That brings up a thought/question. Would it help this to take the feedback from the output of the VAS instead of THE output? It would avoid the problem of spreading out low-harmonic distortion into higher harmonics (is there a name for that?) in the feedback loop.
Since the VAS output is a rather high impedance point, a voltage follower (emitter/source follower and CCS) could be added, and the feedback taken from that. For those on a "low transistor count budget" this could blow it, but as any engineer knows you can't have everything...
I'm afraid you will never find very good complimentarity with JFET's anywhere. They are majority carrier devices one holes one electrons, the mobility difference will always cause something to be amiss. At the very least my observation is that the output characteristics on the P's are always softer, also PFET's do not have impact ionization a direct result of the mobility difference.
Like, if after washing a baby in a water it is still relatively dirty, why do not throw it out together with the dirty water?
JC already showed balanced input to the output stage. It looks smooth on the paper, but is bumpy as usual in the reality. What we discussed here, many pages, how to find more optimal design. However, before doing that we did not stop to decide about optimization criteria, so everyone have own in mind.
It's still not too late to put forward your criteria for consideration.However, before doing that we did not stop to decide about optimization criteria, so everyone have own in mind.
I think Scott & I share very clear common criteria; eg low THD, sensible stability adjustment for different loads & gains. no funnies on overload or startup, no Unobtainium, robust to component variations so flexible with voltages, devices etc, optimised for audio rather than RF, good sound.
We might also admit to some 'show off' criteria like zillon V/us slew so we outperform Golden Pinnae stuff on the important Hirata & Quan tests.
I'd like to do as well as SW-OPA on the above with less current, less devices & less real estate.
You'll excuse me for not adopting 'new' topologies' unless they show improvements on these criteria over tried & tested.
I don't see the point of those last four words, unless this is to be manufactured for the lowest price. One percent resistors are common and cheap, and transistors are cheap enough to buy a handful and measure beta with a $20 DMM to get several pairs matching within two percent or so. It would of course be nice if a design works out to be "robust" but is it really needed?
IMHO, (and this is borne out by one of the few sim results I actually believe 100%) even with 1% resistors and 2% BJTs (and 2% FETs from the Unobtainium FET Shop) you will still need a twiddle pot for 'acceptable' DC offset and output stage quiescent current.
You can put the tweak pot(s) in various places but even in the 'best' positions, some circuits are MUCH more fussy than others. The pot range can be insufficient to allow for your 1% resistors, 2% BJTs etc and at the same time still be too twitchy to adjust accurately.
'.. robust to component variations' usually implies good thermal stability too.
Robust design- Try this in SIM -
Take a look at sim of the circuit I did last year and is in Linear Audio for something different. Did you see it -- I put it up here, at DIYAudio forum, too. Seems to fit your description of what you are looking for. TotalHD+Noise at 1-2 vrms out into >600 Ohm is under (measured as built - not sim) .0005% no dc offset and no thermal drift. No servo. Moderate gnfb. wide open loop BW. Will drive down to 30 Ohms with low distortion -- about 1W. Bias for more, if you like more class A into low Z. Just have to match the cheap devices. Not developed to compete here as a universal OPA but you can play with it's circuit to suit your needs and curiousity and compare results. THx-RNM
Take a look at sim of the circuit I did last year and is in Linear Audio for something different. Did you see it -- I put it up here, at DIYAudio forum, too. Seems to fit your description of what you are looking for. TotalHD+Noise at 1-2 vrms out into >600 Ohm is under (measured as built - not sim) .0005% no dc offset and no thermal drift. No servo. Moderate gnfb. wide open loop BW. Will drive down to 30 Ohms with low distortion -- about 1W. Bias for more, if you like more class A into low Z. Just have to match the cheap devices. Not developed to compete here as a universal OPA but you can play with it's circuit to suit your needs and curiousity and compare results. THx-RNM
Last edited:
I'd love to have a look, Mr. Marsh. Got a more precise link to the circuit? Beach bums don't buy magazines cos kunt reed en rite.
If it sims in the same ball-park as SW-OPA, you're still ahead cos Scott hasn't finished his breadboard.
Affects of dynamicly varying open-loop BW on music-
Reflected load Z artifacts etal from output to Av stage is not unknown. Output 'buffer' circuits vary in their isolation ability to minimize the affects.
But it clearly can affect performance, especially with reactive and changing loads with freq.... such as, power amp and loudspeaker reflecting back any nonelinear Z.
SO. and BUT -->
Can we make a circuit that can modulate just the open-loop bandwidth of the opamp/pwr amplifier and while this is going on, measure IM and other distortions. Especially interesting would be any phase modulation distortion caused by this.... PIM?
Such as via switched C or an injected square wave somewhere that we can control the speed of change in open-loop BW.
Secondarily, which topology in the output stage has the greatest isolation from output to Av stage. [What about no output buffer stage - just off the Av stage.] Thx- RNMarsh
Reflected load Z artifacts etal from output to Av stage is not unknown. Output 'buffer' circuits vary in their isolation ability to minimize the affects.
But it clearly can affect performance, especially with reactive and changing loads with freq.... such as, power amp and loudspeaker reflecting back any nonelinear Z.
SO. and BUT -->
Can we make a circuit that can modulate just the open-loop bandwidth of the opamp/pwr amplifier and while this is going on, measure IM and other distortions. Especially interesting would be any phase modulation distortion caused by this.... PIM?
Such as via switched C or an injected square wave somewhere that we can control the speed of change in open-loop BW.
Secondarily, which topology in the output stage has the greatest isolation from output to Av stage. [What about no output buffer stage - just off the Av stage.] Thx- RNMarsh
Last edited:
WTF ??!!
What axe would that be?
It appeared to me, kgrlee was pointing himself in a direction I had gone before so informed him of the design. Its for him to study and play with and as I said isnt a competitor here.
So, take it easy. Handle it.
I, myself and me.... have zero axe to grind. I am here to participate and contribute and learn from all view points. -R. Marsh
What axe would that be?
It appeared to me, kgrlee was pointing himself in a direction I had gone before so informed him of the design. Its for him to study and play with and as I said isnt a competitor here.
So, take it easy.
Handle it. I, myself and me.... have zero axe to grind. I am here to participate and contribute and learn from all view points. -R. Marsh
Last edited:
No axes, Scott. This is a civilised thread.
Several interesting topologies have been suggested and I've tried to apply the same criteria to the more interesting ones in SPICE land. So far SW-OPA is one of the front runners.
Also unlike certain pseudo Messiah's that shall remain nameless, Marshy has tested his circuit so it has credibility that wannabe SPICE gurus like me lack. But the race is hotting up cos you're about to test the complete SW-OPA.
____________________
Mr. Marsh, any idea when you posted that circuit in diyaudio? I'm looking through your 1000 odd posts but its slow.
Last edited:
What does it all add up to?
I have 45 years of intense learning about something. Scott has 40 years of doing something else. That right there is 95 years of continuous hard work on something. Here we have a common interest that overlaps. Ditto for JC, WJ. More than three centuries doing something which we all overlap here in our common interest areas. I dont begin to assume that I can know what another knows. That knowledge and experience ought to be respected by each and all. More than 200 years of expertise in just the 4 people mentioned above plus the many others who comment here and in other forums (N.Pass) means what is brought up is not to be taken lightly or quickly dismissed. There are a lot of years work behind my sometimes simple sounding statements/style presented as I am just as sure from the others. Thx- R.N.Marsh
I have 45 years of intense learning about something. Scott has 40 years of doing something else. That right there is 95 years of continuous hard work on something. Here we have a common interest that overlaps. Ditto for JC, WJ. More than three centuries doing something which we all overlap here in our common interest areas. I dont begin to assume that I can know what another knows. That knowledge and experience ought to be respected by each and all. More than 200 years of expertise in just the 4 people mentioned above plus the many others who comment here and in other forums (N.Pass) means what is brought up is not to be taken lightly or quickly dismissed. There are a lot of years work behind my sometimes simple sounding statements/style presented as I am just as sure from the others. Thx- R.N.Marsh
Last edited:
Where is it?
gads. Did I write all that? And, I only came here in May this year. I just wanted to help DIY headphone amp builders make the headphone amp and enjoy it. Now here I am spilling my life's guts out to a bunch of strangers on this social medium! Am I that desperate!? [not really]. Wasnt I breast fed?
Kgrlee--- I'll see if I have any better luck finding it. -RNM
Mr. Marsh, any idea when you posted that circuit in diyaudio? I'm looking through your 1000 odd posts but its slow.
gads. Did I write all that? And, I only came here in May this year. I just wanted to help DIY headphone amp builders make the headphone amp and enjoy it. Now here I am spilling my life's guts out to a bunch of strangers on this social medium!
Am I that desperate!? [not really]. Wasnt I breast fed? Kgrlee--- I'll see if I have any better luck finding it. -RNM