Hi All,
I have what I hope is a quick question for the right person(s) here ...
As it is I'm looking for a simple & very good sounding class A/B output stage to be used for a low to medium power amplifier.
The amplifier will be driven from a single-ended voltage amplification stage.
Some specs could be:
- output power ~ 20 watts/8 ohms
- Non-global-feedback
- PSU voltages up to +/- 28 VDC
- preferably doesn't need a servo (voltage amplification stage should be DC stable)
- Preferably an emitter follower solution with proper output biasing scheme ...
- a stable and worry-free solution 🙄
Might one of you help with a link or a small schematic of such an output stage?
Thanks for any help you may give 😉
Jesper
I have what I hope is a quick question for the right person(s) here ...
As it is I'm looking for a simple & very good sounding class A/B output stage to be used for a low to medium power amplifier.
The amplifier will be driven from a single-ended voltage amplification stage.
Some specs could be:
- output power ~ 20 watts/8 ohms
- Non-global-feedback
- PSU voltages up to +/- 28 VDC
- preferably doesn't need a servo (voltage amplification stage should be DC stable)
- Preferably an emitter follower solution with proper output biasing scheme ...
- a stable and worry-free solution 🙄
Might one of you help with a link or a small schematic of such an output stage?
Thanks for any help you may give 😉
Jesper
Last edited:
Based on my experience, for good sound you'll want an OPS which has good PSRR. A diamond buffer gives about the best PSRR I've seen of the more traditional OPSs but I can't help thinking it must be possible to beat the PSRR of a diamond by cascoding the output devices. Maybe in the style used in the beta22 amplifier (which apparently sounds very good). The schematic of that is here - The β22 stereo amplifier
Hi abraxalito,
... thanks for your suggestion - which also made me realize that I need to add to my first post that it should also be NGFB ... sorry for the confusion 😱
BTW, the stage you suggest isn't single-ended as far as I can see ... or might I have missed something here?
Have a good day 😉
Jesper
... thanks for your suggestion - which also made me realize that I need to add to my first post that it should also be NGFB ... sorry for the confusion 😱
BTW, the stage you suggest isn't single-ended as far as I can see ... or might I have missed something here?
Have a good day 😉
Jesper
Not sure what you mean by single ended VAS. Do you mean bootstrapped, rather than using a constant current source or single transistor rather than so called "balanced VAS" configuration?
Here's link to a well known catalogue of basic types that I would call "single ended" if you needed to distinguish from anything with differential outputs etc: Distortion In Power Amplifiers. See section 5.2.2
Here's link to a well known catalogue of basic types that I would call "single ended" if you needed to distinguish from anything with differential outputs etc: Distortion In Power Amplifiers. See section 5.2.2
Now its my turn to be confused. You asked for 'classA/B' but that can only be push-pull, not SE. SE would be pure classA.
So did you really want a single-ended output stage?
@abraxalito:
🙂 ... this communication ...
No, as I write in #1 the VAS will be single-ended whereas I would like the output stage to be push-pull (or the like) and class A/B. So what I need preferably is a schematic for a SE-to-PP configuration ...
@Ian Finch: It's are very basic VAS stage with an NPN & a PNP transistor working counter to eachother voltage-wise. I just need to connect it to an output stage - basically it's a good biasing configuration & SE-to-PP I'm looking for.
Hope this makes it clearer ...
Cheers,
Jesper
🙂 ... this communication ...
No, as I write in #1 the VAS will be single-ended whereas I would like the output stage to be push-pull (or the like) and class A/B. So what I need preferably is a schematic for a SE-to-PP configuration ...
@Ian Finch: It's are very basic VAS stage with an NPN & a PNP transistor working counter to eachother voltage-wise. I just need to connect it to an output stage - basically it's a good biasing configuration & SE-to-PP I'm looking for.
Hope this makes it clearer ...
Cheers,
Jesper
Last edited:
OK so, it's an AB class amp with typical Current Source VAS. This may simply mean a generic amplifier but NGBF would make it very rare, if it could perform with acceptably low noise and distortion whilst sounding as good as people may expect. However, if it were that simple, it would have been done to great commercial success long ago.
It would be helpful to know what the input stage was to be too. The gain and local feedback in the front end play a large part in determining what is possible here.
It would be helpful to know what the input stage was to be too. The gain and local feedback in the front end play a large part in determining what is possible here.
Hi Guys
The standard Lin-Thompson topology used by Self and everyone else falls into the description Jesper provided: asymmetric front-end with single-ended VAS driving a push-pull output stage. The VAS is single-ended by virtue of the fact that only one side is signal-driven - the current-source load for it is not signal-driven.
As far as an output stage in isolation, used in a non-global-feedback amplifier, check out Lineup's power buffer. It uses a diamond internally for gain correction and offset control. The thread shows the use of mosfets or BJTs for the output. In his amp, he has a separate front-end that has feedback around it, driving the separate power buffer. http://www.diyaudio.com/forums/solid-state/222816-new-lineup-idea-power-follower-output-stage.html
Note that the lowest THD will be attained if the output stage is within the feedback loop. Self's Compact blameless has single-digit ppm THD using output-inclusive compensation, aka transitional Miller comp. Just 17 devices including protection, which he counts as 13 devices... You can see the circuit and data for that on his site Signal Transfer Company, or in his book.
Personally I prefer complementary-symmetry front-ends and outputs.
Have fun
The standard Lin-Thompson topology used by Self and everyone else falls into the description Jesper provided: asymmetric front-end with single-ended VAS driving a push-pull output stage. The VAS is single-ended by virtue of the fact that only one side is signal-driven - the current-source load for it is not signal-driven.
As far as an output stage in isolation, used in a non-global-feedback amplifier, check out Lineup's power buffer. It uses a diamond internally for gain correction and offset control. The thread shows the use of mosfets or BJTs for the output. In his amp, he has a separate front-end that has feedback around it, driving the separate power buffer. http://www.diyaudio.com/forums/solid-state/222816-new-lineup-idea-power-follower-output-stage.html
Note that the lowest THD will be attained if the output stage is within the feedback loop. Self's Compact blameless has single-digit ppm THD using output-inclusive compensation, aka transitional Miller comp. Just 17 devices including protection, which he counts as 13 devices... You can see the circuit and data for that on his site Signal Transfer Company, or in his book.
Personally I prefer complementary-symmetry front-ends and outputs.
Have fun
Hi Ian & Struth (remember we've met here before but don't right now have your name present) ...
Thanks for your feedback - I won't, however, have time to look at it today but will get back on Monday ...
Have a good weekend when you get to there ...
Cheers ;-)
Jesper
Thanks for your feedback - I won't, however, have time to look at it today but will get back on Monday ...
Have a good weekend when you get to there ...
Cheers ;-)
Jesper
Although your conditions are not fully met i advise to examine the Circlophone. I made 2 for an active shelf speaker easy to build works instantaneously w/o any adjustment, excellent warm sound. Problem is that idle power dissipation is too high as the size of heat sinks is limited by the speaker enclosure dimensions and this also puts a limit on transformer size. If that is not an issue for you this amp is highly recommended
Hi all,
First - thanks for your feedback ... I've now had time to look into the suggestions you've made and I think Struth's lineup IDEA is close to what I am looking for except that I would also prefer the output stage to be "no-overall-feedback" - hope that makes sense.
Actually, I've been thinking about it myself over the weekend and have come up with this very simple circuitry. Please note that I am aware that this isn't an advanced circuitry - but I know that the input circuitry sounds very well when not driven hard, and since this design is to not take too much time to build I'd like to go with something simple.
So, unless one of you have amendments I think I'll consider Struth's suggestion or the schematic I just posted.
Cheers & thanks again for your suggestions ;-)
Jesper
First - thanks for your feedback ... I've now had time to look into the suggestions you've made and I think Struth's lineup IDEA is close to what I am looking for except that I would also prefer the output stage to be "no-overall-feedback" - hope that makes sense.
Actually, I've been thinking about it myself over the weekend and have come up with this very simple circuitry. Please note that I am aware that this isn't an advanced circuitry - but I know that the input circuitry sounds very well when not driven hard, and since this design is to not take too much time to build I'd like to go with something simple.
So, unless one of you have amendments I think I'll consider Struth's suggestion or the schematic I just posted.
Cheers & thanks again for your suggestions ;-)
Jesper
R.Cordell has a chapter dedicated to Non Global feedback amplifiers.
It is a good read and showed me why I couldn't design something like that.
It is a good read and showed me why I couldn't design something like that.
Member
Joined 2009
Paid Member
It's all the art of tradeoffs. You will be accepting higher distortion than a GNFB amplifier if you want something that is simple without feedback. Nothing wrong with that at all.
Multiple output devices can be a big help, it may not meet your need for simple (?) but more output pairs and you get lower distortion because the current variation through each device is reduced. You also get more heat and a higher power level at which Class A gives way to Class B.
see also: http://www.diyaudio.com/forums/solid-state/174126-best-topology-no-feedback-classab-buffer.html
Multiple output devices can be a big help, it may not meet your need for simple (?) but more output pairs and you get lower distortion because the current variation through each device is reduced. You also get more heat and a higher power level at which Class A gives way to Class B.
see also: http://www.diyaudio.com/forums/solid-state/174126-best-topology-no-feedback-classab-buffer.html
Member
Joined 2009
Paid Member
It's all the art of tradeoffs. You will be accepting higher distortion than a GNFB amplifier if you want something that is simple without feedback. Nothing wrong with that at all.
Multiple output devices can be a big help, it may not meet your need for simple (?) but more output pairs and you get lower distortion because the current variation through each device is reduced. You also get more heat and a higher power level at which Class A gives way to Class B.
see also: http://www.diyaudio.com/forums/solid-state/174126-best-topology-no-feedback-classab-buffer.html
Multiple output devices can be a big help, it may not meet your need for simple (?) but more output pairs and you get lower distortion because the current variation through each device is reduced. You also get more heat and a higher power level at which Class A gives way to Class B.
see also: http://www.diyaudio.com/forums/solid-state/174126-best-topology-no-feedback-classab-buffer.html
I got pretty heavily into poweramp design back in the 1980's while working with some very sharp EE's at Tektronix. I learned about how phase margin works, and current limit devices for the output transistors, that emitter coupled outputs depend much less on negative feedback for correction. More recently I read up on the crossover distortion issue, which varies with temperature and reactive load variations. Advising a specific circuit would be challenging, since any circuit is experimental until you've verified with test equipment that the phase margin is excellent, overload is handled well, and crossover distortion stays non-existent for all operating conditions.
A perfectly balanced symetrical topology looks cool on paper, but will produce a distortion spectrum that is odd harmonics only. It's arguable that an even mix of odd and even harmonic distortion products will sound more natural, and thereby less objectionable. Using a single ended front end outside of the feedback loop in the output stage will help that effort. If I was going to try to keep the circuit "simple", I might look at using MOSFETs for the output devices, biased pretty high to minimize crossover distortion. These are the main issues I'd worry about.
Personally I am quite happy using the LM3886 power amp IC's for poweramps. They're not perfect, but it's a relatively simple project, and they claim to have addressed the crossover distortion issue, maybe better than most discrete poweramps. If you're trying to create a real masterpiece, you're probably in for a fairly time consuming project. One other cool thing about the LM3886 is that it has no turn on transient, so can be direct coupled to a tweeter in a bi or tri amp situation and not damage the tweeter.
A perfectly balanced symetrical topology looks cool on paper, but will produce a distortion spectrum that is odd harmonics only. It's arguable that an even mix of odd and even harmonic distortion products will sound more natural, and thereby less objectionable. Using a single ended front end outside of the feedback loop in the output stage will help that effort. If I was going to try to keep the circuit "simple", I might look at using MOSFETs for the output devices, biased pretty high to minimize crossover distortion. These are the main issues I'd worry about.
Personally I am quite happy using the LM3886 power amp IC's for poweramps. They're not perfect, but it's a relatively simple project, and they claim to have addressed the crossover distortion issue, maybe better than most discrete poweramps. If you're trying to create a real masterpiece, you're probably in for a fairly time consuming project. One other cool thing about the LM3886 is that it has no turn on transient, so can be direct coupled to a tweeter in a bi or tri amp situation and not damage the tweeter.
Hi Bigun & Bob,
Appreciate your comments and suggestions 🙂 Hmmm... there are so many ways in audio that sometimes frustration may just slightly creep in when decisions are to be made
- maybe you also know of something similar ...?
Anyway, as I write
@Bigun: Thanks for the link Bigun ... looks like many interesting and relevant options ... and your comment about the parallelled output devices .. This could actually be an idea ...
@Bob Richards: I've also considered chip amps but given that they operate with GNFB I'm less inclined to use them. As it is I just happen to have had my best audio experiences with low - to - no GNFB ...
So, I will just ponder what to do ...
Wish you a pleasant day 😉
Jesper
Appreciate your comments and suggestions 🙂 Hmmm... there are so many ways in audio that sometimes frustration may just slightly creep in when decisions are to be made
- maybe you also know of something similar ...?Anyway, as I write
my aim here is make a very simple amplifier with good sound quality - and low listening fatique - that I can use to listen to during the day. So I prefer to keep it simple and use tried and tested solutions.
@Bigun: Thanks for the link Bigun ... looks like many interesting and relevant options ... and your comment about the parallelled output devices .. This could actually be an idea ...
@Bob Richards: I've also considered chip amps but given that they operate with GNFB I'm less inclined to use them. As it is I just happen to have had my best audio experiences with low - to - no GNFB ...
So, I will just ponder what to do ...
Wish you a pleasant day 😉
Jesper
Last edited:
Hi Bigun & Bob,
Appreciate your comments and suggestions 🙂 Hmmm... there are so many ways in audio that sometimes frustration may just slightly creep in when decisions are to be made
Anyway, as I write my aim here is make a very simple amplifier with good sound quality - and low listening fatique - that I can use to listen to during the day. So I prefer to keep it simple and use tried and tested solutions.
@Bigun: Thanks for the link Bigun ... looks like many interesting and relevant options ... and your comment about the parallelled output devices .. This could actually be an idea ...
@Bob Richards: I've also considered chip amps but given that they operate with GNFB I'm less inclined to use them. As it is I just happen to have had my best audio experiences with low - to - no GNFB ...
So, I will just ponder what to do ...
Wish you a pleasant day 😉
Jesper
That's why I designed and built a tube type poweramp with 0-15dB of feedback (adjustable), which only goes back to the diff amp phase splitter, leaving the 6SN7 front end (with CC plate load) outside of the feedback loop. The Pushpull output stage is EL34's hooked ultra-linear or triode mode.
Hi again Bob,
... well, without knowing much about tubes it seems to me that what you suggest is an interesting solution, yet in this context I would like to thread known paths and not get "diverted" into new directions - even if they appear enticing ...
So, in the first round thanks again for this outline of your solution - I may keep it in mind for a later point in time ...
Cheers ;-)
Jesper
... well, without knowing much about tubes it seems to me that what you suggest is an interesting solution, yet in this context I would like to thread known paths and not get "diverted" into new directions - even if they appear enticing ...
So, in the first round thanks again for this outline of your solution - I may keep it in mind for a later point in time ...
Cheers ;-)
Jesper
The trouble with paralleling devices is that they can tug-of-war on each other to the degree that they are mis-matched. With FETs in parallel, the input capacitance could be an issue (will be the dominant pole). I've seen people parallel LM3886 chips, and I shuddered... I'm not sure a good poweramp can be "simple". It's about damage control, with high power driving a highly reactive load.
- Status
- Not open for further replies.