diyAudio

diyAudio (https://www.diyaudio.com/forums/index.php)
-   Solid State (https://www.diyaudio.com/forums/solid-state/)
-   -   Stability of an amplifier under capacitive loading (https://www.diyaudio.com/forums/solid-state/359704-stability-amplifier-capacitive-loading.html)

steveu 13th September 2020 03:51 PM

I think knutn is trying to say that if the dominant pole is the output capacitance then more capacitance just makes it more dominant. But, or course, that's a plan that doesn't work well at all. For one reason, real capacitors large enough to create a dominant pole on the output has a self resonance, due to its inductance, that is below the 0dB loop gain frequency. The classic VAS Miller cap works because the self resonance of a ~100pF cap is very high and it is small and can be connected to the VAS transistor with very short wires/traces.
Capacitor Self-resonant Frequency and Signal Integrity | Advanced PCB Design Blog | Cadence

Another reason is that making the naturally dominant pole more dominant gives the best bandwidth vs creating another pole that is so low that it dominates the natural pole of the VAS. Even if the output is the VAS, the naturally dominant pole is the Cbc and not the collector loading.

knutn 13th September 2020 05:18 PM

2 Attachment(s)
i have attached my amplifier with common emitter output, so anyone who cares can play with different loads.

knutn 13th September 2020 05:24 PM

Before anyone asks: Yes, it is a modified (20 W) Hiraga. More details can be found at: Class A Audio Power Amplifier Mimir

spladski 13th September 2020 11:23 PM

Quote:

Originally Posted by knutn (https://www.diyaudio.com/forums/solid-state/359704-stability-amplifier-capacitive-loading-post6336573.html#post6336573)
I'll try to explain why I think it is better with a common emitter output stage.

A 'better' example might backup you up. A slow design with significant distortion would not be 'better' in many people's view.

spladski 14th September 2020 01:47 PM

What is the THD 1V in 200KHz into 8 ohm in parallel with 0.1uF?

mjona 14th September 2020 05:11 PM

Quote:

Originally Posted by profdc9 (https://www.diyaudio.com/forums/solid-state/359704-stability-amplifier-capacitive-loading-post6330880.html#post6330880)
I am working on building up Michael Chua's C300 amplifier which does include a Zobel load on it. I have a SPICE simulation of it going. I am trying to figure out what could go wrong with it through simulation. I have been able to put 10 to 20 nF on it without problems, but once 100 nF is directly loaded on it, oscillations start. I can increase the Miller compensation capacitor to bring the pole down to be stable, but if the loop bandwidth is reduced too much, THD increases. It seems to be in the neighborhood of 0.05% to 0.1% for 20 kHz @ 200 W output to an 8 ohm load (at least in simulation).



Why are you using MJE340 and MJE350 in your Vas - the datasheets don't give any clue as to the fT which makes them unsuited to this application.

profdc9 14th September 2020 05:41 PM

9 Attachment(s)
Ok, to get a better handle on the performance, I did SPICE analysis and repeated the analysis of Bonsai in this presentation to check my work:

Output L_1

The first set of plots is a sanity check to show that the simulation of the circuit from the presentation recapitulates the Bode plot results, and it appears to do so.

The second set of plots is the circuit I have been testing with a 33 pF Miller compensation feedback and MJE340/350 transistors as the VAS and Darlington drivers. The phase margin appears to be about 60 to 70 degrees.

The third set of plots is the circuit with a 100 pF Miller compensation feedback network and MJ15032/33 transistors as the VAS and Darlington drivers with 0.01 uF load and 1 uH series inductance output. The phase margin appears to be about 50 to 60 degrees.

The fourth set is the circuit with the 100 pF Miller compensation feedback network and MJ10532/33 transistors with a 0.1 uF load and 0 uH series inductance output, showing the appearance of the pole due to the capacitance on the output.


I think this might help explain some of the issues here, that the output is not unconditionally stable and probably needs the output inductor, though I think for the MJE10532/33 design 0.03 to 0.04 uF of capacitance can be tolerated before oscillations become possible.

duncan2 14th September 2020 05:54 PM

Your right Mjona online I cant find the Ft but in one equivalent book I have says --equivalent is BD232 which is 3MHz but in my Mullard equivalent book their higher spec one has the BD232 as a line driver in TV,s at 20MHz but I suspect its the lower spec =3MHz Ft as all the other ones near that year of manufacture have 3MHz .


There are different versions of the MJE340 .

mjona 14th September 2020 08:11 PM

Quote:

Originally Posted by profdc9 (https://www.diyaudio.com/forums/solid-state/359704-stability-amplifier-capacitive-loading-post6341269.html#post6341269)
Ok, to get a better handle on the performance, I did SPICE analysis and repeated the analysis of Bonsai in this presentation to check my work:

Output L_1

The first set of plots is a sanity check to show that the simulation of the circuit from the presentation recapitulates the Bode plot results, and it appears to do so.

The second set of plots is the circuit I have been testing with a 33 pF Miller compensation feedback and MJE340/350 transistors as the VAS and Darlington drivers. The phase margin appears to be about 60 to 70 degrees.

The third set of plots is the circuit with a 100 pF Miller compensation feedback network and MJ15032/33 transistors as the VAS and Darlington drivers with 0.01 uF load and 1 uH series inductance output. The phase margin appears to be about 50 to 60 degrees.

The fourth set is the circuit with the 100 pF Miller compensation feedback network and MJ10532/33 transistors with a 0.1 uF load and 0 uH series inductance output, showing the appearance of the pole due to the capacitance on the output.


I think this might help explain some of the issues here, that the output is not unconditionally stable and probably needs the output inductor, though I think for the MJE10532/33 design 0.03 to 0.04 uF of capacitance can be tolerated before oscillations become possible.

Forgetting questions about inductors etc for now 2SA1381 is a better choice for the Vas position because this extends the unity gain bandwidth.

Bonsai has mentioned the rate of decline in gain of 20dB/decade with compensation which is equivalent to an angle of 45 degrees starting from the point where the gain reaches - 3dB (0.707).

If you go with 2SA1381 the -3dB frequency will increase and the point where the phase angle starts to change will also increase - this is around a decade in frequency below the new -3dB point.

The decade rule also applies to compensation in that the Vas pole has should be ten times greater than any other pole to be dominant.

C6 has been mentioned in a different context in a previous post. Bob Cordell has cautioned against using lead capacitors. This could also compromise the dominance of the Vas compensation.

spladski 15th September 2020 01:43 AM

Add cap from C9 collector point to ground, and make it big enough, it should help with the capacitive load. You can also try chucking away C6.


All times are GMT. The time now is 08:52 AM.


Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2021 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 17.65%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2021 DragonByte Technologies Ltd.
Copyright ©1999-2021 diyAudio

Wiki