Tomorrow morning, wake up music, Stevie:
Do I even care about FFBP? Maybe I'll listen to Fleetwood Mac for the rest of the week.
Do I even care about FFBP? Maybe I'll listen to Fleetwood Mac for the rest of the week.
I've heard FFBP has a good sound, something like Brit-pop with a Kraftwerk influence. And some weird Belgian flavoring, dunno, luv me some Anna Karina. Yuuumy.
Christine was the soul of FM. The early blues band was great, the Bob Welsh era was even better, but it wouldda been nuttin without Christine.
Miss you girl,
Chris
Christine was the soul of FM. The early blues band was great, the Bob Welsh era was even better, but it wouldda been nuttin without Christine.
Miss you girl,
Chris
Sometimes OT is OK, especially on a Staturday morning ☕
Happy Holidays to all -
Jan
Happy Holidays to all -
Jan
Demonstration circuit showing that you can have more than one integrator in the loop, which has the advantage that you can get more loop gain at audio frequencies for a given loop bandwidth: https://www.diyaudio.com/community/...-of-1-000-000-000-000-000-p.16936/post-197599
The practice of using more than two dominant poles (~= integrating stages) has never become popular for valve amplifiers, because they tend to oscillate when the valves haven't fully heated up yet. It's used all over the place in sigma-delta modulator design and class-D amplifiers.
Happy Christmas tomorrow!
The practice of using more than two dominant poles (~= integrating stages) has never become popular for valve amplifiers, because they tend to oscillate when the valves haven't fully heated up yet. It's used all over the place in sigma-delta modulator design and class-D amplifiers.
Happy Christmas tomorrow!
Merry Christmas Everybody!
From the beginning of Christmas Day at locations Immediately West of the Dateline,
To the end of Christmas Day at locations Immediately East of the Dateline . . .
There are 48 Hours.
And the beginning of everyday just West of the dateline, to the ending of the same day just East of the dateline is 48 Hours.
Everything is just what it seems to be, until you find out it is different than what you thought it to be.
From the beginning of Christmas Day at locations Immediately West of the Dateline,
To the end of Christmas Day at locations Immediately East of the Dateline . . .
There are 48 Hours.
And the beginning of everyday just West of the dateline, to the ending of the same day just East of the dateline is 48 Hours.
Everything is just what it seems to be, until you find out it is different than what you thought it to be.
I wonder why isn't this whole thread at the same adress with this topic: https://www.diyaudio.com/community/threads/bruno-putzeys-paper-on-negative-feedback.306022/
So many common lines in that paper with your older than a decade insights...did Bruno pay anything to put his name on your studies?
I think I see a relatively simple way to fix FFBP distortion and maybe even some other issues. Conventional N Fdbk adds in a small (inverted) distortion (error term) to the input (plus a gob of inverted signal), so it takes huge forward gain to get rid of a large % of the distortion. We need to get rid of all (or most) of the distortion on the 1st pass to avoid FFBP dist.
Notice that multiplying the input by a number slightly less or slightly greater than 1 can have similar effect as long as the dist. factor is small. So we take the difference between the input and the scaled down output for the control factor (the error) and apply it to the tail current control of a differential stage at the input of the amplifier. Just use a CCS tail with a reference that can be modulated slightly. The amp no longer needs a large gain since there is no inverted signal injection to reduce the gain. A pot in the Fdbk path can adjust the control factor for the best distortion null.
There is a problem however in that the error detected shrinks away as the output cleans up, so some gain may be needed in the Fdbk path to detect the reduced error and maintain control. This could easily go unstable. So I suggest something like an automatic gain control in the error Fdbk path that seeks to boost the error gain up till the error control operates at some set (detected error AC) level proportional to input signal average level. Admittedly somewhat rudimentary. Maybe a diode function gen. in there to set the error correction path gain versus input signal AC.
Notice that multiplying the input by a number slightly less or slightly greater than 1 can have similar effect as long as the dist. factor is small. So we take the difference between the input and the scaled down output for the control factor (the error) and apply it to the tail current control of a differential stage at the input of the amplifier. Just use a CCS tail with a reference that can be modulated slightly. The amp no longer needs a large gain since there is no inverted signal injection to reduce the gain. A pot in the Fdbk path can adjust the control factor for the best distortion null.
There is a problem however in that the error detected shrinks away as the output cleans up, so some gain may be needed in the Fdbk path to detect the reduced error and maintain control. This could easily go unstable. So I suggest something like an automatic gain control in the error Fdbk path that seeks to boost the error gain up till the error control operates at some set (detected error AC) level proportional to input signal average level. Admittedly somewhat rudimentary. Maybe a diode function gen. in there to set the error correction path gain versus input signal AC.
Last edited:
Could just use an integrator of detected distortion to set the error correction path gain. The detected error will go down with correction acting but the integrator will increase gain until no error is detected whatsoever. There is still a problem of how to reduce the integrated up control level when the input signal amplitude goes down. More thought required. And the speed of the gain changes needs to follow input variation rates. We don't have a polarity change available from detected AC error level, but we could have a derivative of sorts.. More work needed.
Maybe error correction path gain just proportional to absolute value of the incoming signal. Except the detected error goes down as the correction activates.
Maybe intractable.
Maybe error correction path gain just proportional to absolute value of the incoming signal. Except the detected error goes down as the correction activates.
Maybe intractable.
Last edited:
That's not surprising, as neither Bruno Putzeys in his article nor I on this forum wrote anything about feedback that had not been known for many decades. I think the error signal overshoot story was missing from his article, hence my comment http://linearaudio.net/sites/linearaudio.net/files/volume1ltemvdg.pdf
For some reason I can't open that link or get a download...
The overshoot signal story is not missing in the 2012 Burning amp presentation though.Anyway...as much as I hate purifi's overpricy marketing technique, his early observations made sense unlike his stance of two years ago when he chose to hop into the same boat with PL kinda rhetorique...It seems that this happens with almost all engineers who choose their hobby as financial mean too...
Here's the exact reason why I'll never be a fan of him:
https://www.diyaudio.com/community/...ollowing-power-cut.374965/page-2#post-7212733
The overshoot signal story is not missing in the 2012 Burning amp presentation though.Anyway...as much as I hate purifi's overpricy marketing technique, his early observations made sense unlike his stance of two years ago when he chose to hop into the same boat with PL kinda rhetorique...It seems that this happens with almost all engineers who choose their hobby as financial mean too...
Here's the exact reason why I'll never be a fan of him:
https://www.diyaudio.com/community/...ollowing-power-cut.374965/page-2#post-7212733
Last edited:
Try going to https://linearaudio.net/letters and scrolling down to Vol 1-2.
The "The F-word" article of Bruno Putzeys appeared in Linear Audio volume 1 from April 2011, so it is older than that 2012 presentation.
The "The F-word" article of Bruno Putzeys appeared in Linear Audio volume 1 from April 2011, so it is older than that 2012 presentation.
Last edited:
While i get instant download link on " here" in the first volumes , apparently i get a blank page in those older volumes .It might be that I'm accessing it from a phone...I'll try later from a computer.It might also be because I live in Romania and some links don't get active here...or just that the links are empty of content... I get though that you had lengthy conversations with Bruno a decade or so ago so you both got to use some common phrases.It struck me that both of you used identical phrases here and there...
Tried the link for vol 1-2 again, but I get a blank page.First volumes from the top down work though...I'll try on a computer soon.
Here's a question for folk with (much, but that's not that difficult) better maths skills than I: the FFBP solution is also an approximation, even in full infinite series drag, because it ignores propagation time. Could this be related to the triode-inherent-feedback question and the common observation that simple single stage degeneration is somehow practically different from a slightly longer loop?
Or, to frame it another way, how important are the very tiny arrival time differences from the loop into the integrator? And then, how important is the integrator's time constant? (These two are not related!) Relevant to valve amplifiers with their good (and monotonic!) distortion levels and their poor high frequency behavior. But maybe generally.
Much thanks, as always,
Chris
Or, to frame it another way, how important are the very tiny arrival time differences from the loop into the integrator? And then, how important is the integrator's time constant? (These two are not related!) Relevant to valve amplifiers with their good (and monotonic!) distortion levels and their poor high frequency behavior. But maybe generally.
Much thanks, as always,
Chris
I believe that the difference between the very tight internal triode anode to grid feedback and a larger loop feedback is that the tighter and smaller the loop, the smaller the phase shift, and the less impact on stability.
Those are practical differences, I don't see any principle difference.
Jan
Those are practical differences, I don't see any principle difference.
Jan