PP MOSFET amp with float supplies

[Kuzmenko]

There is VERY simple MOSFET amp with only 1 dual op-amp and 2 same n-MOSFET.

http://cp.people.overclockers.ru/cgi...fet_bridge.png

Very good THD and thermal stability, no need of precise resistors. Automatic DC balance on output. Only 1 variable resistor (R37) is used for setting quiescent current, so you may choose, would it be class-A or class-AB amp. No need of short protection circuit.

All components are CHEAP, i.e. TL082/IRF540, etc. With suggested supplies (all 15V) it about 20-25W on 4 Ohm load. Pay attention, that not all supplies are connected to the ground! Bipolars are not welcomed here, but you may try something like BDW93 or so...

PS with NE5532 there is possibility to use higher supply volteges, i.e. 22V.

[edl]

the link does not works.

let's try this

[djk]

That circuit was designed for BJTs, with Hexfets you run the risk of thermal runaway.

You need a TIFF view, them click on 'images'

http://patft.uspto.gov/netacgi/nph-P...S=PN/4,229,706

Only one ±22V supply is need for the opamps no matter how many channels you are building.

Maximum of 40V on the floating supplies, two floating supplies need for each channel built.

With 40V supplies you can get about 120W at 4 ohms, 70W at 8.

[Joe Berry]

Originally posted by djk
That circuit was designed for BJTs, with Hexfets you run the risk of thermal runaway.

Maybe not - I think it will depend on whether the opamp's internal output stage bias is thermally compensated. If it is, the bias on the output MOFSETs should drop slightly with warmup, which of course is what you want. It may track even better with everything on a small PC board, mounted with the MOSFETs on the heatsink.

Kuzmenko, what bias are you running, and how is the sound? Looks like a fun project.

[djk]

"Maybe not - I think it will depend on whether the opamp's internal output stage bias is thermally compensated. "

I'm sorry you don't understand how opamps work and how to read a spec sheet for same.

With a HEXFET you have two choices, run it class A or use servo bias.

Use a lateral MOSFET or a BJT or don't bother with this circuit.

[Joe Berry]

djk,

Maybe I should have said that it depends on how the opamp behaves thermally when its output terminal sits at +4V or so rather than near ground, as this could have an effect. I did just check the TL082 spec sheet, and as I expected, it does not address this issue. I imagine that one would just have to build it and see. Have you done so?

[Joe Berry]

On second look, if the opamp holds the output terminals at 0V as it should, and if V12 remains constant over time, the MOSFET bias voltage is determined by Ohm's law and should remain constant regardless of the opamp's thermal behavior. I actually explained this to someone once, but now seem to have forgotten it myself!

Given that, I too would expect to see the MOSFET bias current creep up over time. If this isn't happening in the real circuit, there may be other reasons. Maybe the output terminals are drifting slightly above 0V, or perhaps V12 is sagging gradually over time. Kuzmenko, maybe you can take some measurements and let us know what is really going on here.

[Kuzmenko]

Is it good idea for float sullies to use toroid transformesf for 12V halogene lamps? I can buy it in numerous variants, from 40W to 300W 8-)

Originally posted by djk
That circuit was designed for BJTs, with Hexfets you run the risk of thermal runaway.

I`ve done a lot of simulation using MC7 and can say: there is NEGATIVE thermal runaway of Iq with IRF540. Rose global temp from 1`C to 50`C leads Iq to drop from 300mA to 210mA for example. But I don`t assemble this amp yet, it still in project...

Originally posted by djk
"Maybe not - I think it will depend on whether the opamp's internal output stage bias is thermally compensated. "
With a HEXFET you have two choices, run it class A or use servo bias.
Use a lateral MOSFET or a BJT or don't bother with this circuit.

Opamps have thermo DC offset about few mV. It`s much lower than MOSFET treshold.
Btw, i WILL run it in class-A but for my own reason.
There is thermally compensated Iq control circuits for BJT or strange ;-) MOSFETs:
http://cp.people.overclockers.ru/cgi...dge_thermo.png
or
http://files.people.overclockers.ru/...dge_thermo.png

PS please, let me know, what link works better.

[fscarpa58]

Hello

This is an implementation of US patent 4,229,706,
that is the famous Sumo 9 plus from J.Bongiorno.

Bye

Federico

[fscarpa58]

or its simpler version

THE NINE

unfortunately it uses unavailable output devices.

Federico