Well I have been reading several sources of information, and now I have a few new questions! I hope someone can reply to these.
Regarding the Mosfet front-end, I am looking to improve its O/L performance.
Well, as I have been reading, making R1 twice R5 the HF distortion should decrease! Is this true with my circuit?
If I put a current mirror on the diff pair, should this make even more improvement?
I was reading that the VAS stage is current to voltage amplification. To me implying that the current drive from the VAS is very dependent on the diff pair for it's output current. Based on this should I increase the current through the diff pair?
Will increased current through the diff pair also increase it's O/L gain?
The author shall remain "Blameless"
Mr. Pass, Grey, and others, please comment!
Thanks, DonS
P.S. Maybe I shouldn't read so much!
Regarding the Mosfet front-end, I am looking to improve its O/L performance.
Well, as I have been reading, making R1 twice R5 the HF distortion should decrease! Is this true with my circuit?
If I put a current mirror on the diff pair, should this make even more improvement?
I was reading that the VAS stage is current to voltage amplification. To me implying that the current drive from the VAS is very dependent on the diff pair for it's output current. Based on this should I increase the current through the diff pair?
Will increased current through the diff pair also increase it's O/L gain?
The author shall remain "Blameless"
Mr. Pass, Grey, and others, please comment!
Thanks, DonS
P.S. Maybe I shouldn't read so much!
I confess that I'm not clear on what needs fixing at this point.
Referring to an earlier post about square wave performance--one that showed rounded leading corners on the waveform--I'd like to note that I prefer a 10kHz square wave that has a fairly sharp leading corners when my circuit is running open loop. That, in turn, leads me to look for open loop bandwidth of at least 200kHz. 100kHz bandwidth (I'm still talking open loop) will lead to fairly apparent rounding on the leading corners. All things considered, however, I'd rather see a little rounding than overshoot.
Asking me about negative feedback is a dangerous thing. I try to use as little as possible. To me, 20dB NFB is getting kinda high. Yes, I'm well aware that there are designs out there using far more, but my personal view is that I'd rather flog a circuit into good performance open loop, then use 10dB or less NFB; ideally 0dB, as in none, nada, zippo. Unfortunately, I'm not Nelson, John Curl, or Charles Hansen. I can't just sit down and scribble a design on a napkin that will do such a thing without me having to go back and fine tune the circuit. (And that's assuming that I can find time to work on it between changing diapers and feeding the young un's.)
"Speeding up" a circuit is a pretty broad concept. Ideas:
--Reduce the Gate stopper resistors (R9, R11, R15, R20) on the MOSFETs to 100 Ohms or so. If the circuit oscillates, go back and raise them, but no higher than necessary for stability.
--Use devices with less capacitance than the IRF610s (Q1, Q2).
--Use load resistances as low as you can for the front end differential (R1, R2). In passing, I see that you didn't take me up on my suggestion to use a pot for R1. Don't be surprised if you run into problems when you build your second channel, as the folded cascode MOSFET (Q6) used for the second channel is unlikely to have the same Vgs as this channel. This (potential) problem can be sidestepped if you match the Q6s for each channel to each other. Or to put it more particularly, you need to match Q6 against the values of R1 and R2.
--The VAS lag cap (C21) should be no larger than necessary for stability. A smaller value will give you wider open loop bandwidth. Ditto for C20.
--Reduce or eliminate C1. It's probably unnecessary. If it's there to keep the amplifier from oscillating, cure the oscillation rather than use a high pass filter at the input. Unless I mis-punched the numbers, you're rolling off just over 20kHz. Why? This will be measurably down as low as 5-10kHz.
Just for giggles, I'd suggest increasing the bias for the front end differential. You may need heatsinks for Q1, Q2, and Q3. You'll need to adjust the values of R1 & R2 to compensate. This is a win/win proposition, as the higher bias will make the MOSFETs behave better, and lower resistances in the loads will give you better performance overall.
There was something about buzzing earlier. I believe Nelson suggested looking at the rails. I agree with Nelson. Look at your rails on the scope with no signal. If you see ripple, you've got a problem, because that ripple will effect the circuit. A number of things can cause ripple. The most likely are too little filtering in the power supply or too small a power transformer. The cure will depend on the cause, but it never hurts to put in more capacitance. Well...other than your wallet, anyway.
Grey
EDIT: Please note that we're volunteers, not employees. Sometimes you just need to work through a few of these things on your own.
Referring to an earlier post about square wave performance--one that showed rounded leading corners on the waveform--I'd like to note that I prefer a 10kHz square wave that has a fairly sharp leading corners when my circuit is running open loop. That, in turn, leads me to look for open loop bandwidth of at least 200kHz. 100kHz bandwidth (I'm still talking open loop) will lead to fairly apparent rounding on the leading corners. All things considered, however, I'd rather see a little rounding than overshoot.
Asking me about negative feedback is a dangerous thing. I try to use as little as possible. To me, 20dB NFB is getting kinda high. Yes, I'm well aware that there are designs out there using far more, but my personal view is that I'd rather flog a circuit into good performance open loop, then use 10dB or less NFB; ideally 0dB, as in none, nada, zippo. Unfortunately, I'm not Nelson, John Curl, or Charles Hansen. I can't just sit down and scribble a design on a napkin that will do such a thing without me having to go back and fine tune the circuit. (And that's assuming that I can find time to work on it between changing diapers and feeding the young un's.)
"Speeding up" a circuit is a pretty broad concept. Ideas:
--Reduce the Gate stopper resistors (R9, R11, R15, R20) on the MOSFETs to 100 Ohms or so. If the circuit oscillates, go back and raise them, but no higher than necessary for stability.
--Use devices with less capacitance than the IRF610s (Q1, Q2).
--Use load resistances as low as you can for the front end differential (R1, R2). In passing, I see that you didn't take me up on my suggestion to use a pot for R1. Don't be surprised if you run into problems when you build your second channel, as the folded cascode MOSFET (Q6) used for the second channel is unlikely to have the same Vgs as this channel. This (potential) problem can be sidestepped if you match the Q6s for each channel to each other. Or to put it more particularly, you need to match Q6 against the values of R1 and R2.
--The VAS lag cap (C21) should be no larger than necessary for stability. A smaller value will give you wider open loop bandwidth. Ditto for C20.
--Reduce or eliminate C1. It's probably unnecessary. If it's there to keep the amplifier from oscillating, cure the oscillation rather than use a high pass filter at the input. Unless I mis-punched the numbers, you're rolling off just over 20kHz. Why? This will be measurably down as low as 5-10kHz.
Just for giggles, I'd suggest increasing the bias for the front end differential. You may need heatsinks for Q1, Q2, and Q3. You'll need to adjust the values of R1 & R2 to compensate. This is a win/win proposition, as the higher bias will make the MOSFETs behave better, and lower resistances in the loads will give you better performance overall.
There was something about buzzing earlier. I believe Nelson suggested looking at the rails. I agree with Nelson. Look at your rails on the scope with no signal. If you see ripple, you've got a problem, because that ripple will effect the circuit. A number of things can cause ripple. The most likely are too little filtering in the power supply or too small a power transformer. The cure will depend on the cause, but it never hurts to put in more capacitance. Well...other than your wallet, anyway.
Grey
EDIT: Please note that we're volunteers, not employees. Sometimes you just need to work through a few of these things on your own.
Grey, I have taken so long to reply because of your edit in the last post. I don’t know how to respond. I am sorry you felt that I was ordering you to do something. I certainly didn't mean it that way!
I respect your input very much. You and Mr. Pass have been a great help on my journey.
I was simply seeking help from the man that authored the MiniA project, as my circuit is very similar.
Thank you, for all your help!!!
DonS
I respect your input very much. You and Mr. Pass have been a great help on my journey.
I was simply seeking help from the man that authored the MiniA project, as my circuit is very similar.
Thank you, for all your help!!!
DonS
I'm not sure I see a resemblance between your circuit and the Mini-A, but then I'm short on sleep and...worse yet...only have three brain cells. And two of them didn't show up for work today.
If you're reading Self/Slone, then you're in an entirely different universe from me. I go at things quite differently from those guys. They like class B, I like class A (at least for solid state, I can get decent sound from tubes with class AB, but that's another story [and, yes, I'm aware they've got class A circuits, but that's tucked away in a corner--almost as though they're ashamed of them]). I like low-to-no feedback, they're content with as near to infinite feedback as they can get. Etc. Differences abound. It's not so much that everything they do is incompatible with my way of looking at things...well, okay, maybe it is. You can build an amplifier their way and it will work and music will come out and as far as I know it will be reasonably reliable, but it's kinda like the difference between an Army drill sergeant and a hippy. Those guys want to micromanage every single electron (Sir, yes, Sir!) and I want the poor little things to be free and live fulfilling, happy lives, even if it means moving to Mongolia and herding yaks. Raise them up the right way, and they'll make the right decisions on their own when the time comes. I need not interfere.
(If you get the impression that I'm a square peg in a round hole here in SC, you're right, but this is neither the time nor the place for that discussion.)
As for helping you along your journey, remember that neither Obi-Wan nor Yoda were with Luke the whole way. He had to face Vader alone.
Grey
If you're reading Self/Slone, then you're in an entirely different universe from me. I go at things quite differently from those guys. They like class B, I like class A (at least for solid state, I can get decent sound from tubes with class AB, but that's another story [and, yes, I'm aware they've got class A circuits, but that's tucked away in a corner--almost as though they're ashamed of them]). I like low-to-no feedback, they're content with as near to infinite feedback as they can get. Etc. Differences abound. It's not so much that everything they do is incompatible with my way of looking at things...well, okay, maybe it is. You can build an amplifier their way and it will work and music will come out and as far as I know it will be reasonably reliable, but it's kinda like the difference between an Army drill sergeant and a hippy. Those guys want to micromanage every single electron (Sir, yes, Sir!) and I want the poor little things to be free and live fulfilling, happy lives, even if it means moving to Mongolia and herding yaks. Raise them up the right way, and they'll make the right decisions on their own when the time comes. I need not interfere.
(If you get the impression that I'm a square peg in a round hole here in SC, you're right, but this is neither the time nor the place for that discussion.)
As for helping you along your journey, remember that neither Obi-Wan nor Yoda were with Luke the whole way. He had to face Vader alone.
Grey
First! Thank all of you for helping! It has been quite a journey. This is the first time I have designed an amp myself. Well with a lot of help from this forum, and Mr. Pass, Grey, and JJ especially!
First it was just a front-end, and then it grew to a full amplifier. I had to make a lot of compromises because of the old amp chassis and power supply. However the old beast plays music again and that is what I was after! I have been happily listening to music for over a week now with it. No problems to report.
I wish I could post specs but I do not have the equipment. However, ears are the final judges, and to my ears this is not a bad class a/b amp. It is scalable and would make a good class a amp I think, but not at the supply voltages spec’s. (Too impractical.)
It seems that this thread has come to an end. If someone would like to use this circuit, or has suggestions for improvement, let me know. If not this is my last post for this thread.
Thanks all, DonS
First it was just a front-end, and then it grew to a full amplifier. I had to make a lot of compromises because of the old amp chassis and power supply. However the old beast plays music again and that is what I was after! I have been happily listening to music for over a week now with it. No problems to report.
I wish I could post specs but I do not have the equipment. However, ears are the final judges, and to my ears this is not a bad class a/b amp. It is scalable and would make a good class a amp I think, but not at the supply voltages spec’s. (Too impractical.)
It seems that this thread has come to an end. If someone would like to use this circuit, or has suggestions for improvement, let me know. If not this is my last post for this thread.
Thanks all, DonS
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.