Hi All
Having some thought about the using of screen grid as input instead of control grid of a pentode:
Pentode or beam power tube can be operated as ultra linear mode when screen grid tapped to the plate load at certain %, say 20%,43% or more until 100% which becomes triode connection. The screen grid should be out of phase to the plate output and so constitute a feedback mechanism.....this is what I think, since the result of ultraliner connection is more linear than typical pentode connection. If the % of the plate output is applied correctly to the screen grid, any voltage feedback seems no need to implement neither.
What if we use the screen grid as the signal input instead of the control grid.....
I guess that the signal input to the screen grid will possibly be the same mechanism of ultra linear connection because of the out of phase of the output from the plate, some cancellation possibly arises and this should improve the linearity just like ultra linear connection.
But due to the positive bias of a screen grid and impedance driving problem, a cathode follower should be used to drive the screen grid and this will make things complicated.
Anybody has idea regarding the linearity of screen grid input? Please correct me for my understanding, if any.
Thanks
Tim Chow
Having some thought about the using of screen grid as input instead of control grid of a pentode:
Pentode or beam power tube can be operated as ultra linear mode when screen grid tapped to the plate load at certain %, say 20%,43% or more until 100% which becomes triode connection. The screen grid should be out of phase to the plate output and so constitute a feedback mechanism.....this is what I think, since the result of ultraliner connection is more linear than typical pentode connection. If the % of the plate output is applied correctly to the screen grid, any voltage feedback seems no need to implement neither.
What if we use the screen grid as the signal input instead of the control grid.....
I guess that the signal input to the screen grid will possibly be the same mechanism of ultra linear connection because of the out of phase of the output from the plate, some cancellation possibly arises and this should improve the linearity just like ultra linear connection.
But due to the positive bias of a screen grid and impedance driving problem, a cathode follower should be used to drive the screen grid and this will make things complicated.
Anybody has idea regarding the linearity of screen grid input? Please correct me for my understanding, if any.
Thanks
Tim Chow
Last edited:
Are you sure those curves are for screen grid drive? They look more like "ultralinear" or grid 1 drive (except for the big grid voltages shown). [ Oh, I see what the problem is now. The grid 1 is being held negative there, which causes big kinks. Not a good idea. Grid 1 should be 0V. Grid 2 should swing slightly negative to shut the tube off at high plate V, not grid 1 ]
Here are some pics of a Matsushita 40KG6 in screen grid drive, in "Crazy drive" and in grid 1 drive.
50V/div Horiz, 50 mA/div Vert.
I also have a Sylvania 40KG6 here. I don't recall it looking much different on the tracer, but I could re-check that.
Attachments
Last edited:
Thanks for all reply.
I also guess that screen grid input should be an ultra linear mode but only a very small ratio applied and should be regarded as a negative feed forward. It still can improve the kinks sections of the plate curve even it just small. As refer to smoking-amp's curve for screen input, the gm seems evenly spread over, this should be a linearity indication.
Another added bonus I think for screen input is the frequency response possibly reached at Mhz (I guess....) because now the control grid can be grounded to short circuit the stray capacitance and the plated impedance also improved as compared to pentode mode.
But is it sound excellent as compared to triode or traditional UL amp if negative feedback does not apply.......
I also guess that screen grid input should be an ultra linear mode but only a very small ratio applied and should be regarded as a negative feed forward. It still can improve the kinks sections of the plate curve even it just small. As refer to smoking-amp's curve for screen input, the gm seems evenly spread over, this should be a linearity indication.
Another added bonus I think for screen input is the frequency response possibly reached at Mhz (I guess....) because now the control grid can be grounded to short circuit the stray capacitance and the plated impedance also improved as compared to pentode mode.
But is it sound excellent as compared to triode or traditional UL amp if negative feedback does not apply.......
Hello smoking-amp,
in 1998 Bob Danielak published screen drive curves for 6KG6 / EL509 with g1 prebiased to -33V in an application note for Svetlana. They looked so nice that I got sceptical and plotted (quite some samples, same and different brands) using the same conditions. As it turned out, the curves in the app note were no real plots but simulations.
You write that prebiasing g1 negatively is not a good idea. While this may be true for measuring purposes to show linearity of screen drive alone, it is a very good idea to do so to ensure stable operation conditions in real world amps.
Tim de Paravincini (google his EAR859 screen drive SE amp, circuit was published 09/1994 by Ken Kessler) held a lecture at ETF in Biezenmortel / NL (anno 2008, IIRC) about circuits using sweep tubes like EL509 for audio (including screen drive), nowadays (think 10 years back) quality and production tolerances of new production xKG6/EL509, and so on.
He insisted that g1 had to be prebiased negatively because these fat sweep tubes simply like to run away thermally, especially when operated close to their maximum Pa like in single ended amp circuits (this was not a problem within the operation conditions they were constructed for). And yes, he saw my plots.
By the way, are you doing your plots using rather fast sweeps? If so, please be aware that the tube will not heat up as it would if run f.e in rather constant Pa SE conditions. This usually makes a huge difference in real world concearning power tubes - even a with a generic EL34 or 6L6. Temperature coefficients may be positive or negative or close to zero, depending on construction, brand, charge, whatelse. In real world amps you have to deal with such issues.
Kind regards, Tom Schlangen
in 1998 Bob Danielak published screen drive curves for 6KG6 / EL509 with g1 prebiased to -33V in an application note for Svetlana. They looked so nice that I got sceptical and plotted (quite some samples, same and different brands) using the same conditions. As it turned out, the curves in the app note were no real plots but simulations.
You write that prebiasing g1 negatively is not a good idea. While this may be true for measuring purposes to show linearity of screen drive alone, it is a very good idea to do so to ensure stable operation conditions in real world amps.
Tim de Paravincini (google his EAR859 screen drive SE amp, circuit was published 09/1994 by Ken Kessler) held a lecture at ETF in Biezenmortel / NL (anno 2008, IIRC) about circuits using sweep tubes like EL509 for audio (including screen drive), nowadays (think 10 years back) quality and production tolerances of new production xKG6/EL509, and so on.
He insisted that g1 had to be prebiased negatively because these fat sweep tubes simply like to run away thermally, especially when operated close to their maximum Pa like in single ended amp circuits (this was not a problem within the operation conditions they were constructed for). And yes, he saw my plots.
By the way, are you doing your plots using rather fast sweeps? If so, please be aware that the tube will not heat up as it would if run f.e in rather constant Pa SE conditions. This usually makes a huge difference in real world concearning power tubes - even a with a generic EL34 or 6L6. Temperature coefficients may be positive or negative or close to zero, depending on construction, brand, charge, whatelse. In real world amps you have to deal with such issues.
Kind regards, Tom Schlangen
Last edited:
Sure, why not, it obviously worked for him with the tube charge runs he got at St Petersburg 🙂 Moreover, you certainly noticed, that additional DC feedback is used in the EAR859 to stabilize operating conditions?
Note, g1 @ -33V under SE screen drive conditions for EL509 was suggested by Mr Danielak, working for Svetlana, not by TvP, working for his own.
Tom
Note, g1 @ -33V under SE screen drive conditions for EL509 was suggested by Mr Danielak, working for Svetlana, not by TvP, working for his own.
Tom
Last edited:
In St Petersburg he probably got 6P45S tubes that are compatible with EL509 in sweep duty, but still different. They are frame grid tubes.
The curves I traced are done at 60 Hz on a TEK 576 curve tracer. So I can see where steady class A thermal issues could be another matter. With a current sensing servo, I would think that would be sufficient for class A.
Notice I mentioned that (using 0V on grid 1), grid 2 drive should be able to go negative sufficiently to counter 2X B+ on the plate for full turnoff at opposite peaks.
That should still work for class A bias control on grid 2 with the servo control if the tube can still be turned off at peak 2X B+.
For that -1.5V on g1 on the EAR 859, one needs approx. - (2 x B+)/(Rp gm1) on grid1 to turn the tube off at 2X B+ with 0V on grid2. So approx. - (B+)/30
likely where the -33V figure comes from.
The EAR 859 is probably not holding its tubes off completely at opposite peaks. (re-ignition at 2X B+)
Unless the grid 2 drive goes negative also.
Similarly, for 0V on grid1, one needs approx. - (2 x B+)(Mu_int)/(Rp gm1) available on grid2 to turn the tube off at 2X B+.
.
Notice I mentioned that (using 0V on grid 1), grid 2 drive should be able to go negative sufficiently to counter 2X B+ on the plate for full turnoff at opposite peaks.
That should still work for class A bias control on grid 2 with the servo control if the tube can still be turned off at peak 2X B+.
For that -1.5V on g1 on the EAR 859, one needs approx. - (2 x B+)/(Rp gm1) on grid1 to turn the tube off at 2X B+ with 0V on grid2. So approx. - (B+)/30
likely where the -33V figure comes from.
The EAR 859 is probably not holding its tubes off completely at opposite peaks. (re-ignition at 2X B+)
Unless the grid 2 drive goes negative also.
Similarly, for 0V on grid1, one needs approx. - (2 x B+)(Mu_int)/(Rp gm1) available on grid2 to turn the tube off at 2X B+.
.
Last edited:
- Status
- Not open for further replies.