Münster, a diy copy with modifications of an old Sony, have some suggestions?

[destroyer X]

All idea will be well received and may be tested too.

This schematic, is some "frankstein" copy, from an old Sony, Model 3200F.

That amplifier was very strange in many aspects, some darlington input transistor, called "Paraphase", and a very strange PNP with NPN connection in the output.

I go changing my way, and with simulator aid, i put it to work, no sure how it is working, as i only checked 1Khz sinus result, and 150 watts was too low result that i see i made a bad copy.

I imagine it can go to 240 or little more.

I decided to listen some ideas to help me.

I already could perceive that it needs some regulated input voltage, some CCS or something alike, as voltage fluctuations will affect dinamically the sound.....have to introduce some regulation to input.

Also the VBEs voltages are a little bigger than i like, but may be reasonable... do not know.

It is not the Sony.... was deep changed.....and deeply simplified in many aspects, having room to modifications.

If you see something too much strange, let me know.

regards.

Carlos

[destroyer X]

Here is the hand copy...the scratch schematic.

What a confused thing!

Carlos

[destroyer X]

Here some partial view of the supply.

regards,

Carlos

[destroyer X]

Made with good steel.

Carlos

[destroyer X]

No one tried to help me.

I am not so experienced the way you think, you may have contributions to give to me.

My experience is only assembling and hearing.

Also can feel the way some design will sound...but this fails too.

The modifications:

I made the traditional input voltage reduction using resistor and electrolitic condensers...No optimistic reduction, as sometimes the diy supply loose a lot of volts, i expect to reduction of 10 percent related the bias and main drivers, and made another reduction in the case of 15 percent voltage loss in the differential circuit.

Many parts were included...the supply rail resistor and other simple things that i use my memory to substitute values, those included parts are marked with yellow dots.

Some parts with values changed, are market with red dots

And the green mark is the electrolitic condenser removed.

The amplifier is showing unstability when over driven...and i really do not what to do.

regards,

Carlos

[destroyer X]

I do not know, really!, what is the cause.

regards,

Carlos

[MikeB]

Hi carlos !

Interesting circuit... Took me 1 hour to understand why it clips at
~36v output. It was the diodes, opening into reversebiasing,
shorting the whole circuit. Did you use 1n4002 diodes ? The sim-model
seems to have a maxrating of 100v.
After removing them i had 225watts into 8ohm.

Four other things:

- your 10k at input is shorted
- You really shouldnt have any resistors between diffamp-supply
and vas-supply (Your 330ohm), the voltage at diffampout is relative
to neg-supply, with the 330ohm you move the voltage at this point.
BUT, this gives a very interesting currentfeedback, boosting the
gain and enabling the whole circuit... very strange thing !
Without this resistor the whole amp is too weak...
You would need something like tripledarlington, or more current
in the vas, but then you need a buffer between diffamp and vas...
- You should increase resistors in feedbacknetwork by at least factor
20, ot the 8.2p is much too small, giving unstable circuit.
I would recommend 22k+250, this gives max swing at 0.7v input,
and quite proper feedbackcompensation with these caps.
Or 100k+2.5k, this removes any DC-offset and you don't need
all these resistor at input.
- For this bias in diffamp you should change the 3.9k at diff-legs
to 1.2k, or it is very unbalanced.

I would say, this accidental currentfeedback is something to study,
but the whole amp is some "standard" bootstrapped topology.
I think for this topology the AKSA is already state of the art.

If you are looking for highpower, this one is not the best topology,
you need at least one more buffer somewhere.
(predriver/tripledarlington, or more current into vas and buffer before vas)

Except this "currentfeedback", i would say, don't build this one.

EDIT: I forgot the 5th point, the resistors in VBE-multiplier are too
large, they don't supply enough current to the base, i recommend
2k+1350 instead of the 5k+2k. This gives stable ~100ma bias.

Mike

[destroyer X]

Now i am damn tired, it is too late in that world part.

I read your text fast, but will reflect seriously tomorrow morning, and will make changes, some of them you suggest.

I will complicate a little, two supplies voltages, and will move some parts.

The Frankstein product...using someone leg, other one arm.... but will keep that good sound thopologie, trying to make simple after complications..... will try to make it good and complicated, with all the good practices...and them, i will start to remove here, and there....hehe.... when finished... will be the standard AKSA, and the secrets i will not publish, not really secrets, some details that make's the difference.

I like to make those power machines, because off young boys, they love to melt speaker coils.... i do not use more than 5 watts normally.

regards,

Carlos

[destroyer X]

Will try to remove the diodes, IN5396GP
Differences of Voltage and Current feedback i do not understand, i "feel" some differences hearing, better to current.
Will move VAS current
Will ty 22pf in the feedback network...will also try the 22K plus 250.
Will reduce differential legs load resistors to 1.2

Your attention and effort was appreciated, thank you.

Carlos

[MikeB]

Originally posted by destroyer X
i do not use more than 5 watts normally.

Yes, that's the reason why i make my amps not very powerful, that
makes many things easier. As i have quite efficient 4ohm-boxes, my
24v-amps have more than enough power.

Btw, do not replace feedbackcap AND resistors, use bigger cap OR
bigger resistors ! Having this feedbackcap too big is evil...

Your "currentfeedback" comes from the vas, it modifies negrail for
diffamp. I am not sure what exactly happens, but if you remove it
the amp gets weak...

And yes, that's what i wanted to say, after having finished this one,
you have some AKSA-clone !

Mike