Actually if you squint at the teaser image for the article I Iinked to you have the gist of it. 😉
Jan
That's not really the point, in the context of op-amp with feedback my favorite stimulus is the tri-wave, constant rate of change. The input error signal becomes a square wave, its derivative, and looking at the input and output voltages everything becomes clear. During slew (I don't mean slew limit just a constant rate of change of output voltage with time) the input is a DC voltage representing the tilting of the input stage currents while the output slews.
I'm out and about and at my GF's place, and trying to type on thie silly little vir5ual keyboard.
Ok, right now I don't have formal proofs for what I am saying and have distilled from long experience, but I have derived loose measurements in the past.
For now this is not important, but into the near future yes such measurements are of critical importance.
Immense thank you to Scott for providing me a range of high quality devices that will enable proper measurements in time, nobody in this diyaudio ecosystem that we are all part of will be forgotten.
Perhaps my assertions about changing/cleaning the nature of energy transfer are faulty but I don't think so, time will tell.
In the meantime we have anecdotal/experience evidence only, but indeed this is where science starts.
Friday night was confirmation final trial on a pub PA, with the PA being treated for the last set.
After the show were numerous unsolicited comments on the groove/niceness/enthralling quality of the sound during the last set, and that the sound took a huge jump compared to the previous two sets.
The sound engineer copped compliments from FOUR audience WOMEN during last set and post show, this is unprecedented....we high fived and both laughed the deepest laugh.
More later.....
Dan.
Ok, right now I don't have formal proofs for what I am saying and have distilled from long experience, but I have derived loose measurements in the past.
For now this is not important, but into the near future yes such measurements are of critical importance.
Immense thank you to Scott for providing me a range of high quality devices that will enable proper measurements in time, nobody in this diyaudio ecosystem that we are all part of will be forgotten.
Perhaps my assertions about changing/cleaning the nature of energy transfer are faulty but I don't think so, time will tell.
In the meantime we have anecdotal/experience evidence only, but indeed this is where science starts.
Friday night was confirmation final trial on a pub PA, with the PA being treated for the last set.
After the show were numerous unsolicited comments on the groove/niceness/enthralling quality of the sound during the last set, and that the sound took a huge jump compared to the previous two sets.
The sound engineer copped compliments from FOUR audience WOMEN during last set and post show, this is unprecedented....we high fived and both laughed the deepest laugh.
More later.....
Dan.
This is a very interesting way of looking at it, gives quite a lot of insight! Thanks Scott for sharing!
Jan
That's not really the point, in the context of op-amp with feedback my favorite stimulus is the tri-wave, constant rate of change. The input error signal becomes a square wave, its derivative, and looking at the input and output voltages everything becomes clear. During slew (I don't mean slew limit just a constant rate of change of output voltage with time) the input is a DC voltage representing the tilting of the input stage currents while the output slews.
The example is most clear with a dominant pole type of amplifier which is still very common especially with IC op-amps. You can also use this test to clearly see a small square wave of current going into the inverting input of a "so called" current feedback amplifier while it is making a tri-wave at its output. This is the current that is charging the comp cap, and gee it's coming from the feedback circuit is that why someone called it current feedback?
You can expand this to exploring what happens as you approach slew limit, certainly this applies to a 741 as was used in the TIM paper, and you can also see why FET's are better than un-degenerated bi-polar. This was all part of my training by week two 1974 (not with CFA's yet of course)
Last edited:
In modern power amps you would almost never not degenerate the input pair. Both Self and Cordell discuss this. If you don't, then I would make sure to bandwidth limit the input signal appropriately. I think this is one of the things that lead early (1970's) solid state amplifier problems. I note however that Bart Locanthi was already degenerating the input diff pair for all the right reasons back in the 1960's - the guy was well ahead of his time.
There are only two general ways to increase the slew rate of an amp. Either you degenerate the input bipolar transistors, or you increase the gain-bandwidth of the output stage. A better alternative is to use jfets on the input that automatically lower the input transconductance, yet are pretty quiet as well.
![sub_wall_body1.1_1000[1].jpg](https://www.diyaudio.com/community/data/attachments/545/545782-49e002d6a2e5cd327cf42e23f9f0509f.jpg?hash=SeAC1qLlzT "sub_wall_body1.1_1000[1].jpg")
as I said, 1975 textbook EE - read Roberge
2-pole compensation formally doesn't have a slew rate limit in the region of 2nd order slope
practically there are still parasitic/Miller C of the VAS Q to drive, but the main Ccomp isn't Miller multiplied in 2-pole
so an easy order of magnitude is increase in audio band slew rate limit is avalialbe with only 5-10 degrees added phase shift at the same unity gain intercept
to take advantage of 2-pole compensation you should be using "beta enhanced"/cascode or other high gain "VAS" stage
"TMC" is equivalent to some 2-pole compensation, can give the same slew improvement
of course faster output Q, higher than 500kHz unity gain intercept really makes (either) 2-pole compensation method shine
2-pole compensation formally doesn't have a slew rate limit in the region of 2nd order slope
practically there are still parasitic/Miller C of the VAS Q to drive, but the main Ccomp isn't Miller multiplied in 2-pole
so an easy order of magnitude is increase in audio band slew rate limit is avalialbe with only 5-10 degrees added phase shift at the same unity gain intercept
to take advantage of 2-pole compensation you should be using "beta enhanced"/cascode or other high gain "VAS" stage
"TMC" is equivalent to some 2-pole compensation, can give the same slew improvement
of course faster output Q, higher than 500kHz unity gain intercept really makes (either) 2-pole compensation method shine
Last edited:
In case anyone is interested to take a look, a Roberge pdf can be found here: https://ocw.mit.edu/resources/res-6-010-electronic-feedback-systems-spring-2013/textbook/
I like the last sentence, 1969 and already extending full power BW is good for audio. The poor settling comes with a lot of these multipole compensation techniques so data acquisition folks tend not to use them.
Presumably, root locus plots can provide some intuition about the ringing: http://www.me.ust.hk/~mech261/index/Lecture/Chapter_7.pdf
Strange - I never had any probs with overshoot. If phase and gain margins are adequate you should be good to go. With 2 pole you should be intercepting the ULGF at 20 dB/decade. I BW limit my input signal as well BTW.
Then how do you measure slew rate?
(VoltageSwing / SlewRate) = Risetime
and(Risetime * Bandwidth) = 0.35
thus(VoltageSwing * Bandwidth) = (0.35 * SlewRate)
set the input bandwidth too low and you can't get the amplifier to slew!I think even John working with Walt found 200 kHz lpf fine
the DIM test signal had to be low passed too, 30k or 100kHz were adopted https://www.ap.com/technical-library/dim-30-and-dim-100-measurements-per-iec-60268-3-with-ap2700/
and of course there's the microphones used in music recording - as low as 14 kHz for some vocal mics, low 20 kHz for small condensors, only Earthworks offers a 50 kHz mic for recording as an option in their "Drum Kit"
the DIM test signal had to be low passed too, 30k or 100kHz were adopted https://www.ap.com/technical-library/dim-30-and-dim-100-measurements-per-iec-60268-3-with-ap2700/
and of course there's the microphones used in music recording - as low as 14 kHz for some vocal mics, low 20 kHz for small condensors, only Earthworks offers a 50 kHz mic for recording as an option in their "Drum Kit"
- Status
- Not open for further replies.