some information on class A power amp regualtion :

http://www.tcaas.btinternet.co.uk/index-1.htm

Rather than just try to find a circuit that may not be exactly what you are looking for it is better to actually read an article on the "why" and not just the how then once you have a small understanding of the method it is a simple task to modift the circuit.The power supply section of this page has not only circuits but goes into the details of what worked and where each has limitations.If using a mosfet is your goal as the pass element it is easily substituted for the bipolar transistors.