LM3886 discrete mosfet full 42V regulator

[Mach5]

If you have ever looked for a voltage regulator for the LM3886 rated for the full capacity of the device you know how hard that is to find

Here is my initial design for a mosfet based regulator specific to the LM3886. I am trying to keep this as simple and universal as possible (with good performance still) so anybody can use this.

Any feedback is welcome

note: I am using a +/-48v supply which is already regulated and the LM3886 has decent protection already so the usual safeties are removed for simplicity. your mileage may vary

[Mach5]

Update the component values. I decided to go all mosfet this round

Changing R2 will adjust the output voltage

This is an extremely rugged and quite linear device suitable for all experience levels

theoretical range is from 44v to 100v input
50w power dissipation max

[pyromaniac_]

But what about the negative half of the regulator? Can't figure out how it could look like.

And is one cap at 1000uF really enough? If so I'll use a regulated PSU from now on.

[pacificblue]

Originally Posted by pyromaniac_

But what about the negative half of the regulator? Can't figure out how it could look like.

It would look the same, you only need to rename the rails. 0V becomes V- and V+ becomes 0V.

I wonder what a ±42 V supply should be good for, when the LM3886 cannot cope with the resulting heat dissipation.

[Mach5]

Pyromaniac

one cap at 1000uf is plenty. Don't try to use any fancy caps since this is only for the error amp and will only ever see about 600mV.
C1(andC3) keep the circuit stable. If you really need more capacitance you could add another cap between the regulator-as-drawn and the LM3886

This is capable of running the 84V needed for single supply mode, although depending on your heat sink you may want to double up the mosfets.

For dual supply see the attached schematic

Pacificblue

My intention is to make a regulator which is overbuilt and easy to use(and inexpensive).
You can always adjust the output voltage to whatever your heatsink will handle by changing the values of R2 and R7 and measuring with a voltmeter

[ikoflexer]

Hi Mach5,
I too have worked on a regulator for my now defunct gainclone. But I have to ask, why you want a regulator in the first place. Is your line voltage varying a lot?

It's easier to get worse sound with a regulator, IMHO, than without. If you just want to do it as a learning exercise, it's a good idea.

Now, about this design that you showed here, these are the things that I noticed. Zero dB line regulation. Basically, whatever ripple comes from before the regulator, ends up after the regulator. The output voltage is quite large. A 500mA sine wave current drawn by your amp will induce an almost 500mV sine wave ripple because the regulator can't deal with this demand. Approximate only, in fact the numbers might be wrong. But you get the idea. A 500mV ripple on a 500mA current swing may be audible?
Just my 2c.

[tomchr]

Always use an isolation transformer for anything hooked up to the mains. The turns ratio on the transformer doesn't need to be 1:1 -- and in most cases it's not. But if you want 120 V AC (230 V outside US/Canada), please don't fall for the temptation of just rectifying the power line voltage directly. That's a safety hazard.

~Tom

[geraldfryjr]

also im considering the configuration in post 10 to drive some esl's which require a step up transformer on the output and handle a minimum of 2 ohm load.iron is not cheap these days,and would be alot safer than the direct drive amp powered by a microwave oven transformer that i am considering!

[Mach5]

I would not throw out the advice of the experienced so quickly, there could be some added benefits you miss. Let's start with the basics

120v ac * 1.6 = 192v dc!

This "mach5 regulator" has an absolute max of 100v. True, you could use a higher voltage mosfet but those always have much less current capacity, so you would need quite a few. Their also not so cheap. I estimate $400 in mosfets alone for your application.
You also have to factor the added costs of higher voltage capacitors than would have been required if using a transformer
A transformer really isn't that expensive. Especially if you find one of an old piece of equipment.

Also, regulators convert the excess voltage to heat, so the higher the voltage you start with....and you want 1200 watts! That could easily be 20,000 watts actually used(166 amps!). I've installed commercial ovens that were only 5,00 watts. I don't think the heatsink you would need has yet been made

ESLs need about 2000v or more. The lowest powered one I've ever found was a headset that ran around 300v. So know matter what you will need a tranny if you are going to play with ESLs.

Have you considered winding your own transformer? It's not that difficult

[Mach5]

Originally Posted by ikoflexer

Hi Mach5,
I too have worked on a regulator for my now defunct gainclone. But I have to ask, why you want a regulator in the first place. Is your line voltage varying a lot?

It's easier to get worse sound with a regulator, IMHO, than without. If you just want to do it as a learning exercise, it's a good idea.

Now, about this design that you showed here, these are the things that I noticed. Zero dB line regulation. Basically, whatever ripple comes from before the regulator, ends up after the regulator. The output voltage is quite large. A 500mA sine wave current drawn by your amp will induce an almost 500mV sine wave ripple because the regulator can't deal with this demand. Approximate only, in fact the numbers might be wrong. But you get the idea. A 500mV ripple on a 500mA current swing may be audible?
Just my 2c.

This being a mosfet device and not a transistor I doubt you would get that much variation from current alone.
If there was a ripple before the regulator the error amp M2 would adjust it out. I attached a plot of the input vs. output voltage. You can see it doesn't really care what it's fed so long as it's about 6 volts higher than the desired output. I can't see why you think it can't cope with 500mA

As far as the sound goes, I think you'll find that a "good" regulator can only benefit. Mains can vary up to 10%. I know a lot of other people on this forum feel they have improved their sound with one.