Unusual amp from 1987

[stanislav1957]

Choke instead of a capacitor in the local OOS, try using 0.5-1U . Several turns of 0.1-0.2 mm wire are wound on the housing of the R20 3-10 Ohm resistor

 

[minek123]

I knew at the end you are going to slap a coil on it
==================
Я знал, что в конце концов ты накинешь на него катушку

 

[stanislav1957]

I knew at the end you are going to slap a coil on it
==================
Я знал, что в конце концов ты накинешь на него катушку

я и не снимал.
The loop gain increases, the anti-overload capacity increases, and intermodal modes drop.

 

[minek123]

Maxim, did you try to simulate with speaker capacity added (e.g. 10nF) ?
Please try to simulate square waves with 10nF capacity added, and WITHOUT RL
network at the output.

 

[stanislav1957]

With deep OOS, the response to capacitance is like all such amplifiers . With a multi-loop nested OOS, it is more stable. The more powerful the VC,the more pairs, the better

 

[minek123]

I tried, and on this amp it's impossible to get any kind of non-oscillating square waves without output RL, if even 10n capacitance is added to the speaker.
==================
Я попробовал, и на этом усилителе невозможно получить какие-либо неосциллирующие прямоугольные волны без выходного RL, если к динамику добавить даже 10n емкости.

 

[stanislav1957]

145db of total loop gain. You can't do it without an RL filter

 

[minek123]

Yeah, that's the conclusion: RL is a must on this amp.
2uH + 3 Ohm seems to be working fine.
Caps (C11/C12) can go down to 47pF (from 120pF in my last sim), and Thd at 1kHz is 0 (zero) with 2 pairs of HexFets.

 

[stanislav1957]

на 20кгц смотри
At 20kHz it is better to watch

 

[minek123]

I know, I'm checking at 1kHz, 10kHz and 20kHz.

 

[stanislav1957]

100гц

 

[minek123]

OK, that's it. Moving on to the PCB design.
======
ОК, вот и все. Переходим к дизайну печатной платы.

 

[stanislav1957]

ок.Well. . I will work with the composite in "B" mode for now. On a pair of opamps

 

[minek123]

Please have a look at this plot: this is current flowing through output devices of the quasi hexfet amp.
Upper devices have 'rounded' edges around 0.0V (bottom of the plot), lower devices turn off sharply.
Upper devices are conduction more/longer than lower devices.
Why the difference? Is it good or bad?

 

[stanislav1957]

Check the quiescent current on both outputs and drivers.

 

[minek123]

Outputs:
Upper: 94mA
Lower: 92mA

The bigger the idle current - the upper device conducting longer and longer.
Lower device - no difference.

Drivers:
Upper: 24mA
Lower: 28mA

 

[stanislav1957]

Try placing a capacitor between the emitters of the drivers

 

[minek123]

If I lower idle current (say 20mA per device) these difference in conducting are very small, but still there...
Question: is it good or bad?
I guess it shouldn't hurt if they overlap a little bit..

 

[stanislav1957]

Не повредит

 

[minek123]

>Try placing a capacitor between the emitters of the drivers

It's oscillating with a cap. I tried 1uF and 100nF. Same behavior.
My guess is it's caused by LEDs added by Steve.
It does lower Thd (by half), so perhaps it's OK...
This is kind of experimental-quasi topology
==============
Он колеблется с крышкой. Я пробовал 1 мкФ и 100 нФ. Такое же поведение. Я предполагаю, что это вызвано светодиодами, добавленными Стивом. Он снижает Thd (вдвое), так что, возможно, все в порядке ... Это своего рода экспериментально-квазитопология.