hi MikeB
Thank you very much!!!
honestly I prefered your previous one, but this one is nice too!
but seriously
as you use ac input source the input cap filters HF so at output you have low amplitude of HF. This may 'mask' the local maximum in freq. response (read unstability)
try repeating ac sweep without it
regards
Thank you very much!!!
honestly I prefered your previous one, but this one is nice too!
but seriously
as you use ac input source the input cap filters HF so at output you have low amplitude of HF. This may 'mask' the local maximum in freq. response (read unstability)
try repeating ac sweep without it
regards
That's right, I also had a couple of cases when I had to remake frequency correction in my AB-amplifiers 'in real life', after it seemed to be defined well during simulation (I used Microcap 7 program and models), although p.c.b. layout was done very carefully regarding parasitic capacitances etc.
Anyway, basically this conception looks 'fresh blood' to me and I appreciate your idea Mike. I hope there would be no big problem to eventually add some RC in this amp and it won't shoot its brilliant performance
Anyway, basically this conception looks 'fresh blood' to me and I appreciate your idea Mike. I hope there would be no big problem to eventually add some RC in this amp and it won't shoot its brilliant performance
sam9 said:
It's only a first step to establishing stability...need to perform transient analysis...thereafter:
http://www.enel.ucalgary.ca/People/Haslett/Enel647/ReturnRatioStability/rrSim.pdf
Hi Mike,
I have been thinking about why your amplifier is not oscillating, and I think the answers are
No common emitter stage; ie. less inter-device hf phase shift / feedback from your differential + cascode + mirror + follower output.
The differential is working harder than is often the case when driving a VAS etc.; ie. collector current variation is a much higher proportion of tail current than normal. With higher input impedances for this stage it has less hf capability (gain) for stray voltage field pick-up to influence the cascoded outputs.
From your posts it looks as if you have put the simulator generator directly to the input transistor base; this will give you different amplitude - phase - distortion responses to what happens in real life. You must feed through the input resistors + capacitor.
Jan's comments have been relevent. Where do you measure your output ?
Before the choke is not the same as at the load.
Also have you tried distortion simulation with a virtual loudspeaker in place of the resistor?
Why is the choke fitted, does the amplifier not work without it?
Cheers .......... Graham.
I have been thinking about why your amplifier is not oscillating, and I think the answers are
No common emitter stage; ie. less inter-device hf phase shift / feedback from your differential + cascode + mirror + follower output.
The differential is working harder than is often the case when driving a VAS etc.; ie. collector current variation is a much higher proportion of tail current than normal. With higher input impedances for this stage it has less hf capability (gain) for stray voltage field pick-up to influence the cascoded outputs.
From your posts it looks as if you have put the simulator generator directly to the input transistor base; this will give you different amplitude - phase - distortion responses to what happens in real life. You must feed through the input resistors + capacitor.
Jan's comments have been relevent. Where do you measure your output ?
Before the choke is not the same as at the load.
Also have you tried distortion simulation with a virtual loudspeaker in place of the resistor?
Why is the choke fitted, does the amplifier not work without it?
Cheers .......... Graham.
mikes,
Thanks for the url. I recall asking for info on the topic about 3 years ago and getting little response, but this had not been published at the time.
Since I tend to build on earlier designs that I know work, I've adopted a practice of starting with a Miller cap that large enough that I'm 90% sure will be stable, then when the rest of the design is functioning, I decrease the cap value until either instability sets in (in which case I retreat a step or two) or it's low enough that I don'e believe there is any practacl advantage to going further. Of course, one has to keep track of tail currents at the same time, but there is some latitude so that they don't need to be changed with each step.
If I could do a better prediction in Spice, I could be braver with trying out new ideas.
Thanks for the url. I recall asking for info on the topic about 3 years ago and getting little response, but this had not been published at the time.
Since I tend to build on earlier designs that I know work, I've adopted a practice of starting with a Miller cap that large enough that I'm 90% sure will be stable, then when the rest of the design is functioning, I decrease the cap value until either instability sets in (in which case I retreat a step or two) or it's low enough that I don'e believe there is any practacl advantage to going further. Of course, one has to keep track of tail currents at the same time, but there is some latitude so that they don't need to be changed with each step.
If I could do a better prediction in Spice, I could be braver with trying out new ideas.
Hi graham !
Yes, the idea was to reduce voltagegainstages to reduce delays.
This "melts" diffamp and vas into a single stage.
In sims i only placed the probes at the bases, the feeding was
normal, visible as the input get's a 90° shift for higher freqs.
I placed the probe before outputchoke to be sure HF is not filtered out.
The amp needs outputchoke to be stable into capacitive loads.
I haven't tried speakersim yet, but will do !
Mike
Yes, the idea was to reduce voltagegainstages to reduce delays.
This "melts" diffamp and vas into a single stage.
In sims i only placed the probes at the bases, the feeding was
normal, visible as the input get's a 90° shift for higher freqs.
I placed the probe before outputchoke to be sure HF is not filtered out.
The amp needs outputchoke to be stable into capacitive loads.
I haven't tried speakersim yet, but will do !
Mike
Yes, the idea was to reduce voltagegainstages to reduce delays.
This "melts" diffamp and vas into a single stage.
I think that delay is not so important, if You apply low overall feedback. Use local feedback to set the gain stages linear, and reduce the overall feedback to the minimum. I think, that 12-16dB is enough. Of course the distrtion will be higher, but there is no problem if it goes up to 0.2% from 0.02%....
sajti
This "melts" diffamp and vas into a single stage.
I think that delay is not so important, if You apply low overall feedback. Use local feedback to set the gain stages linear, and reduce the overall feedback to the minimum. I think, that 12-16dB is enough. Of course the distrtion will be higher, but there is no problem if it goes up to 0.2% from 0.02%....
sajti
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.