Old fashioned simple preamp

[x-pro]

Originally posted by PMA
Distortion at 1V, 10kHz (simulated).

Yes, 0.01% is possible for a good combination of values

What was the gain in this case?

Cheers

Alex

[PMA]

Hi Alex,

The gain was 9 dB.

Regards,
Pavel

[x-pro]

Originally posted by PMA
Hi Alex,

The gain was 9 dB.

Regards,
Pavel

Hi Pavel,

here is a result of my simulation with 3 transistors and local NFB only for 9 dB gain and 1V RMS output (somewhat different circuit )

Cheers

Alex

[PMA]

Hi Alex,

excellent!
(somewhat different circuit, as you have said )

P.S.: anyway, your 3rd harm. is higher than mine ...

[x-pro]

Originally posted by PMA
Hi Alex,

excellent!
(somewhat different circuit, as you have said )

P.S.: anyway, your 3rd harm. is higher than mine ...

Pavel, in any case we can not compete to the circuit I attach here, by Michael Kulish from Russia. I made two simulations, one with and one without an output follower, both under the same conditions - 10 kHz, 1 V RMS output, 9dB gain. It is an excellent application of an error-correction way of linearization in a very simple circuit. It works in real life too .

Cheers

Alex

[anatech]

Hi Alex,
Try to simulate a real world load on your outputs and rerun the sims please. I always find unloaded measurements much more promising than loaded ones. Some circuits are much better than others in this respect.

-Chris

[PMA]

Originally posted by x-pro

Hi Pavel,

here is a result of my simulation with 3 transistors and local NFB only

Cheers
Alex

Hi Alex,

though it is maybe not apparent, as the component values are missing, the circuit I posted had also local NFB only. Gain is completely defined by R6/R3 (with respect to hFE, off course). R10 is very low in value and can be omitted completely. R4 is just biasing resistor and its value is like 10M. Input impedance is quite high (about 1M) and R4/R10 do not constitute global feedback.

Cheers,
Pavel

[PMA]

Originally posted by x-pro


Pavel, in any case we can not compete to the circuit I attach here, by Michael Kulish from Russia.

Cheers

Alex

Looks very interesting

[PMA]

Originally posted by x-pro


, by Michael Kulish from Russia. I made two simulations, one with and one without an output follower, both under the same conditions - 10 kHz, 1 V RMS output, 9dB gain. It is an excellent application of an error-correction way of linearization in a very simple circuit. It works in real life too .

Hi Alex,

hereby what I got from the Kulish circuit. No follower and no load. Excellent result, however.

P.S.: I have to add that this is for source impedance of 1k. Distortion very quickly rises with Zout of the signal source. This is very very different from my circuit, which yields quite same distortion even for 100k source impedance.

[x-pro]

Originally posted by anatech
Hi Alex,
Try to simulate a real world load on your outputs and rerun the sims please. I always find unloaded measurements much more promising than loaded ones. Some circuits are much better than others in this respect.

-Chris

The distortion of this circuit is load-independant without the output buffer, only the gain will change if you connect load (with a coupling capacitor) to the (out2) of the second (unbuffered) circuit.

Originally posted by PMA
P.S.: I have to add that this is for source impedance of 1k. Distortion very quickly rises with Zout of the signal source. This is very very different from my circuit, which yields quite same distortion even for 100k source impedance.

I've just copied the values from the original article. Transistors here are run with high currents so the input impedance is fairly low. You may increase the resistor values and reduce that effect thought 10K distortion would rise a bit. Also this is an error-correcting circuit and as such requires balancing of the correction - as you may see I had to adjust one of the values to partially compensate for the buffer distortion. In the same way it is (I think) possible to compensate some of the imbalance due to the source impedance if it is constant.

Cheers

Alex