another nofeedback amp

[fscarpa58]

Hy all

I need some comments and advices about the attached
circuit.
It has no feedback but degeneration.
It is a 150W/8Ohm amp il biased with 200mA per output device.
I am not very happy with the 2° stage since it distorts a quite
and has the lone purpose of give the VAS the rigth static voltage.

Any hint? criticisms?

any suggestions about used devices?

thanks in advance

Federico

[lineup]

I f you would call transistors with unique names,
will be easier to talk about it,
same with resistors and whatever.

I think you would need a couple of compensation caps.
Maybe across 4k resistor to positive rail, and same in negative.
I am not sure is best to set the gain from within bias circuit.

You can try to test your circuit without 8 power transistors.
Replace them with 2 resistors to create a dummy output.
This will tell if problem is in VAS or comes from output stage.

[rozak]

Hello Federico,

My suggestion:
1. Use JFET, MOSFET. Distortion will be almost 2 harmonic and
canceled in simmetrical topology.
2. Current source instead R-35k for independece DC op. point
from VCC.
3. Use buffer before every VAS for lower distortion.
4. You have 2 VAS stages. Make all of them with gain for lower
distortion in second.
5. Power stage need deep class A in no feedback amp.
6. Use DC feedback for lower and constant DC offset.

PSpice model with 0.05% distortion(Vout=20v, 20kHz, 8 ohm):

[myhrrhleine]

It looks like there is feedback around the two voltage gain stages, it just doesn't include the follower-outputs.

[rozak]

Originally posted by myhrrhleine
It looks like there is feedback around the two voltage gain stages, it just doesn't include the follower-outputs.

There is no feedack exept local degeneration.
R25, C3- DC feedback.

[myhrrhleine]

Originally posted by rozak


There is no feedack exept local degeneration.
R25, C3- DC feedback.

Ummm...

Are we looking at the same circuit?
The one Fredrico posted?
I see a feedback path from the second voltage stage,
through a couple of pots and resistors to the base of the first voltage amp.
That's a feedback loop around the two voltage gain stages.

[B Cullingford]

feedback and the circuit Fredrico posted:

I see a ground connection, so no feedback

Comments - I was going to add a little but I will have to think a bit more and maybe try a sim.

When you say it distorts is this from a real or 'sim' amp? That is have you built it?

Bill

edit: extra comment - Rozak I'm not sure that 'deep' class A is needed.

[lineup]

Originally posted by B Cullingford
Rozak I'm not sure that 'deep' class A is needed.

'deep class A' would make it easier to get good result

but should be possible to use class AB output stage, too
with a little thinking

[rozak]

Originally posted by lineup


'deep class A' would make it easier to get good result

but should be possible to use class AB output stage, too
with a little thinking

From my experience class A is always batter in no global
feedback amp.
If you find the way to use AB with the same result
sand a circuit.

[fscarpa58]

Hi,
thank you all for the comments.

First of all: it is a simulation!
Only when things will go at best in the virtual world I'll begin to tranfer
it to the real world.

If you would call transistors with unique names,
will be easier to talk about it,
same with resistors and whatever.

yes lineup, I am sorry. I add a new pic with the relevant information.

I think you would need a couple of compensation caps.

Since there is no feedback I have not to
compensate the open loop response for stability.
Only, I put a couple of caps to
attenuate a peak in the 1MHz region and after many trial I find
the predriver base to be the best locacation.

The output triple has been tested separately and it perform quite well
(not to forget that it is an AB stage). The problem is the second stage
it does not amplify but, for reason of static voltage levels, it can not
be a buffer so it distorts. How much? a little, from an absolute perspective,
but certainly more of the first stage (a buffer with very low dist) ad the third,
the VAS.
I will be happy to avoid it but I do not see how to do.

My suggestion:
1. Use JFET, MOSFET. Distortion will be almost 2 harmonic and
canceled in simmetrical topology.
2. Current source instead R-35k for independece DC op. point
from VCC.
3. Use buffer before every VAS for lower distortion.
4. You have 2 VAS stages. Make all of them with gain for lower
distortion in second.
5. Power stage need deep class A in no feedback amp.
6. Use DC feedback for lower and constant DC offset.

Hi rozac, thank you
1. I never design using FET, I don't know them and yes,
it is a limit of mine but soon I'll try.
2. I am trying to keep the number of acrive devices as little
as possible. Distortion is not affected since it is very slow in that point.
what do you mean with "for independece DC op. point
from VCC"?.
3. the VAS distortion is good.
4. I tried to give the second stage some gain and to lower that of the VAS,
but the results were always worst.
5. Sure, class A is better but true class A gives a number
of problems one for all: heat dissipation. I made a compromise:
about 10-12W/4 Ohm in class A (25W/8 Ohm). This give me an acceptable
performance, expecially in the amount of high order distortion components
(obviously not good as pure class A)
6.Thanks, I have implemented your suggestion and it works fine.
What about the time constant? I read somewhere that interference
with audio signal may occur (J.Curl ?)

I see a feedback path from the second voltage stage,
through a couple of pots and resistors to the base of the first voltage amp.
That's a feedback loop around the two voltage gain stages.

myhrrhleine
I redraw the circuit to avoid such confusion. Now it is more clear
that there no NFB loop.

When you say it distorts is this from a real or 'sim' amp?

a sim, a sim, B Cullingford !!
I analyzed every stage,alone and in the circuit, with res load and with
actual load, and found some facts:
- the major contributors to amp distortion is the output stage when out of class A.
- the second stage follows with a smaller values but I think it cabe done better.
- the other stages have a very very small dist.

From my experience class A is always batter in no global
feedback amp.

rozak, I think class A is always better period.
There is a cost problem however. (maybe not in MOSFET designs, I do not know)

I post this scheme in other forum and I receive some comments.
I'll be happy to hear from you about that.

1) criticisms about the interconnection between gain and bias pots.
2) about the choice of first 3 stages since BC547/557 are 45V devices.
3) about the SOA of MJL4281A/4302A.

I really don't understand the first point since the bias is really not
influenced by the gain pot (just a bit). I think that in a nofeed amp
we need a gain pot to equate the gain of the two channel. Moreover, I have in mind to
add a pot as dinamic balance (2° order harm minimization) as in the attached figure.

I have to say they are right on the second question. devices Q1 and Q2
(and their complementary counterparts) live
at a voltage difference VCE of about 50 Volts, not very variable.
I think to the couple BC546/556 (80Vceo) as a substitute. Any advice?
Q3 instead see a VCE that, at full power, goes from a few Volts up to
100 Volts. Which device is advisable as Q3 in your opinion?

Concerning the SOA I alwais test the amp at full power on a 4 Ohm load (300W)
and trace the V-I curve to verify it is well under the SOA DC limit.
Yes, this is a pure resistive load and I will be happy if some of you
will tell me the values of a reactive load to better simulate the real word
e.g. 4 Ohm paralleled with X uF at 10kHz and 4 Ohm with Y mH serie at 30Hz.

thanks to all

Federico

P.S. for rozak
I went in Israel on my honey moon in 99'. In a nice little village on the sea called
Shavezion. where is Rehovot?