predriver stage in power amps

While I work out my own power amp design I have reached kind of a puzzle situation. I will be using +/-32V supply rails for standard push-pull output stage and there will be two parallel bjt output transistors with Iq of 1.25A per emitter. Now the big question arose when I started thinking which type of predriver stage I should use and more importantly what should be the optimal quiesent current for this stage?

So I looked through all the books and articles etc I own and it seems to me that there isnt much information available. Different designers seem to use current anywhere between 4-100mA. For the layout of this stage there actually seem to be only two versions to choose from (in my case). The first being two push pull transistors with a parallel resistor and cap between emitters and the other version where the emitter resistors of either trasistor are connected to opposite supply rail. The latter seems to be the least explored version...

Anyway. I would like to run the predriver stage also in pure class A. For the quiesent condition the output transistors have an Ib of about 25mA (Ib=Ie/min beta) and as they are in parallel 50mA per side. As the current with signal in any output transitor will roughly be anywhere between 0-2.5A the max Ib for any transistor would be about 50mA. If I want the output stage to keep working in class AB for harder loads I should add additional 10mA to it I suppose.
As I understand all this would mean that the predriver stage quiesent current should be about 120mA. I can't find any designs that use such high current for this stage. They mostly seem to use 5-15mA of current. Is it so because of heat and efficiency considerations? Has anyone experimented if the higher current in such configuration will give better result sonically? That being the ultimate goal anyhow?
Have you read Doug Self's 'Audio Power Amplifier Design Handbook'? If not, try and borrow a copy. It should answer most of your queries as in it he presents the results of a detailed analysis of power amplifier stages, including the output stage.

Basically, compound feedbaback pairs for the output are more linear than the darlington arrangement and this is what he uses in his Trimodal Class-A design. This design incorporates the ability to switch to Class-B working and has a novel quiescent current control circuit. The circuit was first published in Electronics World in May/June 1995. If you would like a copy of the circuit diagram please email me.

VAS driver stage bias

A good place to start with VAS driver stage bias is around the 5ma to 10ma range. Ideally you should have a look at the transfer curves of the device you wish to use and set the current to the most linear part of that curve. Any driver loading outside of this stage should be handled by a buffer stage in between the VAS and O/P stage.
This way you don't have to worry to much about the loading affects of the O/P stage...
Another good way of buffering the VAS as outlined in Douglas Self's book is to add a Beta enhancement stage between the error amp stage and the VAS.

I hope this helps

Anthony Holton