Hi All,
For some time now an acquaintance of mine has been using a single-ended class A amplifier - with much satisfaction as I understand him - although this amplifier topology has a rather steep distortion increase when playing loud(er). With this in mind he has asked me if something could be done to amend this high distortion level while maintaining the sound qualities that he finds this amplifier topology to have.
FYI I have attached a schematic of the amplifier circuitry as it overall looks today.
I have been considering something like boot strapping or a Sziklai output stage (right side schematic drawing), however, I wonder if there could be some kind of "error correction topology" (i.e. not global feedback) that would work for single-ended circuitries (and still sound excellent)?
In my thinking the main issue is the output stage where the distortion gets quite high even at moderate playback levels (please note here that I am aware that CCS'es could be substituted for the resistors R1, R2, R9 and thus improve distortion figures. In this context I would hope to find some "single ended error correction topology").
Might one/some of you have some insight into this that you could share?
Cheers & thanks for your inputs,
Jesper
For some time now an acquaintance of mine has been using a single-ended class A amplifier - with much satisfaction as I understand him - although this amplifier topology has a rather steep distortion increase when playing loud(er). With this in mind he has asked me if something could be done to amend this high distortion level while maintaining the sound qualities that he finds this amplifier topology to have.
FYI I have attached a schematic of the amplifier circuitry as it overall looks today.
I have been considering something like boot strapping or a Sziklai output stage (right side schematic drawing), however, I wonder if there could be some kind of "error correction topology" (i.e. not global feedback) that would work for single-ended circuitries (and still sound excellent)?
In my thinking the main issue is the output stage where the distortion gets quite high even at moderate playback levels (please note here that I am aware that CCS'es could be substituted for the resistors R1, R2, R9 and thus improve distortion figures. In this context I would hope to find some "single ended error correction topology").
Might one/some of you have some insight into this that you could share?
Cheers & thanks for your inputs,
Jesper
Attachments
Most likely, the 'sound" you enjoy is the "triode" distortion that happens in simple amplifiers with no feedback, and it comes part and parcel with power limitations. Note that the circuit on the left is also unlikely to center the bias without some fiddling, again producing even harmonic distortion but very limited output. A single ended class A amp is the worst of the worst for efficiency, and a push-pull class-A amp would be twice as efficient, actually more given that an active output could give twice as much output voltage (+6dB) instead of splitting the supply between the speaker and the pull-down resistor.
It is not insignificant that "single ended" amplifiers are popular with novices because they are difficult to destroy. An efficient amp is also one that can "blow up" in an instant. Building a "real amplifier" requires a lot of caution and patience, particularly with start-up testing.
The human hearing system is logarithmic, which means:
1. The power required for impress is unlimited and easily destroys speakers.
2. A useful sound level can be very modest, only a few milliwatts.
I suggest you focus on making something that works and not on any audiophile notion. You can do that later after you have lots of experience.
It is not insignificant that "single ended" amplifiers are popular with novices because they are difficult to destroy. An efficient amp is also one that can "blow up" in an instant. Building a "real amplifier" requires a lot of caution and patience, particularly with start-up testing.
The human hearing system is logarithmic, which means:
1. The power required for impress is unlimited and easily destroys speakers.
2. A useful sound level can be very modest, only a few milliwatts.
I suggest you focus on making something that works and not on any audiophile notion. You can do that later after you have lots of experience.
The distortion when playing louder is clipping on the negative part of the waveform. The output needs more bias, and then it will run up against low efficiency.
The circuit is a poor power amplifier. Build one of the amplifiers on this site. They are all much better.
Ed
The circuit is a poor power amplifier. Build one of the amplifiers on this site. They are all much better.
Ed
A Taylor or White scheme could improve linearity. You can go further and use this scheme: https://www.diyaudio.com/community/...nd-double-the-performance.184895/post-2500006
If you want to keep your inefficient resistor loading, you can get rid of Q2, 6, 7, 8 and just keep the error-correcting engine.
You can go overboard, and adopt UniGaBuf: https://www.diyaudio.com/community/...er-cut-out-to-be-a-leader.185501/post-2511348
In this example, a single diode (D2) is used for error correction:
Another possibility is to use the Tringlotron topology: https://www.diyaudio.com/community/...sed-tringlotron-amplifier.165131/post-2325621
The tringlotron has also been applied to chipamps: the chtringlunator: https://www.diyaudio.com/community/...-regulator-chip-amplifier.175827/post-2340491
Etc, etc.
If you want to keep your inefficient resistor loading, you can get rid of Q2, 6, 7, 8 and just keep the error-correcting engine.
You can go overboard, and adopt UniGaBuf: https://www.diyaudio.com/community/...er-cut-out-to-be-a-leader.185501/post-2511348
In this example, a single diode (D2) is used for error correction:
Another possibility is to use the Tringlotron topology: https://www.diyaudio.com/community/...sed-tringlotron-amplifier.165131/post-2325621
The tringlotron has also been applied to chipamps: the chtringlunator: https://www.diyaudio.com/community/...-regulator-chip-amplifier.175827/post-2340491
Etc, etc.
It's a video amplifier actually, well documented for years now - local feedback to unity gain, can have some gain too but as a SE main amp not most suitable.
It needs the buffer from the left side, at least.
If you need error correction, local or overall, amplification is needed before feedback can be applied. But there is a limited 'gain window'.
Hi All ... & many thanks for your many helpful and constructive replies 🙂 ... Looks very interesting, indeed!
However, reading through them I realize that I need a bit of time to "decipher" what is going on inside the various circuitries - so for now I would say that I (personally) do not need further replies (but of course feel welcome to continue the thread as you like ...)
Again, many thanks for your feedbacks!
Cheers, Jesper
However, reading through them I realize that I need a bit of time to "decipher" what is going on inside the various circuitries - so for now I would say that I (personally) do not need further replies (but of course feel welcome to continue the thread as you like ...)
Again, many thanks for your feedbacks!
Cheers, Jesper
Hi all,
First again thanks for your feedbacks a little while back - they have been really inspiring ... and also slightly "scary" as I have come to realize that my familiarity with amplifier design "building blocks" is (too) limited, I think.
As it is the Kulish cell that @dreamth mentioned above has been the one I have mostly focused on because it appeared to offer a very significant reduction of distortion while at the same time being practical to work with. So I have made a good number of simulations with the Kulish structure and besides being quite discerning with the value of the R3 resistor it IMHO simulates remarkably well. 2H values of - 110 to - 125 dB for a 2-4 Vpp output signal with a 2-4 times amplification is not too difficult (though it takes some fiddling with values to find the best value for this R3).
However, when actually building and testing the circuitry I find that the distortion level typically is 15-20 dBs worse than the simulated result(s). Particularly surprising was the discrepancy between a 2N3819 (Q1) and 2SC3324 (Q2) combination which could not be adjusted to better than ~40 dB less than the simulated values (I used the LTSpice 2N3819 and Toshiba 2SC3324 models). For convenience here's a link to the Kulish cell structure:
https://www.diyaudio.com/community/threads/single-darlington-line-preamp.74861/page-3#post-858266
I reckon there can be some discrepancy between simulation FFTs and reality - but would you have any suggestions as to what would be reasonable expectations?
Also, I have this not-yet-materialized image in my head of a "3 transistor Kulish + Sziklai" structure that would give low-distortion amplification with a high input impedance and a low output impedance (which I personally think is one of the less optimal aspects of the Kulish structure, i.e. the highish output impedance of R3+R4) ... I have tried to simulate some possible ways of combining the Kulish with a Sziklai but not with a feasible result - might someone else here have had more constructive/similar thoughts in this direction?
And then, reading about linuxguru's development of and progress with a partly Kulish based amplifier on pages 4 & 5 in this link (courtesy dreamth):
https://www.diyaudio.com/community/...ass-a-ef-o-p-stage.114249/page-4#post-1419114
... it makes me wonder if there exists like a "library" of, well, "building blocks" that describe various amplifier "parts" that can be used to "assemble" an amplifier - with descriptions and explanations of what they do/how they work? I am familiar with Douglas Self's Small Signal Amplifier book, Cordell's Power Amplifier Book, and Kolinummi's Audio Amplifier book ... but it would be great to have an overview of options in one place so that LTP, CFP, Wilson's mirror, current conveyor, etc, etc. were gathered in one not too complex text ... But I reckon I am dreaming here ...
@Citizen124032:
Any chance you would have a link to a paper on how a modest gain can be added to the Sziklai?
Well, thanks for reading & wishing you a good day/evening wherever you may be 😉
Jesper
First again thanks for your feedbacks a little while back - they have been really inspiring ... and also slightly "scary" as I have come to realize that my familiarity with amplifier design "building blocks" is (too) limited, I think.
As it is the Kulish cell that @dreamth mentioned above has been the one I have mostly focused on because it appeared to offer a very significant reduction of distortion while at the same time being practical to work with. So I have made a good number of simulations with the Kulish structure and besides being quite discerning with the value of the R3 resistor it IMHO simulates remarkably well. 2H values of - 110 to - 125 dB for a 2-4 Vpp output signal with a 2-4 times amplification is not too difficult (though it takes some fiddling with values to find the best value for this R3).
However, when actually building and testing the circuitry I find that the distortion level typically is 15-20 dBs worse than the simulated result(s). Particularly surprising was the discrepancy between a 2N3819 (Q1) and 2SC3324 (Q2) combination which could not be adjusted to better than ~40 dB less than the simulated values (I used the LTSpice 2N3819 and Toshiba 2SC3324 models). For convenience here's a link to the Kulish cell structure:
https://www.diyaudio.com/community/threads/single-darlington-line-preamp.74861/page-3#post-858266
I reckon there can be some discrepancy between simulation FFTs and reality - but would you have any suggestions as to what would be reasonable expectations?
Also, I have this not-yet-materialized image in my head of a "3 transistor Kulish + Sziklai" structure that would give low-distortion amplification with a high input impedance and a low output impedance (which I personally think is one of the less optimal aspects of the Kulish structure, i.e. the highish output impedance of R3+R4) ... I have tried to simulate some possible ways of combining the Kulish with a Sziklai but not with a feasible result - might someone else here have had more constructive/similar thoughts in this direction?
And then, reading about linuxguru's development of and progress with a partly Kulish based amplifier on pages 4 & 5 in this link (courtesy dreamth):
https://www.diyaudio.com/community/...ass-a-ef-o-p-stage.114249/page-4#post-1419114
... it makes me wonder if there exists like a "library" of, well, "building blocks" that describe various amplifier "parts" that can be used to "assemble" an amplifier - with descriptions and explanations of what they do/how they work? I am familiar with Douglas Self's Small Signal Amplifier book, Cordell's Power Amplifier Book, and Kolinummi's Audio Amplifier book ... but it would be great to have an overview of options in one place so that LTP, CFP, Wilson's mirror, current conveyor, etc, etc. were gathered in one not too complex text ... But I reckon I am dreaming here ...
@Citizen124032:
Any chance you would have a link to a paper on how a modest gain can be added to the Sziklai?
Well, thanks for reading & wishing you a good day/evening wherever you may be 😉
Jesper
A Sziklai does amplify, but it is depending on how it is configured in the circuit. It is also named a transconductance amplifier, and can be done with bipolars, j-fets, mosfets and combinations of them.
In (older) application notes (TI, Harris, Linear, Siliconix), several circuits (amplifiers and others) were well documented and explained. Try to find them though most of them survive in paper form on forgotten shelves.
Sziklai's are applied in the Hiraga amplifiers, and there is a lot of it available.
Video amplifiers are used in video recorders (now most are digital, but find the analog ones), satellite receivers (broadband amps with up to 8+MHz flat bandwidth) and various measument devices like lock-in amplifiers.
There is also a lot in the Horowitz & Hill, I'm selling mine if you hit a deal (pm).
Others here may supply you with countless links and hints.
Designing electronics is like cooking: know your ingredients and select the best for the ultimate dish.
In (older) application notes (TI, Harris, Linear, Siliconix), several circuits (amplifiers and others) were well documented and explained. Try to find them though most of them survive in paper form on forgotten shelves.
Sziklai's are applied in the Hiraga amplifiers, and there is a lot of it available.
Video amplifiers are used in video recorders (now most are digital, but find the analog ones), satellite receivers (broadband amps with up to 8+MHz flat bandwidth) and various measument devices like lock-in amplifiers.
There is also a lot in the Horowitz & Hill, I'm selling mine if you hit a deal (pm).
Others here may supply you with countless links and hints.
Designing electronics is like cooking: know your ingredients and select the best for the ultimate dish.
@Citizen124032: & thanks for your feedback.
As it is I was/am hoping to find a "text" of some kind that deals with these error correction circuitries, i.e. low-distortion circuitries without global feedback, in like "one volume" or "one cookbook" so that I can get an overview & understanding of the possible circuitries & principles before being specific about certain topologies. Anyway, I will check out transconductance amplifier and see where it leads - thanks for the tip 🙂
Regarding your Horowitz & Hill book - I will just consider - borrow in from the library - (and then I may get back to you on PM later).
Cheers,
Jesper
As it is I was/am hoping to find a "text" of some kind that deals with these error correction circuitries, i.e. low-distortion circuitries without global feedback, in like "one volume" or "one cookbook" so that I can get an overview & understanding of the possible circuitries & principles before being specific about certain topologies. Anyway, I will check out transconductance amplifier and see where it leads - thanks for the tip 🙂
Regarding your Horowitz & Hill book - I will just consider - borrow in from the library - (and then I may get back to you on PM later).
Cheers,
Jesper