100watt MOSFET Power Amplifier

[emmi]

Will this amplifier works?

[CBS240]

Not for very long. The outputs will go into thermal runaway and fail. Those type of fets need to have thermal feedback. If you use Lateral type fets, as the circuit is designed to use, there will be no problem with thermal compensation.

R17 is wrong, 0.6V/47K~13uA. I suspect the IPS tail current is suppose to be a bit more than this.

[Ron AKA]

The Hafler DH-200 100 watt per channel DH-200 MOSFET amp works well. You can find the circuit drawing and component listings in the manual link below.

Hafler DH-200 Manual

[Nico Ras]

R17 should be around 390 ohms then all should work fine. The MOSFETs should be 2SK1058 and 2SJ162

[emmi]

Thank you all

[nigelwright7557]

Originally Posted by CBS240

Not for very long. The outputs will go into thermal runaway and fail. Those type of fets need to have thermal feedback. If you use Lateral type fets, as the circuit is designed to use, there will be no problem with thermal compensation.

R17 is wrong, 0.6V/47K~13uA. I suspect the IPS tail current is suppose to be a bit more than this.

Swap the pot for a Vbe multiplier and put the transistor on the heatsink.
I have used the irfp240/9240 in dozens of amps.

[astankov]

For better thermal compensation don't use bjt transistor. Use the same type as the outputs like IRF 610/530/540 or other. It tracks temperature much better but if you decide to use bjt will work too.

[ljm_ljm]

I suggest that in most cases, use BJT will be better.

They have a minimum VBE. The smallest pay more, and the power of the largest distortion utilization.

And often they better stability.

[CBS240]

The bias spread between the two gates will be 2 X Vgsth, about 7 to 8V. BJTs have different temperature coefficients than Fets, largly due to the higher transconductance around cut off at the region bias will be set. Using a BJT may overcompensate the Fets, but including an emitter resistor will decrease the gain and temp co. It may require some trial and error to get it just right. IMO, the circuit should slightly overcompensate the output stage bias thermal responce. Since the bias for mosfets should be set above the transconductance droop region, around 150mA, a slight overcompensation should not create problems. Besides, an undercompensated stage is obviously a problem. It becomes a problem to drive the fets all the way to the rail because of the large Vgs required to turn them on. Usually a higher tier rail voltage for the IPS and VAS is required, or a bootstrapped driver stage.

[ingenieus]

With right sort of MOSFETs you don't need a Vbe multiplier. The design below (by Randy Slone) shows how.

The design looks OK, except for problem with the current source pointed out by Nico.