FET+Bipolar follower biasing (To mix, or not to mix?)

[Wavebourn]

Hi everyone,
I need your opinion.

I am building a hybrid amp and need a powerful voltage follower for 500W on 2 Ohm load.
I think of using one complementary pair of FETs driving bunch of complementary pairs of bipolar transistors.

Now, this is the question.

However, idle current of bipolars may be regulated as usual by a transistor on heatsink, for FETs this transistor may stabilize some more voltage. The question is, is it safe enough, can requirements to bias FETs screw down bipolars? What tricks may be done? Does someone have any experience with such topologies? Does an idle current of FETS need to be high enough up to the self-temperature stable point?

To mix, or not to mix? This is the question... Before I start blowing up transistors...

[Nelson Pass]

Using Mosfet as drivers for bipolars carries some additional
complexity as the temperature coefficient of the Vgs of the
Mosfets is large compared to the Bipolars. The easiest way
around this is to but a bias network across the Bases of the
Bipolar banks or otherwise regulate the bias voltage across
the Bases. A good example of this would be my patent
#4,752,745 where an opto-isolator is used to feedback the
bias across the output bank, which would largely eliminate
the effect of the Vgs drift.

[Wavebourn]

Originally posted by Nelson Pass
Using Mosfet as drivers for bipolars carries some additional
complexity as the temperature coefficient of the Vgs of the
Mosfets is large compared to the Bipolars. The easiest way
around this is to but a bias network across the Bases of the
Bipolar banks or otherwise regulate the bias voltage across
the Bases. A good example of this would be my patent
#4,752,745 where an opto-isolator is used to feedback the
bias across the output bank, which would largely eliminate
the effect of the Vgs drift.

Thanks Nelson;
I thought already of 3 possibilities:

1. Idle current of FETs equal to zero temperature coefficient point, then bias bipolars as usual;
2. Idle current of FETs more for the reverse coefficient to cancel bipolar's coefficient;
3. separated biases...


...and... while answering you I come accross one more solution I like more: the single FET follower and CCS. Wow!

Thank you, you've kicked my brain in the right direction!

Anatoliy

[Nelson Pass]

Originally posted by Wavebourn
1. Idle current of FETs equal to zero temperature coefficient point, then bias bipolars as usual;
2. Idle current of FETs more for the reverse coefficient to cancel bipolar's coefficient;
3. separated biases...

1 That is usually a very high current.

2 That is an even higher current.

3 Most likely candidate.

and of course the SE biased follower comes under (3)

[Wavebourn]

Originally posted by Nelson Pass


1 That is usually a very high current.

2 That is an even higher current.

3 Most likely candidate.

and of course the SE biased follower comes under (3)

..even better, much simplier while more stable.
Output current is already limited in one direction, in other direction it may be limited by a single zener that as well protects the gate; simplier servo loop with single integrator that don't affect bias, and lower input capacitance.

[mlloyd1]

not if you use lateral mosfet drivers instead of hexfet/vertical?

Originally posted by Nelson Pass


1 That is usually a very high current.

2 That is an even higher current.
...

[Wavebourn]

Originally posted by mlloyd1
not if you use lateral mosfet drivers instead of hexfet/vertical?

Good idea, thanks! But I don't have them, and already made a PCB with SE driver.