The resistor can help if there are any series resonances across the load. The cable has a series resonance if unterminated, the voicecoil capacitance may resonate with preceeding inductance (not sure how severe this can be since the voicecoil is very resistive), etc.
Also a .22R resistor could be fairly inductive.
Also a .22R resistor could be fairly inductive.
That is exactly what JLH did in his 80 W mosfet amp, 0.22 ohm instead of output inductor, and it sounds excellent to my ears.
What does it matter considering the resistance of the voice coil (no to mention a passive crossover) Re, which is around 5 Ohm for an 8 Ohm nominal impedance driver ? It seems that JLH just wanted to avoid the slight overshoot which happens with an output inductor (after this one), and pure capacitive loads.
What does it matter considering the resistance of the voice coil (no to mention a passive crossover) Re, which is around 5 Ohm for an 8 Ohm nominal impedance driver ? It seems that JLH just wanted to avoid the slight overshoot which happens with an output inductor (after this one), and pure capacitive loads.
That resistor is protection against capacitive load to the certain degree (cabling). There is no output inductor.
...Over a number of output devices, the physical dimensions start to limit the achievable (with decent stability margins) ULGF. My amp with 8 pairs is slightly under 1MHz
What parameter(s) do you think set the limit?
I don't see that actual transmission delay is likely to be relevant.
I think we are far from the limits set by unavoidable stray inductance and capacitance, but perhaps I miss your point.
Best wishes on the expiry of your "sin bin" time
David
CFA relevance is that it's the OPS that sets the speed limit. Faster OPS may use more of that "CFA" potential.
I dont think its stability margins that are the issue.
Its best to separate loop stability (loop gain and phase relationship) from parasitic instability (unwanted coupling mechanisms introducing oscillation) and treat them as two separate issues - although I commented elsewhere that the two do interact and one can trigger problems in the other domain.
For modelling purposes, you can insert approximate trace interconnect inductances into the output stage device leads to get a measure of potential parasitic instability problems.
If you google Ovidiu Popa, Syn08 or PGP Amp, you will see some interesting work that was done in this area.
Its best to separate loop stability (loop gain and phase relationship) from parasitic instability (unwanted coupling mechanisms introducing oscillation) and treat them as two separate issues - although I commented elsewhere that the two do interact and one can trigger problems in the other domain.
For modelling purposes, you can insert approximate trace interconnect inductances into the output stage device leads to get a measure of potential parasitic instability problems.
If you google Ovidiu Popa, Syn08 or PGP Amp, you will see some interesting work that was done in this area.
Last edited:
I dont think its stability margins that are the issue.
Its best to separate loop stability (loop gain and phase relationship) from parasitic instability (unwanted coupling mechanisms introducing oscillation) and treat them as two separate issues - although I commented elsewhere that the two do interact and one can trigger problems in the other domain.
For modelling purposes, you can insert approximate trace interconnect inductances into the output stage device leads to get a measure of potential parasitic instability problems.
If you google Ovidiu Popa, Syn08 or PGP Amp, you will see some interesting work that was done in this area.
Hello Bonsai,
I agree with you Bonsai, but I think that one of the problems with Ovidiu Popa, Syn08 analysis is that his output transistor models are not accurate enough to simulate the parasitic oscillations.
Another thing output devices characteristic vary even when they are matched from the factory, which means some amps work and some have a parasitic oscillation.
Regards
Arthur
Last edited:
Testing this ,Ostripper Gnome CFA ,with 2 pairs mjw21195,6 as output trans. amplifier oscillate badly.Hello Bonsai,
I agree with you Bonsai, but I think that one of the problems with Ovidiu Popa, Syn08 analysis is that his output transistor models are not accurate enough to simulate the parasitic oscillations.
Another thing output devices characteristic vary even when they are matched from the factory, which means some amps work and some have a parasitic oscillation.
Regards
Arthur
Testing again with one pair 2sa1943,c5200 everything is ok.The same with 2sa1295,c3264 (all ok).
Thimios.
Testing this ,Ostripper Gnome CFA ,with 2 pairs mjw21195,6 as output trans. amplifier oscillate badly.
Testing again with one pair 2sa1943,c5200 everything is ok.The same with 2sa1295,c3264 (all ok).
Thimios.
I dont find that surprising, the MJWs are slower devices, the amp is probably compensated for highish ULGF so lower distortion can be achieved.
I think manso is correct - you cannot replace fast output devices with slow ones without risking loop instability problems. You usually need to re compensate if you do that. I expect OS designed the comp for a specific set of output devices.
Arthur, I think the value in Syn08's effort is that he shows the reader how to do it and the result provides some indication of the possible problems. Of course, the results are more accurate the better the models.
Arthur, I think the value in Syn08's effort is that he shows the reader how to do it and the result provides some indication of the possible problems. Of course, the results are more accurate the better the models.
I think manso is correct - you cannot replace fast output devices with slow ones without risking loop instability problems. You usually need to re compensate if you do that. I expect OS designed the comp for a specific set of output devices.
Arthur, I think the value in Syn08's effort is that he shows the reader how to do it and the result provides some indication of the possible problems. Of course, the results are more accurate the better the models.
I sim simple VSSA with TIP3055 and TIP2955 (double emitter follower OPS), slew rate is 35V/uS. I change output transistor using MJL3281 and MJL1302, slew rate is 55V/uS.
The strength of CFA is being simple and give high slew rate. If we use slow output transistor it will loose the strength, then it is better to build VFA
Testing this ,Ostripper Gnome CFA ,with 2 pairs mjw21195,6 as output trans. amplifier oscillate badly.
Testing again with one pair 2sa1943,c5200 everything is ok.The same with 2sa1295,c3264 (all ok).
Thimios.
Hello Thimios,
Can you give a link to the complete Gnome CFA that you are talking about.
Regards
Arthur
Last edited:
I sim simple VSSA with TIP3055 and TIP2955 (double emitter follower OPS), slew rate is 35V/uS. I change output transistor using MJL3281 and MJL1302, slew rate is 55V/uS.
The strength of CFA is being simple and give high slew rate. If we use slow output transistor it will loose the strength, then it is better to build VFA
Hello Bimo,
Can you post the Unity loop gain of your amp with different ouput devices,
Regards
Arthur
Hello Bimo,
Can you post the Unity loop gain of your amp with different ouput devices,
Regards
Arthur
I sim for learning purpose only and never build these amp .
Attachments
Yes, here http://www.diyaudio.com/forums/solid-state/248105-slewmaster-cfa-vs-vfa-rumble-148.htmlHello Thimios,
Can you give a link to the complete Gnome CFA that you are talking about.
Regards
Arthur
Post#1476
Here.http://www.diyaudio.com/forums/solid-state/248105-slewmaster-cfa-vs-vfa-rumble-23.htmlHello Thimios,
What about the output stage , how many emmitter followers are used before the power output devices , how do you bias the outputs its not shown in post #1476 .
Regards
Arthur
Post #223
I used one pair out.
Here.http://www.diyaudio.com/forums/solid-state/248105-slewmaster-cfa-vs-vfa-rumble-23.html
Post #223
I used one pair out.
Hello Thimios,
You are using power output devices QNJW0281g and QNJW0302g as pre drivers for the output stage, is this correct.
Regards
Arthur
No,i used 2sc4793 ,2sa1837 as pre drivers.Hello Thimios,
You are using power output devices QNJW0281g and QNJW0302g as pre drivers for the output stage, is this correct.
Regards
Arthur
- Home
- Amplifiers
- Solid State
- CFA Topology Audio Amplifiers