Should i try this ?

[MikeB]

Hi !

I just designed a new ClassAB-Amp, but maybe it's a little
strange...
What do you think, should i give this a try ?
The THD in PSpice says 0.0005% @ 2.8v 4ohm 1khz

Its specially the inputcircuit that i haven't seen yet,
maybe i violate some semiconductor-rules ?

cu,
Michael

[sajti]

Hi,

there is nothing strange in Your amplifier. There are some unusual solution, such as Q29, and the 666ohms in the VAS stage.
But I think it will work, If You put together.

sajti

[DarkOne]

Looks good, but why did you use current source against Q29?
You should use emmiter follower with Q30(as Q29) and place temperature compensation and bias circuit between colectors of Q31 and Q32. I think that it can be much better.

[MikeB]

Originally posted by DarkOne
Looks good, but why did you use current source against Q29?
You should use emmiter follower with Q30(as Q29) and place temperature compensation and bias circuit between colectors of Q31 and Q32. I think that it can be much better.

I thought, do it a different way... ;-)
In fact, the construction with the current source against Q29
showed up much better results. Q29/Q30 form a simple current
gain. Because of the constant current through the emitter of Q29
this transistor has much better linearity and has nearly a constant
current at its base, which reduces distortions in the VAS.
Especially the high harmonics above K3 seem to be eliminated.

The strange part in this circuit (i thought) is the inputmodel, using
complementary jfets.

Mike

[Tony]

q29/q30 forms a buffer circuit between the vas and output stage right?

[MikeB]

Originally posted by joan2
q29/q30 forms a buffer circuit between the vas and output stage right?

Yes, i wanted to avoid a tripledarlington, and this kind of buffer
seems to be very friendly to the VAS.
The 2nd advantage is, that it doesnt use complementary BJTs...
So no matching of devices here :-)

Mike

[traderbam]

why is R31 a difference value from R32?

[MikeB]

Originally posted by traderbam
why is R31 a difference value from R32?

It's not easy to match transistors in PSpice, so i compensated
with these differences. In real world, this would be a poti for
adjusting DC-Offset. With perfectly matched pairs, these two
would be identical.
I will try to match these in PSpice, this should greatly reduce K2,
i hope...

[traderbam]

Good way to adjust dc offset. However, a side-effect is to change the feedback factor between the two halves, which may or may not be what you want.

[MikeB]

No, i don't really want that, i think that this difference creates
many K2's, that's why the transistors should be matched.
But a minimal difference shouldn't do any harm.
I just ordered the jfet's (spuro.de), i will try this amp the next
days...