Connecting g3 to the plate (also g2) for triode mode will give just a bit lower plate impedance than g2 only to plate for triode mode. One thing possibly useful is that less current will be reflected back to g2 with g3 positive, so go on thru to the plate instead, for less g2 heating in triode mode. (Usually g3 is so sparsely spaced as to not make any appreciable difference. ) 6CL6 is a normal gridded g3 pentode.
Using triode configuration for a cathode follower there, and then connecting g3 to the plate as well, makes no sense at all however. One wants a pentode configuration (g2 bootstrapped above the cathode by a Zener) to get the lowest cathode follower output impedance. Maybe they didn't have Zener diodes back in the 50s.
Using triode configuration for a cathode follower there, and then connecting g3 to the plate as well, makes no sense at all however. One wants a pentode configuration (g2 bootstrapped above the cathode by a Zener) to get the lowest cathode follower output impedance. Maybe they didn't have Zener diodes back in the 50s.
It doesn't make much difference as g3 rarely has more than a teensy amount of gain; it would take hundreds of volts, thousands even, to make a difference. In any case, that difference is a linear one: a proportional offset to Vg1.
(The effective Vg1 seen by the cathode, and therefore what sets cathode current through the diode equation, is the sum of all electrode voltages, divided by their amplification factors. That is, Ik = f(Vg1 + Vg2 / mu2 + Vg3 / mu3 + ...), where f(x) is the diode transfer function, a 3/2 power law. mu_g2g1 is the most important, which is why, when g2 is tied to plate, the cathode current varies so strongly with plate current: plate resistance. mu_g3g1 is very small indeed, and mu_pg1 even smaller, which is why g3 has so little effect, and why pentode plate resistance is extremely high, away from the saturation region.)
No electrodes have a problem with that connection, because once the electron beam passes g2 (where the effective triode-mode plate is), the electric field is about zero, and the electrons drift freely (no acceleration) until they strike the anode. A small number will strike g2 and g3, proportional to their cross sectional areas. The plate carries the majority of current, because the areas of the other electrodes are small.
I don't recommend this for dual-control type pentodes (or even-more-grid tubes), where g3 does have a significant cross section and therefore will dissipate some power.
Tim
(The effective Vg1 seen by the cathode, and therefore what sets cathode current through the diode equation, is the sum of all electrode voltages, divided by their amplification factors. That is, Ik = f(Vg1 + Vg2 / mu2 + Vg3 / mu3 + ...), where f(x) is the diode transfer function, a 3/2 power law. mu_g2g1 is the most important, which is why, when g2 is tied to plate, the cathode current varies so strongly with plate current: plate resistance. mu_g3g1 is very small indeed, and mu_pg1 even smaller, which is why g3 has so little effect, and why pentode plate resistance is extremely high, away from the saturation region.)
No electrodes have a problem with that connection, because once the electron beam passes g2 (where the effective triode-mode plate is), the electric field is about zero, and the electrons drift freely (no acceleration) until they strike the anode. A small number will strike g2 and g3, proportional to their cross sectional areas. The plate carries the majority of current, because the areas of the other electrodes are small.
I don't recommend this for dual-control type pentodes (or even-more-grid tubes), where g3 does have a significant cross section and therefore will dissipate some power.
Tim
As I just said, the electric field beyond the screen is zero. So the force is zero, and no more electrons strike that grid than its size allows
When the grid is zero, the beam decelerates some as it passes the screen, but the suppressor grid is too widely spaced to cause cutoff. Rather, the beam is pinched away from the suppressor wires. It then accelerates towards the plate, striking at the usual velocity.
Tim
When a pentode is triode connected there is no need for a supressor grid at all because g2 and anode is connected together and there is no potential difference between g2 and anode. So in triode mode g3 has no use. Therefore it is logical to connect g2, g3 and anode together.
In case of pentode connection, when g3 is connected to anode, then a tetrode with anode current kink is formed.
The quite accurate comment in this thread so far which stood out to me.
Why do they often wire the pentode as a triode when used as a cathode follower?
If you simply wire it as a pentode with a voltage divider to the screen (g2), as you normally would (for a pentode gain stage) then you have negative feedback when you consider the cathode to screen signal voltage. The result of that is that it is the triode characteristics of the tube which apply, not the pentode characteristics.
To get the pentode characteristic back you need to have no ac signal between cathode and the screen. Hence bootstrapping. AC connecting the two via cap (or zener). This adds a small bit of complexity, so most designers simply wire it as a triode. Pentode Cathode Followers (with bootstrapped screen) are quite a bit more linear and offer lower output impedance than triode. My view is that the small amount of extra effort and the couple of extra components required is worth it.
Next step after that is a CCS load, but that is another thread.
Cheers,
Ian
The transconductance doesn't change, because when used as a cathode follower the output gets both the anode and g2 current whether it is wired as pentode or triode. The difference is the anode impedance: lowish for triode wiring, high for pentode wiring. The anode impedance appears in parallel with the output load, so pentode CF wiring (with a bootstrapped g2) gives slightly higher output impedance but also higher gain.
Think you're right, DF96. In a cathode follower, where anode & g2 each has low impedance connection to B+ or thereabouts, the valve can't know whether it is a triode or a pentode, it will just have an operating point, and gm determined by that operating point.
In which case, triode strapping or not makes little difference to output impedance in a cathode follower made from a pentode.
However because pentode valve types can have significantly higher gm than native triode types, pentode types should offer lower output impedance in cathode followers.
LD
In which case, triode strapping or not makes little difference to output impedance in a cathode follower made from a pentode.
However because pentode valve types can have significantly higher gm than native triode types, pentode types should offer lower output impedance in cathode followers.
LD
"The difference is the anode impedance: lowish for triode wiring, high for pentode wiring. The anode impedance appears in parallel with the output load, so pentode CF wiring (with a bootstrapped g2) gives slightly higher output impedance but also higher gain."
I think this is not quite correct. When g1 pulls up +, to drive the cathode load positive, the effect of the decreasing plate to cathode V is to -decrease- the current draw slightly. (not much however, due to actual pentode Rp being up around 150K Ohms for 6CL6) So output Z is increased slightly (from 1/gm) for triode and pentode. (or gm1 effectively dropped by 1/150K, about 7 uMho)
Connecting g3 to the plate too would have a similar effect, but maybe lose another 50 uMho off gm1. And g2 to plate (so triode mode) would lose another 1/Mu or about 550 uMho. (6CL6 gm1 starting with 11000 uMho at 30 mA according to the datasheet)
So 6CL6 pentode follower Zout 1/(11000-7) = 91 Ohms (not including pull down R in parallel)
and 6CL6 triode follower Zout 1/(11000-7-50-550) = 96 Ohms (not including pull down R in parallel)
The 6CL6 in the amplifier mentioned is running at much less than 30 mA however, in order to drive the large class AB2 output grid voltage swing (6CL6 dissipation limited). (assuming a similar design to the Fisher 55A, which has a 210 Ohm follower output Z)
So the effects of g2, g3 and plate are much greater with gm1 reduced at low current. (210 Ohm output Z would be ballpark 4800 gm1 instead of 11000)
see Driver Section:
The End of an Era: The Fisher 55-A | Page 2 | Audiokarma Home Audio Stereo Discussion Forums
210 Ohms Zout is somewhat problematic for driving the output grids positive (high order distortions). The PL802 or a 12GN7 frame grid video tube would work better there (about 28 Ohms native Zout). Or even better, a Mosfet follower (around a few Ohms Zout).
I think this is not quite correct. When g1 pulls up +, to drive the cathode load positive, the effect of the decreasing plate to cathode V is to -decrease- the current draw slightly. (not much however, due to actual pentode Rp being up around 150K Ohms for 6CL6) So output Z is increased slightly (from 1/gm) for triode and pentode. (or gm1 effectively dropped by 1/150K, about 7 uMho)
Connecting g3 to the plate too would have a similar effect, but maybe lose another 50 uMho off gm1. And g2 to plate (so triode mode) would lose another 1/Mu or about 550 uMho. (6CL6 gm1 starting with 11000 uMho at 30 mA according to the datasheet)
So 6CL6 pentode follower Zout 1/(11000-7) = 91 Ohms (not including pull down R in parallel)
and 6CL6 triode follower Zout 1/(11000-7-50-550) = 96 Ohms (not including pull down R in parallel)
The 6CL6 in the amplifier mentioned is running at much less than 30 mA however, in order to drive the large class AB2 output grid voltage swing (6CL6 dissipation limited). (assuming a similar design to the Fisher 55A, which has a 210 Ohm follower output Z)
So the effects of g2, g3 and plate are much greater with gm1 reduced at low current. (210 Ohm output Z would be ballpark 4800 gm1 instead of 11000)
see Driver Section:
The End of an Era: The Fisher 55-A | Page 2 | Audiokarma Home Audio Stereo Discussion Forums
210 Ohms Zout is somewhat problematic for driving the output grids positive (high order distortions). The PL802 or a 12GN7 frame grid video tube would work better there (about 28 Ohms native Zout). Or even better, a Mosfet follower (around a few Ohms Zout).
Last edited:
Umm, on second thought, it looks like one has to take into account the static plate resistance for Zout as well, since it shunts the output.
So the triode case acts like it has a lower static resistance in parallel with the 1/gm1 drive, but less effective gm to drive that. So less voltage and current swing out, but similar Zout to the pentode. (essentially g2 to plate lowers Zout by the same amount (from shunting) as it increases Zout via reducing the effective gm1) (g3 to plate would just incrementally continue the same effect)
The Fisher Amps are short on driver/follower stage current drive to achieve the voltage swing needed for class AB2 of the output tubes. So a pentode like follower would be better there. A Mosfet would be best.
So the triode case acts like it has a lower static resistance in parallel with the 1/gm1 drive, but less effective gm to drive that. So less voltage and current swing out, but similar Zout to the pentode. (essentially g2 to plate lowers Zout by the same amount (from shunting) as it increases Zout via reducing the effective gm1) (g3 to plate would just incrementally continue the same effect)
The Fisher Amps are short on driver/follower stage current drive to achieve the voltage swing needed for class AB2 of the output tubes. So a pentode like follower would be better there. A Mosfet would be best.
Last edited:
Thanks, all very interesting. I've wondered about AB2 and positive g1 for the output pair, but been dissuaded by knowing neither likely g1 current flow, nor likely max g1 power dissipation for usual output valve types ? That thread has some good insight, but any idea where to find out more about g1 limiting values for common valve types ?
Thx
Thx
Some tubes are spec'd specifically for grid 1 positive drive (like 4D32). And some (seemingly unlikely) tube datasheets will give some data on that. 6CB5 for example:
https://frank.pocnet.net/sheets/084/4/4D32.pdf
4D32 specss 50W for plate, 10 Watt for g2 (probably has a cooling fin), and 0.75 Watt for g1
https://frank.pocnet.net/sheets/137/6/6CB5.pdf
Lacking any data, I would guess maybe 1/10 to 1/20 of the g2 Watts spec might be safe for grid 1. (some big tubes are only rated for less than 1 Watt on grid 1) Check the tube for a cooling fin up top, on grid 1, that would be a good sign. (the cooling fin might be for g2 however, have to distinguish) And/or two base pins connected to grid 1 (on each grid support rod) is another good sign. (connect heavy wire to those socket pins for cooling)
6HJ5 is a TV sweep tube with a big cooler fin up top on grid 1 and it has the double base pins for grid 1. (nothing in the data sheet about positive grid 1 however) Looks like quite a few TV sweeps have the double grid 1 base pins. 21HB5A has the double base pins and -some- of the tubes have the grid 1 cooler. RCA 21LG6A similarly.
Crazy Drive (or Twin Drive) (g2 driven + and g1 driven + via a resistor) can operate with less than 0.1 Watts on grid 1. And using TV Sweep tubes, may use a similar drive V amplitude for the g2 as a typical audio tube would use for g1 drive in AB2 mode. (its way more linear and efficient too)
.
https://frank.pocnet.net/sheets/084/4/4D32.pdf
4D32 specss 50W for plate, 10 Watt for g2 (probably has a cooling fin), and 0.75 Watt for g1
https://frank.pocnet.net/sheets/137/6/6CB5.pdf
Lacking any data, I would guess maybe 1/10 to 1/20 of the g2 Watts spec might be safe for grid 1. (some big tubes are only rated for less than 1 Watt on grid 1) Check the tube for a cooling fin up top, on grid 1, that would be a good sign. (the cooling fin might be for g2 however, have to distinguish) And/or two base pins connected to grid 1 (on each grid support rod) is another good sign. (connect heavy wire to those socket pins for cooling)
6HJ5 is a TV sweep tube with a big cooler fin up top on grid 1 and it has the double base pins for grid 1. (nothing in the data sheet about positive grid 1 however) Looks like quite a few TV sweeps have the double grid 1 base pins. 21HB5A has the double base pins and -some- of the tubes have the grid 1 cooler. RCA 21LG6A similarly.
Crazy Drive (or Twin Drive) (g2 driven + and g1 driven + via a resistor) can operate with less than 0.1 Watts on grid 1. And using TV Sweep tubes, may use a similar drive V amplitude for the g2 as a typical audio tube would use for g1 drive in AB2 mode. (its way more linear and efficient too)
.
Last edited:
A decrease in valve current means less current to drive the load, which means less gain. Looking at it another way, a decrease in valve current is equivalent to an increase in load current - as they go in opposite directions.smoking-amp said:
As I said, the anode impedance appears in parallel with the load. You can regard it as part of the load or part of the CF, whichever is most convenient but the outcome is the same.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.