Thanks to moderators for transferring the posts from another thread.
Sander, we can now continue in the discussion of time domain behavior of the ExtremA and its step response. I would greatly appreciate if Bruno could join us as well.
My post with measurements is here:
http://www.diyaudio.com/forums/solid-state/96853-extrema-class-strikes-back-28.html#post4118436
Sander, we can now continue in the discussion of time domain behavior of the ExtremA and its step response. I would greatly appreciate if Bruno could join us as well.
My post with measurements is here:
http://www.diyaudio.com/forums/solid-state/96853-extrema-class-strikes-back-28.html#post4118436
There is no RC filter in the original Sander/Bruno design. It is so slow, that RC time constant would probably affect audio band, if it should prevent the circuit from slewing.
It is not about music only, as Bruno had said. It is also about glitches, D/A residuals, EMI interference. The input SR may be easily exceeded and there is nothing to prevent it.
It is not about music only, as Bruno had said. It is also about glitches, D/A residuals, EMI interference. The input SR may be easily exceeded and there is nothing to prevent it.
Member
Joined 2009
Paid Member
Seems like the greater the current output capacity of the amp, the better it will show on the test under the loads. So, what would be those minimum power output parameters needed to show up as a good amplifier with the AudioGraph loadbox?
THx-RNMarsh
Last edited:
Current capacity increases with load. Speaker + cable are not a pure real resistive load (in general) so you can't set reserve current demand assuming it is. If some have found that in practice this power cube concept is useless fair enough. I remember Bob Cordell discussing a session observing clipping behavior with normal listening and an amp that should have been adequate, or am I remembering wrong? BTW I'm certain I would be happy I don't like it loud or with bass.
Volts / Resistance = Current
• 20 volts / 8 ohms = 2.5 amps
• 20 volts / 4 ohms = 5 amps
• 20 volts / 2 ohms = 10 amps (assuming the amplifier can provide 10 amperes)
Power = Voltage x Current. So the three loads would receive this amount of power:
• 8-ohm load: 20 volts x 2.5 amps = 50 watts.
• 4-ohm load: 20 volts x 5.0 amps = 100 watts.
• 2-ohm load: 20 volts x 10.0 amps = 200 watts.
As you can see, the amplifier power doubles every time the load impedance is halved. That’s at low levels, where the power supply is not forced to produce much current.
Last edited:
First and last statement regarding Extrema, and my involvement.
I might clarify my silence. Mr Sassen and I have something of a personal tiff since the Extrema was originally published and of which after all these years he still hasn't fathomed the cause. It is fairly succinctly analysed in this post (from 2009, #22):
http://www.diyaudio.com/forums/solid-state/155181-my-extrema-built-whew-3.html#post1999925
Other than that I would like to state in public what I've so far only said in private. The Extrema was a summer geek-out project, a stunt if you will, for the purpose of publication in an Elektor special edition. Design-wise the Extrema is not my proudest moment. It shows all the oversights you would expect from a new design that was only ever started, never finished. It's also woefully complicated for the performance. These days, if called to design a non-class D amp I would make it a moderate bandwidth optimally biased class AB affair with an Ncore style higher order loop round it. Much simpler, much better performance too. I would never recommend anyone looking to quickly DIY a great amp to go and build an Extrema.
I drafted a schematic. Mr Sassen drafted boards which I reviewed and after bouncing them back and forth a bit, Mr Sassen built up a prototype. Then we had a (rather quick) measurement and debug session and Mr Sassen respun the boards and built the second prototype. He then published the design in Elektor. Authors: "Sander Sassen and Bruno Putzeys". In that order. I felt this to be a bit sassy for someone who has only the sketchiest insight in how the circuit works (or alternatively, doesn't). Otherwise he'd have been perfectly able to serve up cogent and technically meaningful answers to various reasonable queries about the amp's idiosyncrasies.
After the article was published he built the second channel but managed to sell the thing before I could even hear it. I did get a set of his boards and output transistors so that I could build my own if I wanted. Suffice to say I have never heard the Extrema. Ever. Mr Sassen then set up a small business selling boards to DIY for which I haven't seen a penny (just today he wrote to explain that this was merely in compensation for his efforts) and as the post I linked above attests, paid little effort to attribute the design. He only did that whenever he was confronted with annoying questions. I have naïvely allowed myself to be used to help Mr. Sassen shine the spotlight on himself. Simply put, he got the glory, I got the dreck. You will understand that I felt, and still feel, little compunction letting him tread water ever since. I sympathise deeply with people who are, as a result, still struggling with the many issues about Extrema that were never fixed. For future takers: if you are planning to build an Extrema, don't.
---
Responding to the question whether an amplifier should happily handle large supersonic signals in order to classify as a good audio amplifier: from a practical perspective often yes but fundamentally no. In the classical Miller-style design supersonic performance correlates reasonably well with sonic performance, for reasons I explained in "the F word"* aka "why there is no such thing as too much feedback". But even then the simple case of degeneration shows that slew rate in itself is a poor predictor of performance. How well it handles realistic (audio-bandwidth) slew rates is the only thing that matters. The only extension to that would be some allowance for DSD type noise, but the astonishing amount of outband noise produced by some ideologically inspired DACs (unfiltered R2R for example) is a design problem for the DAC. If it causes trouble, the DAC's where the issue ought to be fixed.
This is only tangentially related to the question whether it's a good sign for an amp to hunt around before suddenly settling of course. It's not, clearly. This kind of behaviour happens easily with higher order control loops and is known as "integrator wind-up". It's an occasionally pesky problem but it's always fixable, at least in something as straightforward as an audio amp. Doing so on the Extrema is left as an exercise for Mr. Sassen.
I hope you will accept that, as explained above, I'll not respond further to this thread or any thread about the Extrema project.
_____________
*Please check "why there's no such thing as too much feedback" on EDN or support Jan Didden's Linear Audio by buying the 1st edition in which it was originally published. It may well answer your questions regarding slew rate, THD and the 1970's Great Numbers War.
I might clarify my silence. Mr Sassen and I have something of a personal tiff since the Extrema was originally published and of which after all these years he still hasn't fathomed the cause. It is fairly succinctly analysed in this post (from 2009, #22):
http://www.diyaudio.com/forums/solid-state/155181-my-extrema-built-whew-3.html#post1999925
Other than that I would like to state in public what I've so far only said in private. The Extrema was a summer geek-out project, a stunt if you will, for the purpose of publication in an Elektor special edition. Design-wise the Extrema is not my proudest moment. It shows all the oversights you would expect from a new design that was only ever started, never finished. It's also woefully complicated for the performance. These days, if called to design a non-class D amp I would make it a moderate bandwidth optimally biased class AB affair with an Ncore style higher order loop round it. Much simpler, much better performance too. I would never recommend anyone looking to quickly DIY a great amp to go and build an Extrema.
I drafted a schematic. Mr Sassen drafted boards which I reviewed and after bouncing them back and forth a bit, Mr Sassen built up a prototype. Then we had a (rather quick) measurement and debug session and Mr Sassen respun the boards and built the second prototype. He then published the design in Elektor. Authors: "Sander Sassen and Bruno Putzeys". In that order. I felt this to be a bit sassy for someone who has only the sketchiest insight in how the circuit works (or alternatively, doesn't). Otherwise he'd have been perfectly able to serve up cogent and technically meaningful answers to various reasonable queries about the amp's idiosyncrasies.
After the article was published he built the second channel but managed to sell the thing before I could even hear it. I did get a set of his boards and output transistors so that I could build my own if I wanted. Suffice to say I have never heard the Extrema. Ever. Mr Sassen then set up a small business selling boards to DIY for which I haven't seen a penny (just today he wrote to explain that this was merely in compensation for his efforts) and as the post I linked above attests, paid little effort to attribute the design. He only did that whenever he was confronted with annoying questions. I have naïvely allowed myself to be used to help Mr. Sassen shine the spotlight on himself. Simply put, he got the glory, I got the dreck. You will understand that I felt, and still feel, little compunction letting him tread water ever since. I sympathise deeply with people who are, as a result, still struggling with the many issues about Extrema that were never fixed. For future takers: if you are planning to build an Extrema, don't.
---
Responding to the question whether an amplifier should happily handle large supersonic signals in order to classify as a good audio amplifier: from a practical perspective often yes but fundamentally no. In the classical Miller-style design supersonic performance correlates reasonably well with sonic performance, for reasons I explained in "the F word"* aka "why there is no such thing as too much feedback". But even then the simple case of degeneration shows that slew rate in itself is a poor predictor of performance. How well it handles realistic (audio-bandwidth) slew rates is the only thing that matters. The only extension to that would be some allowance for DSD type noise, but the astonishing amount of outband noise produced by some ideologically inspired DACs (unfiltered R2R for example) is a design problem for the DAC. If it causes trouble, the DAC's where the issue ought to be fixed.
This is only tangentially related to the question whether it's a good sign for an amp to hunt around before suddenly settling of course. It's not, clearly. This kind of behaviour happens easily with higher order control loops and is known as "integrator wind-up". It's an occasionally pesky problem but it's always fixable, at least in something as straightforward as an audio amp. Doing so on the Extrema is left as an exercise for Mr. Sassen.
I hope you will accept that, as explained above, I'll not respond further to this thread or any thread about the Extrema project.
_____________
*Please check "why there's no such thing as too much feedback" on EDN or support Jan Didden's Linear Audio by buying the 1st edition in which it was originally published. It may well answer your questions regarding slew rate, THD and the 1970's Great Numbers War.
Last edited:
Thank you for your point of view. I understand completely that you will not respond further to this thread. On the other hand, the circuit was released and sold to public.
In fact I have no questions regarding slew rate, I am pretty familiar with it and with relations between time and frequency domains. Based on that, based on my education and based on my professional background I state that the time domain behavior of this amplifier is insufficient and unfortunately degrades the potential of the amp defined by its very low THD when the circuit remains in class A. The output stage is excellent in this regard, it is also excellent in time domain and it would deserve well engineered front-end, which is not the case here.
Oh my, dear Bruno, talk about showing your true colors ...
We had a brief exchange of emails today where you threatened to vent your frustrations on a public forum when I politely asked you whether you wanted to give your views on an issue presented by Pavel (PMA).
I guess you didn't have the decency to discuss any frustrations in private, as I suggested earlier today. Despite the fact that I picked up on, from the tone of your email, that something was bothering you. I'm sorry to see you opted to discuss this on a public forum as you leave me no other choice than to respond to your post (something I'm sure you must've understood whilst posting your reply).
Things, at least in my opinion, transpired a little differently than how Bruno has chosen to portray them here and despite having no real desire to air any dirty laundry, allow me to give the readers my perspective.
The ExtremA began as a cool idea Bruno and I cooked up on a late evening Skype session somewhere in 2006. I mentioned it would be nice to have a really high-performance DIY amplifier in the next Elektor Audio Special (for which I was the editor in chief) and Bruno mentioned he had some design ideas he'd like to try out.
Bruno sent me some drafts, I subsequently worked those into working simulations in LTspice, and we were off to something, In a matter of weeks I ordered parts for and built the first, relatively simple, prototype and drove out all the way to Belgium to evaluate the design at Bruno's lab at home. We refined a few things and I set to work on incorporating these changes into a 2nd round of prototypes, for which I purchased another set of parts, and spent a considerable amount of time constructing and testing.
The cycle repeated, with me driving out to Hypex in Groningen to do another round of testing. This 2nd prototype performed a lot better, pretty close to where we were aiming for actually, so this design was drafted as a PCB by yours truly. I ordered PCBs and the 3rd set of prototypes was ready for scrutiny a few weeks later. I again drove out to Hypex in Groningen to have the prototype measured by Bruno.
This design is what we finally settled on as it did everything we set out to do in terms of measured distortion and output power.
The rest is pretty much in the public domain already. The article got published in the Elektor Audio Special with the authors indeed being listed as 'Sander Sassen and Bruno Putzeys', the order of which was never discussed and in my mind not of any importance. In hindsight I would've been perfectly happy with them being in reverse order if that is of such importance to Bruno.
I subsequently ordered a small batch of the final PCBs, so I could sell these, basically at cost, to interested DIY-ers and sent a set of PCBs and matched output transistors to Bruno at my expense. The set of ExtremA amplifiers I built for myself I sold in the week after the Elektor Audio Special appeared in local kiosks. There's nothing weird about that frankly, I had an acquaintance come over for a listening session and he offered me a four figure sum of money for the set, an offer I couldn't refuse as at the time I was having a difficult time making ends meet.
I've never set up 'a small business selling boards' as Bruno puts it, and I have offered, from the get go, to share in any revenues originating from that.
Frankly, in the brief exchange of emails we've had today, I have once more offered to send Bruno 1/2 of the revenues for all PCBs sold to date if that's what's bothering him, I have yet to get a reply from him on that offer. In all honesty there are less than 100 sets sold the world over, so we're not talking about large sums of money here.
Take note that all of the costs associated with the prototypes, PCBs, transformers, heatsinks, etc. etc. have all been covered by me, Bruno has never offered, nor bothered, to pay a penny. I thus felt justified in thinking that any revenues from selling PCBs would go towards reimbursing some of those costs first. We never discussed this, in hindsight perhaps we should've, as apparently it is something that bothers Bruno.
What surprises me is that after publication Bruno has never put much effort towards supporting the design, other than a handful of posts and emails, to aid those that are building the ExtremA. I've spent the last 7 years supporting the design and those aspiring to build it. If he indeed felt he was helping to shine the spotlight on me why didn't he say so earlier?
I also find it disappointing to see that initially Bruno himself was also advocating the ExtremA as being an exemplary amplifier, and now suggests people to stay away, as the design has issues that were never fixed. Isn't addressing these issues something Bruno needs to tackle, as he's so keen on putting the spotlight on himself as the designer and states I lack the competence to do that? If the design is lacking in some aspects, isn't that partly his responsibility?
I would like to end this longish reply with an apology; I typically do not vent in public, certainly not about something which to this day I am proud to have been a part of. I sincerely wish Bruno would have discussed these frustrations with me earlier, rather than wait 7 years to vent them on a public forum.
Bruno, You know I love you for the genius you are, although your personality needs a little getting used to, but so does mine, or so I am told. Despite that fact I always had the distinct impression we got along pretty well, so I am truly sorry to see us part ways in this manner.
I wish you the best of luck and success in your endeavors.
Respectfully,
Sander Sassen.
We had a brief exchange of emails today where you threatened to vent your frustrations on a public forum when I politely asked you whether you wanted to give your views on an issue presented by Pavel (PMA).
I guess you didn't have the decency to discuss any frustrations in private, as I suggested earlier today. Despite the fact that I picked up on, from the tone of your email, that something was bothering you. I'm sorry to see you opted to discuss this on a public forum as you leave me no other choice than to respond to your post (something I'm sure you must've understood whilst posting your reply).
Things, at least in my opinion, transpired a little differently than how Bruno has chosen to portray them here and despite having no real desire to air any dirty laundry, allow me to give the readers my perspective.
The ExtremA began as a cool idea Bruno and I cooked up on a late evening Skype session somewhere in 2006. I mentioned it would be nice to have a really high-performance DIY amplifier in the next Elektor Audio Special (for which I was the editor in chief) and Bruno mentioned he had some design ideas he'd like to try out.
Bruno sent me some drafts, I subsequently worked those into working simulations in LTspice, and we were off to something, In a matter of weeks I ordered parts for and built the first, relatively simple, prototype and drove out all the way to Belgium to evaluate the design at Bruno's lab at home. We refined a few things and I set to work on incorporating these changes into a 2nd round of prototypes, for which I purchased another set of parts, and spent a considerable amount of time constructing and testing.
The cycle repeated, with me driving out to Hypex in Groningen to do another round of testing. This 2nd prototype performed a lot better, pretty close to where we were aiming for actually, so this design was drafted as a PCB by yours truly. I ordered PCBs and the 3rd set of prototypes was ready for scrutiny a few weeks later. I again drove out to Hypex in Groningen to have the prototype measured by Bruno.
This design is what we finally settled on as it did everything we set out to do in terms of measured distortion and output power.
The rest is pretty much in the public domain already. The article got published in the Elektor Audio Special with the authors indeed being listed as 'Sander Sassen and Bruno Putzeys', the order of which was never discussed and in my mind not of any importance. In hindsight I would've been perfectly happy with them being in reverse order if that is of such importance to Bruno.
I subsequently ordered a small batch of the final PCBs, so I could sell these, basically at cost, to interested DIY-ers and sent a set of PCBs and matched output transistors to Bruno at my expense. The set of ExtremA amplifiers I built for myself I sold in the week after the Elektor Audio Special appeared in local kiosks. There's nothing weird about that frankly, I had an acquaintance come over for a listening session and he offered me a four figure sum of money for the set, an offer I couldn't refuse as at the time I was having a difficult time making ends meet.
I've never set up 'a small business selling boards' as Bruno puts it, and I have offered, from the get go, to share in any revenues originating from that.
Frankly, in the brief exchange of emails we've had today, I have once more offered to send Bruno 1/2 of the revenues for all PCBs sold to date if that's what's bothering him, I have yet to get a reply from him on that offer. In all honesty there are less than 100 sets sold the world over, so we're not talking about large sums of money here.
Take note that all of the costs associated with the prototypes, PCBs, transformers, heatsinks, etc. etc. have all been covered by me, Bruno has never offered, nor bothered, to pay a penny. I thus felt justified in thinking that any revenues from selling PCBs would go towards reimbursing some of those costs first. We never discussed this, in hindsight perhaps we should've, as apparently it is something that bothers Bruno.
What surprises me is that after publication Bruno has never put much effort towards supporting the design, other than a handful of posts and emails, to aid those that are building the ExtremA. I've spent the last 7 years supporting the design and those aspiring to build it. If he indeed felt he was helping to shine the spotlight on me why didn't he say so earlier?
I also find it disappointing to see that initially Bruno himself was also advocating the ExtremA as being an exemplary amplifier, and now suggests people to stay away, as the design has issues that were never fixed. Isn't addressing these issues something Bruno needs to tackle, as he's so keen on putting the spotlight on himself as the designer and states I lack the competence to do that? If the design is lacking in some aspects, isn't that partly his responsibility?
I would like to end this longish reply with an apology; I typically do not vent in public, certainly not about something which to this day I am proud to have been a part of. I sincerely wish Bruno would have discussed these frustrations with me earlier, rather than wait 7 years to vent them on a public forum.
Bruno, You know I love you for the genius you are, although your personality needs a little getting used to, but so does mine, or so I am told. Despite that fact I always had the distinct impression we got along pretty well, so I am truly sorry to see us part ways in this manner.
I wish you the best of luck and success in your endeavors.
Respectfully,
Sander Sassen.
Last edited:
wow ...
sorry things were sour.
that's too bad; i was looking forward to a good expert level exchange on this topic and expecting some good learnings to come out of it.
mlloyd1
sorry things were sour.
that's too bad; i was looking forward to a good expert level exchange on this topic and expecting some good learnings to come out of it.
mlloyd1
Guys, stick to technical discussion.
I'd like to take a stab at it, but I'm a bit rusty as I don't design amplifiers on a daily basis, nor have the wealth of experience Bruno has, but alas, lets have a look. It needs no further explanation that the step response isn't particularly clean.
One thing that's clear from the simulations is that the slewrate limitation as reported by Pavel (and verified in simulation as well as in the real world) is mostly due to the two 10R/1nF RC networks that are at the input of the two output stages.
They are R40/C12 and R41/C13 in the schematic above. They are driven from the predrivers which is a Sziklai transistor configuration, biased to ~4mA. These are Q38/Q39 and Q37/Q40 in the schematic below, their output is PRE+ and PRE-.
Biasing these higher will result in more drive capability, and hence better slewrate, the downside is that dissipation will also rise. Changing R34/R35 to 470R rather than 680R raises the bias current to 10mA. Simulation shows that this doesn't solve the problem though.
We can obviously lower the value of C12/C13 in the RC networks, but doing that to a large extent makes the amplifier become unstable, so we'll also need to compensate elsewhere. Dropping C12/C13 to 220p shows improvement, and no unstability, but there's still overshoot and subtle ringing on the output, dropping to 100pF makes the amplifier unstable. So lets keep these at 220pF for now.
There's two other RC networks though, R38/C10 and R39/C11, if C10/C11 are made larger in value that should straighten things out again. Doubling their value from 33pF to 68pF indeed does the trick. Below is the amplifier with the above change and the step response stimulus (PULSE(-1 1 0 100n 100n 50u 100u). The THD values are slightly better in the simulator than the original with these changes.
I unfortunately do not have an ExtremA here I can test this on, so I'm hoping Pavel would be able to try this out in practice. It requires swapping out a total of four capacitors.
If anyone would like to offer different suggestions I'd be happy to hear them.
Sander, I greatly appreciate that you have returned to the circuit. Regarding what I can do and I have have already done with the ExtremA, please let me summarize.
I have built another amplifier with your, slightly modified, output stage. It is a SE version with opamp frontend, please see the schematics:
This circuit has very low THD, that confirmed YOUR THD measurements (I have shown the measurements earlier in this thread). This circuit also has smooth, RC – like step response, that is not slew rate limited, shown again:
--------------------------------------
Then, about 3 weeks ago, I was asked by one of the builders here, to help with measurements of the original, i.e. yours and Bruno's design. That's how I found the problem with step response, which I have also simulated with very similar result as the measured one.
I have tried to reduce C4 and C9 (original names per your webpage, now C12 and C13 in the post above) capacitors from 1nF to 100pF. The result looked promising, step response was clean (though slew rate limited), but – THD rose to some 0.2%. Though oscilloscope has not shown any oscillations, the circuit must have been oscillating at least locally. So I have put back the original 1nF capacitors back. I do not have your original amplifier design yet, it is back at the owner, which is about 300km from my place.
Later I have been playing with simulations as well and came to similar conclusion as you have made. The C4 and C9 should be no smaller than 220pF. The modified value of two further capacitors should help as well, the last thing to do is to verify it in the real amplifier, I hope the guys who have the amplifier now might try it.
Thanks again for your co-operation, I believe we are on the right track now.
I have built another amplifier with your, slightly modified, output stage. It is a SE version with opamp frontend, please see the schematics:
An externally hosted image should be here but it was not working when we last tested it.
This circuit has very low THD, that confirmed YOUR THD measurements (I have shown the measurements earlier in this thread). This circuit also has smooth, RC – like step response, that is not slew rate limited, shown again:
An externally hosted image should be here but it was not working when we last tested it.
--------------------------------------
Then, about 3 weeks ago, I was asked by one of the builders here, to help with measurements of the original, i.e. yours and Bruno's design. That's how I found the problem with step response, which I have also simulated with very similar result as the measured one.
I have tried to reduce C4 and C9 (original names per your webpage, now C12 and C13 in the post above) capacitors from 1nF to 100pF. The result looked promising, step response was clean (though slew rate limited), but – THD rose to some 0.2%. Though oscilloscope has not shown any oscillations, the circuit must have been oscillating at least locally. So I have put back the original 1nF capacitors back. I do not have your original amplifier design yet, it is back at the owner, which is about 300km from my place.
Later I have been playing with simulations as well and came to similar conclusion as you have made. The C4 and C9 should be no smaller than 220pF. The modified value of two further capacitors should help as well, the last thing to do is to verify it in the real amplifier, I hope the guys who have the amplifier now might try it.
Thanks again for your co-operation, I believe we are on the right track now.
Last edited:
Intersting OPS.
In the sx-Amp, I use a single transistor to regulate the OPS current (700mA per pair). I used this same (single transistor) technique for the buffer stage in my Symphony preamp to get ppm level class A performance in to 32 Ohms. Later, I used a two transistor regulator with current source load very similar to the design above - I posted up a circuit in the Headphone forum.
It of course does not handle the power that the ExtremA does, but the distortion in class A is in low ppb (simmed) - around -130 dB ref 1V out, and still -120 dB at ~3 V out. So the basic OPS concept can deliver outstanding class A performance.
In the sx-Amp, I use a single transistor to regulate the OPS current (700mA per pair). I used this same (single transistor) technique for the buffer stage in my Symphony preamp to get ppm level class A performance in to 32 Ohms. Later, I used a two transistor regulator with current source load very similar to the design above - I posted up a circuit in the Headphone forum.
It of course does not handle the power that the ExtremA does, but the distortion in class A is in low ppb (simmed) - around -130 dB ref 1V out, and still -120 dB at ~3 V out. So the basic OPS concept can deliver outstanding class A performance.
@Pavel,
I like what you've done here, I've always wondered how the output stage would fare with an opamp frontend, never had time to take a closer look though. This, with a bit more work, can easily be turned into a bridged configuration with balanced input. That will help with overall efficiency, which, even on the ExtremA, isn't great courtesy of the class-A OPS, we're at about 40% with the ExtremA.
I would certainly be interested in exploring this further, the design of a poor-man's ExtremA would be a possibility. One of the complaints I've always heard about the current design is that there's too many parts and it is too complex.
One approach could be to take a (balanced) instrumentation amplifier (dual opamp) as an input stage (with all the required EMI/RFI measures) then subsequently two of the amplifier stages with an opamp frontend and round this off by tying a DC servo around the whole amplifier.
Anyway, I'm getting ahead of myself here. I'll see if I can find someone with an ExtremA locally so the proposed changes can be worked in and verified. Thanks so much for your help and suggestions thusfar.
I like what you've done here, I've always wondered how the output stage would fare with an opamp frontend, never had time to take a closer look though. This, with a bit more work, can easily be turned into a bridged configuration with balanced input. That will help with overall efficiency, which, even on the ExtremA, isn't great courtesy of the class-A OPS, we're at about 40% with the ExtremA.
I would certainly be interested in exploring this further, the design of a poor-man's ExtremA would be a possibility. One of the complaints I've always heard about the current design is that there's too many parts and it is too complex.
One approach could be to take a (balanced) instrumentation amplifier (dual opamp) as an input stage (with all the required EMI/RFI measures) then subsequently two of the amplifier stages with an opamp frontend and round this off by tying a DC servo around the whole amplifier.
Anyway, I'm getting ahead of myself here. I'll see if I can find someone with an ExtremA locally so the proposed changes can be worked in and verified. Thanks so much for your help and suggestions thusfar.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.