Hi Stein,
Being 'skilled in the arty', of course you will know the answers.
Besides, most of the questions were already answered.
Your welcome. I hope it will help.
Cheers,
E.
Hello Edmond,
>> But, as with any change, I have to simulate all loops all over again (a lot of work!), in particular the clamping loops. Probably some changes will be needed.
When you refer to loops do you mean every loop in the amp literally. Sorry if it seems like a stupid question, but you have a lot of compensation I dont normally see in other circuits which implies a more thorough stability analysis like testing all the loops to see if they are stable and eliminate parasitic oscillations.
Regards
Arthur
>> But, as with any change, I have to simulate all loops all over again (a lot of work!), in particular the clamping loops. Probably some changes will be needed.
When you refer to loops do you mean every loop in the amp literally. Sorry if it seems like a stupid question, but you have a lot of compensation I dont normally see in other circuits which implies a more thorough stability analysis like testing all the loops to see if they are stable and eliminate parasitic oscillations.
Regards
Arthur
Hi Arthur,
Indeed, I mean every loop. Most of them I've already simmed separately, but if you put the latest modification all together, much chance that I will overlook something. Right now, this happened already with the over-current clamp: Ib of the pre-drivers becomes too high when the protection kicks in.
Cheers,
E.
Indeed, I mean every loop. Most of them I've already simmed separately, but if you put the latest modification all together, much chance that I will overlook something. Right now, this happened already with the over-current clamp: Ib of the pre-drivers becomes too high when the protection kicks in.
Cheers,
E.
Phoenix13
Below some loop responses.
1st pic: global NFB loop at R7
2nd pic: CFB loop at R26 (global NFB loop disabled)
3rd pic: local Miller loop at C20 respectively C21 (other loops disabled)
Notice that the phase margin of all these loops are close to 90 degrees (as it ought to be).
Cheers,
E.
edit: I've doubled the values of the shunt compensation caps: C26=C28=47pF & C29=C30=220pF.
Below some loop responses.
1st pic: global NFB loop at R7
2nd pic: CFB loop at R26 (global NFB loop disabled)
3rd pic: local Miller loop at C20 respectively C21 (other loops disabled)
Notice that the phase margin of all these loops are close to 90 degrees (as it ought to be).
Cheers,
E.
edit: I've doubled the values of the shunt compensation caps: C26=C28=47pF & C29=C30=220pF.
Attachments
Last edited:
Hi Edmond,
This looks really great. Am I to understand that we are getting close to a "first cut"? I gather that there will be some additional fine tunning in real life circuit.
One question though. What about DC-servo for different op-amps? Should that be included as optional?
I will have a closer look and get back.
Best regards,
Mogens
This looks really great. Am I to understand that we are getting close to a "first cut"? I gather that there will be some additional fine tunning in real life circuit.
One question though. What about DC-servo for different op-amps? Should that be included as optional?
I will have a closer look and get back.
Best regards,
Mogens
Hi Edmond,
Please ignore message #88. I did a little searching this morning, on my way to work, and can see that this has been discussed in other threads and judging from the values of the emitter resistors I think you are aware of the issue.
I just remembered reading about the AES paper in the 80's and remembered that thermal stability of course was a problem. It was the ThermalTrack devices that triggered me.
Best regards,
Mogens
Please ignore message #88. I did a little searching this morning, on my way to work, and can see that this has been discussed in other threads and judging from the values of the emitter resistors I think you are aware of the issue.
I just remembered reading about the AES paper in the 80's and remembered that thermal stability of course was a problem. It was the ThermalTrack devices that triggered me.
Best regards,
Mogens
Last edited:
Hi Mogens,
Thx.
I really hope it's close to a "first cut" and I don't find another detail that might be improved.
Indeed, the need for a servo depends on the specifications of the op-amp. When using an AD797, for example, you will definitely need a servo. There are a few more things to make the amp complete: a PSU and offset monitor circuit (TA7317P + relay). What about to combine the latter together with the servo and Zobel coil on a separate PCB, as putting the whole amp on a single PCB makes the project (more or less) unmanageable?
Cheers,
E.
bias recovery after clipping
Hi Mogens,
As for voltages clipping, the bias is rock stable. As for current limiting, the bias is also unaffected, except above 10kHz. In that case the OPS gets a bit under-biased, depending on the values of C32 and C33 (btw, I decreased them to 22nF). Not a serious problem I would think. (If C32=C33=0 then the OPS becomes over-biased.)
Cheers,
E.
Hi Mogens,
As for voltages clipping, the bias is rock stable. As for current limiting, the bias is also unaffected, except above 10kHz. In that case the OPS gets a bit under-biased, depending on the values of C32 and C33 (btw, I decreased them to 22nF). Not a serious problem I would think.
(If C32=C33=0 then the OPS becomes over-biased.)Cheers,
E.
Hi Edmond,
That's great. It might be a good idea to make a separate board with the peripherals. I'm not familiar with the protection chip. I will have a look. Perhaps also a hf-detection (oscillation) circuit. I have included this in the past, in some of mine designs.
I have been speculating a little about the PCB form factor, but not come to any finale decision yet. I also want to be able to add a active crossover board, which would be nice if I could mount that on top of the amp pcb.
Best regards,
Mogens
That's great. It might be a good idea to make a separate board with the peripherals. I'm not familiar with the protection chip. I will have a look. Perhaps also a hf-detection (oscillation) circuit. I have included this in the past, in some of mine designs.
I have been speculating a little about the PCB form factor, but not come to any finale decision yet. I also want to be able to add a active crossover board, which would be nice if I could mount that on top of the amp pcb.
Best regards,
Mogens
Clipping
Hi Arthur,
I had issues with the TT-version, i.e. Phoenix13.
But I managed to circumvent the convergence problems by simplifying the circuit a bit:
using a constant bias voltage, fixed +/- 15v PSU and a simplified op-amp model.
It appeared that the phase margin of the voltage clamp was too low: only 35 degrees.
However, after a subtle modification of the frequency compensation, it's much, much better: 110 degrees! See 1st pic.
The only thing I don't like is an ugly phase transition at 25MHz, though I'm not sure if it's harmful anyhow.
BTW, the new frequency compensation is really new and exciting. So I like to investigate this one further (and find the cause of the ugly phase transition).
The 2nd pic shows the output at Vi = 2Vpk & f=20kHz. Impeccable for a NDFL amp, I would say.
Cheers,
E.
Hi Arthur,
I had issues with the TT-version, i.e. Phoenix13.
But I managed to circumvent the convergence problems by simplifying the circuit a bit:
using a constant bias voltage, fixed +/- 15v PSU and a simplified op-amp model.
It appeared that the phase margin of the voltage clamp was too low: only 35 degrees.
However, after a subtle modification of the frequency compensation, it's much, much better: 110 degrees!
See 1st pic.The only thing I don't like is an ugly phase transition at 25MHz, though I'm not sure if it's harmful anyhow.
BTW, the new frequency compensation is really new and exciting. So I like to investigate this one further (and find the cause of the ugly phase transition).
The 2nd pic shows the output at Vi = 2Vpk & f=20kHz. Impeccable for a NDFL amp, I would say.
Cheers,
E.
Attachments
Hello Edmond
>>BTW, the new frequency compensation is really new and exciting. So I like to investigate this one further (and find the cause of the ugly phase transition).
When you apply the new compensation scheme can you revert to more complex models for opamp and +/-15V supply and run the circuit without convergence issues.
By the way I am very interested in your new compensation scheme.
Regards
Arthur
>>BTW, the new frequency compensation is really new and exciting. So I like to investigate this one further (and find the cause of the ugly phase transition).
When you apply the new compensation scheme can you revert to more complex models for opamp and +/-15V supply and run the circuit without convergence issues.
By the way I am very interested in your new compensation scheme.
Regards
Arthur
SuperClamp
Hi Arthur,
I don't know yet. Perhaps I can avoid the convergence troubles by adding some 'catch diodes' or so.
So am I! In the next few days I will post more info about it.
It's about an adaptive frequency compensation that automatically adjusts (lowers) the loop BW when the clamp becomes active.
Cheers,
E.
Hi Arthur,
I don't know yet. Perhaps I can avoid the convergence troubles by adding some 'catch diodes' or so.
So am I! In the next few days I will post more info about it.
It's about an adaptive frequency compensation that automatically adjusts (lowers) the loop BW when the clamp becomes active.
Cheers,
E.
Convergence troubles
Hello Edmond
I loaded Phoenix13 into the simetrix simulator and the circuit works fine accept until the clamp is engaged, and you are right the clamp is the culprit and not stable ( but you already new this).
What subtel modifications did you make to get stable operation.
How are you going with the adaptive frequency compensation that automatically adjusts (lowers) the loop BW when the clamp becomes active.
Regards
Arthur
Hello Edmond
I loaded Phoenix13 into the simetrix simulator and the circuit works fine accept until the clamp is engaged, and you are right the clamp is the culprit and not stable ( but you already new this).
What subtel modifications did you make to get stable operation.
How are you going with the adaptive frequency compensation that automatically adjusts (lowers) the loop BW when the clamp becomes active.
Regards
Arthur
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.