We did, for power amplifiers I still think max output being only 2X the standing current is problematic.
A lot of people wonder what is bad about an IC/opamp if the numbers are so good. So, it is good just to point out where the descrete circuit advantage lie -- the output stage, most often.
But now there are some good buffer IC to add-on to fix the 'problem' . However some of those have same issues of cross-over distortion and so it is a mixed bag again.
The light load on an opamp OPS, plus operating it in class A, by using a compl class A fet buffer does seem useful.
THx-RNMarsh
But now there are some good buffer IC to add-on to fix the 'problem' . However some of those have same issues of cross-over distortion and so it is a mixed bag again.
The light load on an opamp OPS, plus operating it in class A, by using a compl class A fet buffer does seem useful.
THx-RNMarsh
Last edited:
What the hell about class AB, and their distortion artifacts at high frequencies no human ears can notice ?
If it measures (harmonics in the 20-20000 range, remembering Fletcher and Munson curves), and sounds good enough ?
More than this. if an amp is set to work in class A for the average level of the music, i believe we can find an advantage to entering in class B during the short high transients. Percussions, kick drums, bass attacks.
Compression of the musical content is a problem during all the process of recording creation and reproduction. Including thermal effects in speakers coils.
I found ( a very unconventional and personal point of view) that the little added distortion when the amp goes to class B during those peaks, if they are short enough to not let the time to our brain to analyse-it, can lighten them in a way, adding some little more 'presence' to them.
I often used over level recording (on analog tapes) for kick drums to add what producers founded a very natural 'skin impact' ;-)
And, you are right, dvv, full class A for high power amps makes the things "big and very impractical".
If it measures (harmonics in the 20-20000 range, remembering Fletcher and Munson curves), and sounds good enough ?
More than this. if an amp is set to work in class A for the average level of the music, i believe we can find an advantage to entering in class B during the short high transients. Percussions, kick drums, bass attacks.
Compression of the musical content is a problem during all the process of recording creation and reproduction. Including thermal effects in speakers coils.
I found ( a very unconventional and personal point of view) that the little added distortion when the amp goes to class B during those peaks, if they are short enough to not let the time to our brain to analyse-it, can lighten them in a way, adding some little more 'presence' to them.
I often used over level recording (on analog tapes) for kick drums to add what producers founded a very natural 'skin impact' ;-)
And, you are right, dvv, full class A for high power amps makes the things "big and very impractical".
Last edited:
For a headphone amp, I use power mosfets.
I would like to point out that under some conditions, there isn't much of a problem, especially if you don't design extremely noise circuits. So, a feedback resistor of 10K and above and a following load that is not excessive, (such as a spectrum analyzer) the results, even with a typical IC output stage are usually pretty good.
However, MANY real time audio designers try to get the lowest noise, so they often use 500 ohms or less as a load on a pretty good IC, in order to keep the gain where they want it, and the noise as low as possible. This is where the problem shows up. The only 'good' solution is to buffer the IC that is subjected to this requirement, BUT many suggestions here take up much more space than the IC, itself, even though they would work. It is impractical, in many cases, to squeeze the buffer into the space available, so it is often not done. Guess what? It is difficult to find any problem, except by ear. So, people put out audio products with more 'compromise' in them than the specs imply.
I would like to point out that under some conditions, there isn't much of a problem, especially if you don't design extremely noise circuits. So, a feedback resistor of 10K and above and a following load that is not excessive, (such as a spectrum analyzer) the results, even with a typical IC output stage are usually pretty good.
However, MANY real time audio designers try to get the lowest noise, so they often use 500 ohms or less as a load on a pretty good IC, in order to keep the gain where they want it, and the noise as low as possible. This is where the problem shows up. The only 'good' solution is to buffer the IC that is subjected to this requirement, BUT many suggestions here take up much more space than the IC, itself, even though they would work. It is impractical, in many cases, to squeeze the buffer into the space available, so it is often not done. Guess what? It is difficult to find any problem, except by ear. So, people put out audio products with more 'compromise' in them than the specs imply.
Well, that's the essence of class A, and unless you cheat in some way (yes, there are creative ways of cheating, and I am not the last to use them), there is no way round.
Let's see this in perspective: even a maximally efficient class B tube amplifier is likely to have an overall efficiency lower than a class A semi amp (not a SE, resistor loaded type though)
Make a comparison with a low noise SOTA class A design that measures below -120dBr and has no high order harmonics, and then tell me. You might be surprised. No Alephs or XAs, please.
I think you gentlemen are overthinking the point. It's now come to who makes a nicer circuit. Remember, the starting point was how to improve the sound of common op amps (esoteria excluded).
I had my mind on SIMPLE solution, which are guaranteed to work almost anywhere under reasonable conditions. See Picture 1 for the basic idea, which has been implemented by so many people so far I long lost count. Picture 2 is the alternative, and such circuits are used in reVox devices. Picture 3 is a simple derivative, or expansion if you like, on the basic there for those occasions you feel you need LOTS of current for whatever reason.
Just one note: please disregard the values shown, I just dropped them in out of the blue, they are there for illustration purposes only.
Regarding Picture 3, a friend just as curious as I am and I once used OPA 2604 fed off +/- 22V lines, made the first pair (drivers) BD 139/140 and the ouput devices 2SC5200/2SA1943. We ended up with a true power amplifier, limited to about 20 Wrms and generally not sounding bad at all. Hardly a pinnacle model, but so simple and cheap that it's a Godsend for say active loudspeakers, or some such.
These are hardly SOTA circuits, but then, they are simple to make, use standard parts and only a few of them, which makes them reliable as there is less to go wrong and need less work, plus a more modest cash outlay. Yet, they will transforme ANY standard linear op amp, at least have done so in my experience.
I had my mind on SIMPLE solution, which are guaranteed to work almost anywhere under reasonable conditions. See Picture 1 for the basic idea, which has been implemented by so many people so far I long lost count. Picture 2 is the alternative, and such circuits are used in reVox devices. Picture 3 is a simple derivative, or expansion if you like, on the basic there for those occasions you feel you need LOTS of current for whatever reason.
Just one note: please disregard the values shown, I just dropped them in out of the blue, they are there for illustration purposes only.
Regarding Picture 3, a friend just as curious as I am and I once used OPA 2604 fed off +/- 22V lines, made the first pair (drivers) BD 139/140 and the ouput devices 2SC5200/2SA1943. We ended up with a true power amplifier, limited to about 20 Wrms and generally not sounding bad at all. Hardly a pinnacle model, but so simple and cheap that it's a Godsend for say active loudspeakers, or some such.
These are hardly SOTA circuits, but then, they are simple to make, use standard parts and only a few of them, which makes them reliable as there is less to go wrong and need less work, plus a more modest cash outlay. Yet, they will transforme ANY standard linear op amp, at least have done so in my experience.
I didn't know that harmonics issues with increased reservoir caps is very well known as stated by Jan. Kamis, just trust your ears, you cannot rely on experts
I like overrated transformer, but not caps.
I don't want to invest my time doing "controlled experiment". I just do a "hit and miss" approach. My assumption is that for every pcb trace, power supply, impedance, capacitor brand, circuit implementation, speaker used, bla-bla-bla, there is optimum value of capacitors. Doing "hit and miss" is easier than calculating the right caps for me.
Just like you, Kamis, when I replaced the caps and suddenly I don't like the sound, I allow for some "breaking-in" time then if no improvement I will change it.
"Latest" opinion is that as long as you parallel many of the same size capacitors, the more the better. Now I don't believe this as my ears have different experience.
Also, by doing a "hit and miss", I often put paralleled small value FKP capacitor and I like the sound. It could be a coincidence as it is strange how it can affect the sound (and often it is not the only change that I do at one time). But as long as I have so many components in stock (and I'm afraid they will "die" from being unused) I don't feel sorry making this (putting FKP bypass cap) as my new tradition
Depends on how literal you define it, even your example forces commutation linearly between the two halves of the output using a lot of local feedback and then fails when one side cuts off. So in essence why is that different from a power law commutation as long as neither side cuts off?
Are-you sure the difference is not related to dynamic, power rails side ? (higher, but more constant consumption ?).
You are talking about -120dB. of course, it is reassuring, if you can. Adding any kind of signal, related or not, to an existing musical content, do you think you can even notice his presence at -100dB with the mask effect ?
You know, like dither in a 16bits DA converter ;-)
- Status
- Not open for further replies.