Leach - mixing Power Transistors

[Arco]

Hi,
I have one question:
Can I mix different type of power transistors in Leach amp Power stage?

I Checked Toshiba (2SA1943/2SC5200) and Motorola (MJL21193/MJL21194) separatelly.
I found different sonic characteristics of this devices and... I want to join high speed nature of Toshiba with true power of Motorola.
I found that Toshiba transistors warms more than Motorola and I suspect that thermal compensation can make a problem...

Dismounting amp module for real test is... twister...


Please forgive me my english...

Regards,
Arco

[john curl]

This is dangerous. However, with a one-off, it might work OK. Right now I am struggling with current hogging with the SAME type output transistors in parallel, just different batches, in my JC-1 power amp. Yes, I have plenty of heatsink, etc. I'm amazed that we have even seen a problem. If your emitter resistors are large enough, and the betas are not too different, it will probably work.

[sajti]

I agree with Mr. Curl. It might work. But You need at least 0.5ohm (or .68-1ohm) emitter resistors to get same current, and I recommend to use small series resistors too, at the base of the output devices. (2.2-4.7ohms/1W)

Sajti

[peranders]

I doubt that you can achieve positive effects with mixing different types in the same output stage. Maybe you can do as Denon did in an old POA amp. They had two different amps which were connected in parallel.

I think this is a typical "free" idea but the reality will be present. It's hard enough to get matched devices working. Epsecially when you intend to use high speed transistors TOGETHER with slower ones, I think you are asking for trouble.

My statement is only a design engineer hunch but it's no harm in testing.

[Arco]

Thank you for all answers.
Reply in general is - it's dangerous but... possible!
I should check this configuration.

I have Toshiba & Motorola matched transistors (with hFE 100).
I can change emiter and base resistors - no problem, but please tell me now: what about thermal compensation for mixed transistors?

Arco

[sajti]

Use same heatsink for all the output devices. Put the bias diodes on this heatsink, as recommended.

Sajti

[mikek]

Originally posted by john curl
This is dangerous. However, with a one-off, it might work OK. Right now I am struggling with current hogging with the SAME type output transistors in parallel, just different batches, in my JC-1 power amp. Yes, I have plenty of heatsink, etc. I'm amazed that we have even seen a problem. If your emitter resistors are large enough, and the betas are not too different, it will probably work.

If you are indeed THE John Curl of JC-1 fame, then i would suggest this forum is well on its way to becoming the leading online dialogue resource for audio designers worldwide.

On a technical note, examination of the Curl-designed HCA220011 gain-block schematic, reveals it to be a derivative of the electrocompaniet circuit shown here:

http://home.online.no/~tsandstr/schematics_2.htm

I have found stability of 2nd stage quiescent current to be rather a problem, being overly sensitive to the tolerances of first stage collector resistors, and second stage tail resistors.

Has anyone else had any experiance with this circuit, or attempted to incorporate current mirrors in each of the complementary diff. pair to eliminate second harmonic distortion?

[mikek]

Originally posted by john curl
Right now I am struggling with current hogging with the SAME type output transistors in parallel, just different batches, in my JC-1 power amp.

Current hogging, even with emitter ballast?? Almost certainly local oscillation.....

Try base resistors mounted as close as possible to each of your output transistors.