Would this g2 feedback technique also work if the triode stage was a concertina instead of a cathode follower? Or would the feedback loop unbalance the concertina?
Oops, found my own answer in this post:
http://www.diyaudio.com/forums/tubes-valves/120888-magic-phase-splitter-simplep-p.html
http://www.diyaudio.com/forums/tubes-valves/120888-magic-phase-splitter-simplep-p.html
"Or would the feedback loop unbalance the concertina? "
The screen load is normally quite low Z and would seriously unbalance the concertina. Could put a follower in between to drive the screen. Would look a bit odd with one follower driving another though. Could try two concertina stages, with one driving the screen and the other doing the splits.
More risque, would be to put a CCS load on the pentode. Since its plate and screen would be tracking, the screen would be drawing near to constant current then (constant fraction of constant plate current). This would look like a hi-Z load then at the screen. Whether this would come out accurate enough to not AC unbalance the concertina would require some testing. But the DC screen current (sourced by the concertina then) would impose a noticeable DC offset in the concertina's top load resistor, eating up more headroom. All in all, probably not a good idea that way.
I recall Tubelab did something a while back using feedback to the screen from a concertina, but I think the intention there was for LF DC servoing of the concertina op point.
The screen load is normally quite low Z and would seriously unbalance the concertina. Could put a follower in between to drive the screen. Would look a bit odd with one follower driving another though. Could try two concertina stages, with one driving the screen and the other doing the splits.
More risque, would be to put a CCS load on the pentode. Since its plate and screen would be tracking, the screen would be drawing near to constant current then (constant fraction of constant plate current). This would look like a hi-Z load then at the screen. Whether this would come out accurate enough to not AC unbalance the concertina would require some testing. But the DC screen current (sourced by the concertina then) would impose a noticeable DC offset in the concertina's top load resistor, eating up more headroom. All in all, probably not a good idea that way.
I recall Tubelab did something a while back using feedback to the screen from a concertina, but I think the intention there was for LF DC servoing of the concertina op point.
"http://www.diyaudio.com/forums/tubes...simplep-p.html "
Yes, that's it. The 1 uF cap sets the gain very high for DC, so servo's the Op point of the pentode plate and concertina grid. The dual 100K resistor divider in the feedback path sets the AC closed loop gain at the concertina as 2X the g2/g1 Mu of the pentode. Good idea to include a stopper resistor on the pentode g2 and Mosfet gate.
Yes, that's it. The 1 uF cap sets the gain very high for DC, so servo's the Op point of the pentode plate and concertina grid. The dual 100K resistor divider in the feedback path sets the AC closed loop gain at the concertina as 2X the g2/g1 Mu of the pentode. Good idea to include a stopper resistor on the pentode g2 and Mosfet gate.
Last edited:
If you replace the pentode with a cascode instead, the "screen" input for feedback from the concertina becomes Hi-Z, so no extra follower is needed.
Problem then is the B+ voltage eaten up by the cascode to drive a concertina directly. I would suggest, using a Mosfet for the top of the cascode. A Mosfet would work well for the concertina stage too. Probably incite Mosphobia by then though.
Problem then is the B+ voltage eaten up by the cascode to drive a concertina directly. I would suggest, using a Mosfet for the top of the cascode. A Mosfet would work well for the concertina stage too. Probably incite Mosphobia by then though.
As I ponder different feedback methods, another one has popped up. I have not searched so if this has been covered, my apologies.
In our general avoidance of pentodes, we may be missing a better way to introduce feedback through the "other" linear element. The Supressor Grid. It was used for AVC in the past as an easier and more linear way to introduce compression than using other methods. Supressor grid modulation was used in some AM transmitters. It has been mentioned that the screen grid is less than linear for introducing feedback.
Could this be used as a linear way to use the UL feedback tap? This grid may need to be negatively biased so a direct UL connection might not be used. Could some other feedback loop be used to better advantage than the more commonly seen ones? Plate to g3 without the need for a high gm driver or input transformer as in plate to g1? Voltage feedback from a pot across the output transformer primary, giving a variable % feedback?
I have not thought these through, just tossing out suggestions. The more experienced of you think way faster than me.
In our general avoidance of pentodes, we may be missing a better way to introduce feedback through the "other" linear element. The Supressor Grid. It was used for AVC in the past as an easier and more linear way to introduce compression than using other methods. Supressor grid modulation was used in some AM transmitters. It has been mentioned that the screen grid is less than linear for introducing feedback.
Could this be used as a linear way to use the UL feedback tap? This grid may need to be negatively biased so a direct UL connection might not be used. Could some other feedback loop be used to better advantage than the more commonly seen ones? Plate to g3 without the need for a high gm driver or input transformer as in plate to g1? Voltage feedback from a pot across the output transformer primary, giving a variable % feedback?
I have not thought these through, just tossing out suggestions. The more experienced of you think way faster than me.
I started a new thread on using G3.
http://www.diyaudio.com/forums/tubes-valves/160240-suppresor-grid-used-feedback.html#post2066647
http://www.diyaudio.com/forums/tubes-valves/160240-suppresor-grid-used-feedback.html#post2066647
The coarsely spaced suppressor grid has next to no effect in most tubes except holding down secondary emission.
"Dual control" pentodes have a fine mesh g3 that can be used for control, but typically has only around a gm of 500 or so. Take a look at 6LE8 or 6BV11, 6GY6, 6HZ6, 6AS6, 6DB6, 6DT6 data sheets. Also, some curves here:
http://www.diyaudio.com/forums/tubes-valves/159892-nice-triode-5.html#post2062926
http://scottbecker.net/tube/sheets/135/6/6LE8.pdf
The g2 linearity is no worse than for triode internal plate feedback. Just hard to drive the low Z g2. My earlier point about g2 linearity was that if one is going to use NFBK, one would normally use the most linear (versus the input signal already on g1) control points: g1 or cathode for applying that feedback. The only advantage of g2 feedback is that one can preserve "triode sound" even with gobs of NFBK in order to get low output Z for the speaker. Ie, if one wants to get SE sound with solid bass response, Mr g2 is your man.
Don
"Dual control" pentodes have a fine mesh g3 that can be used for control, but typically has only around a gm of 500 or so. Take a look at 6LE8 or 6BV11, 6GY6, 6HZ6, 6AS6, 6DB6, 6DT6 data sheets. Also, some curves here:
http://www.diyaudio.com/forums/tubes-valves/159892-nice-triode-5.html#post2062926
http://scottbecker.net/tube/sheets/135/6/6LE8.pdf
The g2 linearity is no worse than for triode internal plate feedback. Just hard to drive the low Z g2. My earlier point about g2 linearity was that if one is going to use NFBK, one would normally use the most linear (versus the input signal already on g1) control points: g1 or cathode for applying that feedback. The only advantage of g2 feedback is that one can preserve "triode sound" even with gobs of NFBK in order to get low output Z for the speaker. Ie, if one wants to get SE sound with solid bass response, Mr g2 is your man.
Don
I should qualify my above comment about g2 linearity. As long as the plate V and g2 V track each other, g2 linearity is close to triode linearity. This is possible when g2 current is a constant fraction of plate current. If the plate does not track g2, then screen current distortion will be significant (screen current being subtracted from cathode current to leave plate current) and the plate output becomes distorted.
g3 modulation works by deflecting plate current back to the g2 grid. This may not be very linear due to the electron velocity spread from passing earlier grids on the way. I will try this on the curve tracer to see. In any case, g2 gets rather hot when a lot of plate current gets bounced back to it.
g3 modulation works by deflecting plate current back to the g2 grid. This may not be very linear due to the electron velocity spread from passing earlier grids on the way. I will try this on the curve tracer to see. In any case, g2 gets rather hot when a lot of plate current gets bounced back to it.
Here is a VCA circuit that shows the principle anyway. If I knew spice I would give it a try but I've not mastered that program yet.
Audio Synthesis via Vacuum Tubes
Audio Synthesis via Vacuum Tubes
Can this scheme be modified if I need to add gnfbk at the cathode or does it just fall apart?
- Status
- Not open for further replies.