150W MOSFET Amplifier with IRFP250x2

[russel]

Originally Posted by apexaudio

BIAS 100mA
THD< 0,05%; TIM< 0,07%
F 20Hz-20kHz +/-0,1dB; 2Hz-250kHz +0/-3dB
Damping >200
S/N >108dB

I am afraid these figures are all but meaningless unless the conditions they are taken under are quoted as well.

[apexaudio]

Figures taken from Elektor

[apexaudio]

Current limiter is set on 3,3A. For IRFP250 resistor 0,22R 5W must be replaced with 0,15R 5W.

[apexaudio]

Originally Posted by latala

I am interested in your amplifier can you please give furthere details
regards Trev

If you interested about PCB, I take time to design new version with over current limiter.

[kenpeter]

Really? I had assumed BF871's purpose was already to fix one cold emitter's
voltage drop across the pair of .22R by stealing bias current away from the
driving pair. That alone should assure a runaway-less output current that is
pretty much forced to follow class A rule of I1+I2=Constant. But this might
only happen when there is far more idle current dialed-in than desirable.

If you bias less than the full linear class A rule, perhaps AB by mosfet square
law rule??? Then that limiter transistor would be turned off most of the time.
I'm not sure sonic goodness of letting a transistor sometimes turn completely
off. If its always off, thats OK. Its still there from a safety standpoint.

I think you can get close to AB efficiency without ever turning anything off.
The class I speak of is Hyperbolic A. If you exchange each of those 0.22R for
0.1R + MBR745 Schottky in series. And bias CC up high enough that the limiter
is always in effect as a shunt regulator.

Holding voltage drop across the sensing resistors and Schottkys to constant.
Will give back a very square law curve, following rule of cold Schottky Diodes
in the sensing totem, rather than square laws of the MOSFETs. Neato! Cause
now nothing ever turns off, nor runs away with hot output devices.

The hyperbolic A curves are close to AB. You might set idle current ~600mA?
I can draw a picture tomorrow if it helps. We aren't talking bout mods severe
enough to need a new layout if one exists (I'm assuming it does)... Just two
diodes in series with slightly different choice of sense resistors.

[apexaudio]

This is Yorkville amplifier schematics with IRFP250.

[kenpeter]

Yeah, sorta like that. You see Yorkville Q9+Q10 sensing a similar function.

I don't think it needs to be that complex. I see upper half bootstraped like
Aleph, Lower half not needin' bootstraps cause its Quasi. And the whole
ball of crossover correction driven from between emitters like an Allison.

How close will yours pull to the upper rail without a bootstrap or separate
supply voltage for the gate drive?

[apexaudio]

About 7V, but simplicity for DIY is priority.

[knairkuttan]

Originally Posted by apexaudio

This two old modules (about 20 years) same circuit without curent limit, with BF422/BF433 and BUZ21x4 stil in use work on +/- 36 DC.

how many waats rms ca give it? can I use it as subwoofer amp plz give me details

[kenpeter]

Originally Posted by apexaudio

About 7V, but simplicity for DIY is priority.

This is what happens when I try to "fix" something what ain't broke.
At first, I couldn't get my mod to behave as hoped/described above.
But after screwin' round several hours, I come up with this instead???

Its almost a Visch glitchless class B crossing, plus a small pair of
constant currents loads that prevent anything but Schottkys from
actually turning off. Yes, like B , yet somehow not B , what???
I think it may be thermally stable, though not for AB usual reasons.
Except for those constant current, its trying very hard to shut off.

I think you can still see a lot of Apex's original design and topology
deserving credit. Certainly no longer anything one could hope to
patch into an existing layout. I won't distract this thread further
with this tangent. I just thought you might want to see this crazy
ideas you inspire drawing till 2:57AM on a night before a workday.

Those currents were plotted at the verge of clipping at 22KHz.
They look even nicer at 1KHz, but that would be cheating...
And it will tolerate a little clipping without goin' banannas.