power amp with transformer output

[techie]

First of all big thanks to audioguru who helped me out with an previous amp design. Sadly i couldn't use it because transformer has just too big dynamic resistance.
I researched and worked on my own and i came up with an alternative design. I managed to get everything working properly but just need to modify one last thing.

I need to do a 180 degrees phase shift of signal before the base of the
transistor Q14. After that just trim the voltage gain a bit and I'm done.

I need help with that phase shifting. I already tried using various combinations in a place of the lover Darlington pair without any success. I tried compound pair and various combinations of transistors npn and pnp with and without resistors.
I thought of using an OP amp to do the phase shift, but thats a really lame solution.

 

[Willbe]

I need to do a 180 degrees phase shift of signal before the base of the
transistor Q14.

Q14 is supposed to be the PNP complement of a 2N3904, same with Q13. . .no?
You'll also have crossover distortion because there are no emitter resistors.
BTW, these final four transistors only supply current gain; Q15 does the gain.

 

[techie]

I tried all combinations of transistors with and without emitter and collector resistors but still simulation results turn out horrible. Maybe its a bug in Multisim ...
Theoretically changing Q14 to 2N3906 (PNP complement of 2N3904), putting a resistor in its collector and connecting the base of Q13 between collector and resistor of Q14 should work. But simulation doesn't agree to that...

 

[MrAl]

Hi there,

Yes i see your design which looks like a roll your own design wont work
without a phase shifter, you are very right about that, but i dont see
what is so lame about using an op amp or two? There is a little more
that goes into a high quality amplifier like this than meets the eye, and
some of that has to do with the amount of distortion on the output.
Using op amps, you can control the transistors pretty well.

If you want to stick with transistors then perhaps you can try just
building an inverter (that's 180 degree phase shift) with a gain of 1
and use that to drive the lower section. The trick is to get the gain
to be 1, and have it stay that way even if some change in ambient
temperature. The top gets driven the way it is i guess, and the
bottom gets driven with the inverter.

I cant say that this thing is going to sound that good though, but at
least you will be on your way to a better amplifier.

You also have to do a quick check on quiescent bias current to make
sure it is not enormous, which a simulator wont alert you to and
in real life would draw wayyyy too much current to be feasible.

The way to beat crossover distortion is to keep the 'off' transistor
actually turned on a little, even when it is supposed to be off.
The trick here is to make sure the off current is high enough to
keep the transistor in the linear region but not so high that it eats
up power and heats up the transistors.

It's understandable that the circuit doesnt work as is, because the
transformer is not being driven with the correct phases. Once this
is corrected you might have a workable design in progress.

 

[Ubergeek63]

First of all big thanks to audioguru who helped me out with an previous amp design. Sadly i couldn't use it because transformer has just too big dynamic resistance.

Just how much audio power and what quality are you trying to get?

 

[audioguru]

The output transistors should have a low value emitter resistor and DC negative feedback to control their idle current.

The circuit drives a motor so maybe the very high distortion doesn't matter.
The first simulation accurately shows the high distortion (about 40%) of a transistor without any negative feedback.
Here is my simulation:

 

[flat5]

Hello audioguru,

May I ask what specific software you have used for the simulation?

 

[Hero999]

He uses LTSpice which can be freely downloaded from Linear Technology's website.