Think of this thread as a 'many ways to skin a cat' thing...
Apart from the regular pentode, ultralinear, and triode connections, oft mentioned, there are other laternative ones, of course, not practicable for all pentodes, but certainly worth mentioning.
I am aware that triodes reign supreme in audio
but there are many qualified people on this forum that could shed some much needed light on the use of pentodes, apart from the uaual output stages and current sources.
Things like screen grid drive were discussed (but no harm done in mentioning them again here), but more data is welcome: after all, even for triode connections, one can find two variants (g1 = control, g2 strapped to plate, g3 to catkode, as well as g2 and g3 strapped to cathode). Space charge mode I have yet to find a mention of, and I am sure there is more
Apart from the regular pentode, ultralinear, and triode connections, oft mentioned, there are other laternative ones, of course, not practicable for all pentodes, but certainly worth mentioning.
I am aware that triodes reign supreme in audio
but there are many qualified people on this forum that could shed some much needed light on the use of pentodes, apart from the uaual output stages and current sources.Things like screen grid drive were discussed (but no harm done in mentioning them again here), but more data is welcome: after all, even for triode connections, one can find two variants (g1 = control, g2 strapped to plate, g3 to catkode, as well as g2 and g3 strapped to cathode). Space charge mode I have yet to find a mention of, and I am sure there is more
Steve Bench has covered some of them here.
Basically, you have (1) g1 drive with g2, g3 held at a constant votlage, (2) g2 drive with g1, g3 held at a constant voltage, and (3) g3 drive with g1, g2 held at constant voltage. Whether the constant voltages are postive, negative or grounded increases the number of possibilities further (for instance, space charge connection is option two with g1 at positive bias, while screen drive is option two with g1 at negative bias or at 0V).
Then you start getting into the more perculiar options. You can hold all of the grids constant at various voltages and apply the signal to the cathode (cathode drive/grounded grid). Or even weirder, apply the signal to a negatively biased anode and tap the output at a positively biased grid (inverted operation). There are still probably more... and we still haven't got into applying feedback to various input points or distributed loading (ultralinear, cathode feedback etc.)
Basically, you have (1) g1 drive with g2, g3 held at a constant votlage, (2) g2 drive with g1, g3 held at a constant voltage, and (3) g3 drive with g1, g2 held at constant voltage. Whether the constant voltages are postive, negative or grounded increases the number of possibilities further (for instance, space charge connection is option two with g1 at positive bias, while screen drive is option two with g1 at negative bias or at 0V).
Then you start getting into the more perculiar options. You can hold all of the grids constant at various voltages and apply the signal to the cathode (cathode drive/grounded grid). Or even weirder, apply the signal to a negatively biased anode and tap the output at a positively biased grid (inverted operation). There are still probably more... and we still haven't got into applying feedback to various input points or distributed loading (ultralinear, cathode feedback etc.)
Thanks for your input... I forgot to add that there are probably differences in 'unconventional' usage of a regular and a beam pentode, which increases the number of considerations when connecting a given tube.
What i would love to see are some real world experiences and examples, or even opinions.
Here are a couple of examples:
Does G2 drive always have to be with positive bias, and, is negative G2 bias only used in a space charge configuration.
Can G3 drive be used in beam pentodes? Can inverted operation be used in beam pentodes, usin G3 as output?
What is the difference between using a G2-to-plate, G3-to-cathode triode connection with respect to G3 and G2 both tied to plate?
What about G2 drive with G1 at 0V, G3 connected to plate, this should be one of the triode modes...
Which tubes are suitable (or have been used in) the various configurations...
Etc...
Steve Bench and others do a good job of covering the more usual connections, but still only mention others. It would be interesting to read about them in a comparative manner, for and against, and all that
Just trying to get a good discusion going. Surely a pentode, with so much connecting options at hand, is worth at least the same consideration as a triode
PS, doesn't this topic make you go do some experiments with a curve tracer
What i would love to see are some real world experiences and examples, or even opinions.
Here are a couple of examples:
Does G2 drive always have to be with positive bias, and, is negative G2 bias only used in a space charge configuration.
Can G3 drive be used in beam pentodes? Can inverted operation be used in beam pentodes, usin G3 as output?
What is the difference between using a G2-to-plate, G3-to-cathode triode connection with respect to G3 and G2 both tied to plate?
What about G2 drive with G1 at 0V, G3 connected to plate, this should be one of the triode modes...
Which tubes are suitable (or have been used in) the various configurations...
Etc...
Steve Bench and others do a good job of covering the more usual connections, but still only mention others. It would be interesting to read about them in a comparative manner, for and against, and all that
Just trying to get a good discusion going. Surely a pentode, with so much connecting options at hand, is worth at least the same consideration as a triode
PS, doesn't this topic make you go do some experiments with a curve tracer
Hi,
In "Radio engineering handbook" by F. E Terman first published 1932 several "Special connections for conventional tubes" are mentioned, in addition to normal triode connection of pentodes, (G2 and G3 connected to anode) also another variant is shown with drive to G1 and G2 connected together and G3 connected to anode, this connection have higher mu. Also inverted connection of triodes are described, this connection was invented by Terman in 1928, (so it is not a new invention).
(For those who doesn't know who Terman was it can be mentioned that he was professor for a long time at Stanford some of his students where The Varian brothers, (inventors of the klystron) and W.R Hewlett among others.)
Regards Hans
In "Radio engineering handbook" by F. E Terman first published 1932 several "Special connections for conventional tubes" are mentioned, in addition to normal triode connection of pentodes, (G2 and G3 connected to anode) also another variant is shown with drive to G1 and G2 connected together and G3 connected to anode, this connection have higher mu. Also inverted connection of triodes are described, this connection was invented by Terman in 1928, (so it is not a new invention).
(For those who doesn't know who Terman was it can be mentioned that he was professor for a long time at Stanford some of his students where The Varian brothers, (inventors of the klystron) and W.R Hewlett among others.)
Regards Hans
A small signal pentode can be run as a triode, with g2 as the anode and the plate (and g3) grounded. Dissipation is limited to what g2 will tolerate, so it only workd well at a point where the signal is small, such as in a phono preamp.
I tried it with a 6AU6, following a tone stack incurred about 12dB of attentuation. The signal was small at that point and I didn't want to use a current-guzzling tube, so I tried the 6AU6 with g2 as anode. Results were good and the additional shielding from the grounded plate helped avoid hum.
I tried it with a 6AU6, following a tone stack incurred about 12dB of attentuation. The signal was small at that point and I didn't want to use a current-guzzling tube, so I tried the 6AU6 with g2 as anode. Results were good and the additional shielding from the grounded plate helped avoid hum.
> Does G2 drive always have to be with positive bias
You can do ANYTHING you want. But the plate's positive voltage alone is not going to suck electrons with all those negative grids in the way. If you want some output at the plate, you need enough grids positive so that the cathode feels a positive voltage field.
> is negative G2 bias only used in a space charge configuration.
Well, hold on. The so-called space-charge tubes are special, don't work like ordinary tubes, and ordinary tubes won't be too happy biased as space-charge.
In the Conventional Thermionic Valve, the cathode is hot enough to boil electrons. But it is very hard to get them to break free of the cathode surface. In addition to great heat (and magic minerals), you typically need a high voltage to get even a little current.
High voltage is awkward in car radios, but 12V power is cheap. The best-known class of space-charge tubes were made for this use. To get 10mA of plate current and room for a control grid with just 12V plate voltage, they would need a monster 10-Watt cathode. But they found another trick. Build a close-space Diode and slam the full 12V across it. It is designed to pass about 100mA. It has holes (may be a wound "grid") and about 10% or 10mA of its current misses the diode plate and wanders out where we can do something with it. It is in effect a "cathode cloud", but because these electrons are not tied to a physical cathode they move more easily. Now we can use a positive plate to attract them, a negative grid (counting the 100mA diode "plate" as G1, the control grid is G2), and maybe a screen or suppressor grid (though these may not be needed at low gain and low voltage).
Note that 90% of the "plate" power we put in is not available to the load. It is in effect extra heater power, but injected after the electrons are freed from the cathode surface.
Ordinary tubes are not made to work this way, and may fail fast with 12V*100mA= 1.2 Watts of power in a grid made for delicate control.
> a pentode, with so much connecting options at hand
The Dead Men who invented/discovered tetrodes and pentodes didn't know how to use them, and tried all possible ways. Their heirs were not stupid either, and tried nearly everything. You don't hear much about the "unusual" connections because they do nothing good, or nothing that a Standard connection does not do better.
"Better" is of course a matter of taste. Many tubes were perverted for more G1 gain at the cost of less linearity. Since G2 was not usually considered an input (though see Radiotron 3rd, or several screen-modulated AM transmitters) it is often more linear. But good G1 gain means low G2 gain, so driving G2 to full output will strain any practical driver.
I'm surprised I have not seen any "Plate-In Plate-Out" amplfiers. Voltage linearity is excellent.
You can do ANYTHING you want. But the plate's positive voltage alone is not going to suck electrons with all those negative grids in the way. If you want some output at the plate, you need enough grids positive so that the cathode feels a positive voltage field.
> is negative G2 bias only used in a space charge configuration.
Well, hold on. The so-called space-charge tubes are special, don't work like ordinary tubes, and ordinary tubes won't be too happy biased as space-charge.
In the Conventional Thermionic Valve, the cathode is hot enough to boil electrons. But it is very hard to get them to break free of the cathode surface. In addition to great heat (and magic minerals), you typically need a high voltage to get even a little current.
High voltage is awkward in car radios, but 12V power is cheap. The best-known class of space-charge tubes were made for this use. To get 10mA of plate current and room for a control grid with just 12V plate voltage, they would need a monster 10-Watt cathode. But they found another trick. Build a close-space Diode and slam the full 12V across it. It is designed to pass about 100mA. It has holes (may be a wound "grid") and about 10% or 10mA of its current misses the diode plate and wanders out where we can do something with it. It is in effect a "cathode cloud", but because these electrons are not tied to a physical cathode they move more easily. Now we can use a positive plate to attract them, a negative grid (counting the 100mA diode "plate" as G1, the control grid is G2), and maybe a screen or suppressor grid (though these may not be needed at low gain and low voltage).
Note that 90% of the "plate" power we put in is not available to the load. It is in effect extra heater power, but injected after the electrons are freed from the cathode surface.
Ordinary tubes are not made to work this way, and may fail fast with 12V*100mA= 1.2 Watts of power in a grid made for delicate control.
> a pentode, with so much connecting options at hand
The Dead Men who invented/discovered tetrodes and pentodes didn't know how to use them, and tried all possible ways. Their heirs were not stupid either, and tried nearly everything. You don't hear much about the "unusual" connections because they do nothing good, or nothing that a Standard connection does not do better.
"Better" is of course a matter of taste. Many tubes were perverted for more G1 gain at the cost of less linearity. Since G2 was not usually considered an input (though see Radiotron 3rd, or several screen-modulated AM transmitters) it is often more linear. But good G1 gain means low G2 gain, so driving G2 to full output will strain any practical driver.
I'm surprised I have not seen any "Plate-In Plate-Out" amplfiers. Voltage linearity is excellent.
PRR said:
That is perfectly understood. the question is, I suppose, how do you make a sort of triode out of a pentode, and it's mostly related to power pentodes. Small signal pentodes can be amazingly good triodes when wired in the classical G2=P, G3=C mode. The small question remains, what difference you get when G3=P...
Actually, I understand that though thanks for the excellent description.
Here's where the question comes from:
Typically in sweep tubes (and rarely in AF power pentodes) you get a Vg2) vs Ip) diagram at G1=0V or, as I have found, more often at G1 = -1V. However, in these tubes, G1 is not a flimsy thing. The question really is, biasing g1 positive staying inside the gate current limits, how low would G2 have to be biassed (and could it even go negative for some tubes?). The idea behind it is a reduction of G2 current. One would also assume that strapping G3 to plate would make a difference here tough typically we are talking about beam tubes so this may not be so clear - for instance, Vg3 max is given as 30V for a PL519.
The reason why I mentioned this as 'space charge mode' is because G1 being positive, it 'accelerates' electrons from the cathode, similar to space charge tubes.
Yes, this was really my point too - it's really a question of trying to design with tubes you have rather than theones you want
when ideas such as these are considered...Yes, like in the monode tube, a tube with just one electrode
These are also amazingly linear
> biasing g1 positive staying inside the gate current limits, how low would G2 have to be biassed (and could it even go negative for some tubes?).
VG1 and VG2 are semi-locked by the Mu, G1 to G2.
Take a typical power tube in typical use. Mu(G1-G2) is say 10. G2 is at 400V, G1=-24V gives a nice idle current, G1= 0V to -48V for a good swing.
Or: (-24V)+(+400V/10) gives a +16V field off the cathode to urge the idle current, (0V)+(+400V/10)= +40V for peak current, (-48V)+(+400V/10) +8V for near-cutoff current.
If we make G1 zero V all the time, then G2 must be 160V for idle, 400V for peak, +80V for near-cutoff.
If we make G1 +10V all the time, then G2 must be 60V for idle, 300V for peak, -20V for near-cutoff. That is not quite right: G1 is now sucking ~10mA away from the cathode, so the Plate current is probably 10mA less than with negative grid.
You could of course just set up a tube with a +/- G1 bias supply, an adjustable G2 supply and a high-volt signal, and try it on the bench. In fact when working outside the conventional techniques, that's the only way to be sure nothing odd happens. In conventional work, tube makers were careful to minimize (or document) any odd behavior, but when you do things strange you may discover strange things.
> The idea behind it is a reduction of G2 current.
In the simple pentodes with plate voltage not much lower than G2 voltage, G2 current tends to be a somewhat constant fraction of plate current. So "reducing G2 current" implies a reduction in plate current. But plate current is our precious output, which we do NOT want to reduce.
For large plate peak current with large power output and good efficiency, the plate goes very low when G2 goes very high. Large peak G2 currents seem unavoidable. Using the positive G1 trick, the G2 at the peak may be 300V instead of 400V; meanwhile we hope the plate has gone down near 50V. I don't know if getting peak G2 current down to 75% of before is much of an improvement.
When you get into the aligned-grid "beam-G3" pentodes, all bets must be off. What they have done is try to reduce G2 current to zero and G3 stray current to zero. Small changes in field conditions or factory winding cause large variations in actual current. Get too clever and you un-do the careful cancellations designed into these tubes. I suspect G2 (and maybe G3/beam-electrode) current could soar far beyond what the designer built them for. For example: in the align-grids 6L6, the negative voltage on G1 casts shadows where the G2 wires are, reducing G2 current. Make G1 positive, and what happens? Actually not a disaster: such tubes are run with positive peaks in radio work. But peak G2 current is not an issue there: it is heavily bypassed.
> how do you make a sort of triode out of a pentode
So use a triode. If power and efficiency were important goals, you would be using transistors. Or if transistors have not been invented yet, you would use 6L6 power pentodes. As far as I can tell, a G2-driven power pentode works just like a triode with a Mu of about 1. Yes, low Mu in a triode means lower Rp and more power, but also butt-breaking drive voltage. For reasonable assumptions, Mu of about 4 or 5 is "best". See '50, 2A3, 300B, etc. For tubes with high plate voltage rating and low plate power rating, a higher Mu is workable: 807/6L6 is an OK triode if you run a high B+ and high load resistance. Since gain is always less than Mu, and plate swing can approach twice the supply, the grid swing even at Mu=4 can approach the supply rails too close for low driver distortion; why make things worse? And worse again by taking drive current? Still worse because the drive current is a complex function of both G2 and Plate voltage and current.
Or look at it another way. They could have omitted G1 and G3/beam, and told you that G2 was the control grid, and gotten a "new" tube type. No new tooling. Why didn't they? Because of what you are finding: it is just hard work to control a tube with a grid that has such poor control.
Some of the bigger H-sweep tubes have G2 Mu near 5, conservative plate dissipation ratings, and huge plate voltage ratings. These would make marvvy G1-control power triodes if worked push-pull to cancel their curvature. (And in SE, the curve just causes "high" 2nd harmonic, not a problem to the ear.)
VG1 and VG2 are semi-locked by the Mu, G1 to G2.
Take a typical power tube in typical use. Mu(G1-G2) is say 10. G2 is at 400V, G1=-24V gives a nice idle current, G1= 0V to -48V for a good swing.
Or: (-24V)+(+400V/10) gives a +16V field off the cathode to urge the idle current, (0V)+(+400V/10)= +40V for peak current, (-48V)+(+400V/10) +8V for near-cutoff current.
If we make G1 zero V all the time, then G2 must be 160V for idle, 400V for peak, +80V for near-cutoff.
If we make G1 +10V all the time, then G2 must be 60V for idle, 300V for peak, -20V for near-cutoff. That is not quite right: G1 is now sucking ~10mA away from the cathode, so the Plate current is probably 10mA less than with negative grid.
You could of course just set up a tube with a +/- G1 bias supply, an adjustable G2 supply and a high-volt signal, and try it on the bench. In fact when working outside the conventional techniques, that's the only way to be sure nothing odd happens. In conventional work, tube makers were careful to minimize (or document) any odd behavior, but when you do things strange you may discover strange things.
> The idea behind it is a reduction of G2 current.
In the simple pentodes with plate voltage not much lower than G2 voltage, G2 current tends to be a somewhat constant fraction of plate current. So "reducing G2 current" implies a reduction in plate current. But plate current is our precious output, which we do NOT want to reduce.
For large plate peak current with large power output and good efficiency, the plate goes very low when G2 goes very high. Large peak G2 currents seem unavoidable. Using the positive G1 trick, the G2 at the peak may be 300V instead of 400V; meanwhile we hope the plate has gone down near 50V. I don't know if getting peak G2 current down to 75% of before is much of an improvement.
When you get into the aligned-grid "beam-G3" pentodes, all bets must be off. What they have done is try to reduce G2 current to zero and G3 stray current to zero. Small changes in field conditions or factory winding cause large variations in actual current. Get too clever and you un-do the careful cancellations designed into these tubes. I suspect G2 (and maybe G3/beam-electrode) current could soar far beyond what the designer built them for. For example: in the align-grids 6L6, the negative voltage on G1 casts shadows where the G2 wires are, reducing G2 current. Make G1 positive, and what happens? Actually not a disaster: such tubes are run with positive peaks in radio work. But peak G2 current is not an issue there: it is heavily bypassed.
> how do you make a sort of triode out of a pentode
So use a triode. If power and efficiency were important goals, you would be using transistors. Or if transistors have not been invented yet, you would use 6L6 power pentodes. As far as I can tell, a G2-driven power pentode works just like a triode with a Mu of about 1. Yes, low Mu in a triode means lower Rp and more power, but also butt-breaking drive voltage. For reasonable assumptions, Mu of about 4 or 5 is "best". See '50, 2A3, 300B, etc. For tubes with high plate voltage rating and low plate power rating, a higher Mu is workable: 807/6L6 is an OK triode if you run a high B+ and high load resistance. Since gain is always less than Mu, and plate swing can approach twice the supply, the grid swing even at Mu=4 can approach the supply rails too close for low driver distortion; why make things worse? And worse again by taking drive current? Still worse because the drive current is a complex function of both G2 and Plate voltage and current.
Or look at it another way. They could have omitted G1 and G3/beam, and told you that G2 was the control grid, and gotten a "new" tube type. No new tooling. Why didn't they? Because of what you are finding: it is just hard work to control a tube with a grid that has such poor control.
Some of the bigger H-sweep tubes have G2 Mu near 5, conservative plate dissipation ratings, and huge plate voltage ratings. These would make marvvy G1-control power triodes if worked push-pull to cancel their curvature. (And in SE, the curve just causes "high" 2nd harmonic, not a problem to the ear.)
Biasing g1 positive staying inside the gate current limits, how low would G2 have to be biassed (and could it even go negative for some tubes?).
PRR, thanks once again for clarifying the issue in a straightforward manner! This sort of discussion is exactly what I've been looking for in ths thread.
The idea behind it is a reduction of G2 current.
Yes, not much of an improvement but in some cases enough - enough to use a driver tube you already have handy I'm not really looking for revolutionary discoveries here
Again some good insight. Still, looking at examples of sweep and similar tube G2 drive (which is mostly what I'm getting at anway), the G2 current stays more or less within sane values.
how do you make a sort of triode out of a pentode
Of course, that would be the obvious solution. The problem is that it is VERY diffcult to get them at sane proces here. So, as I said, one has to figure out how to make use of what one has, i.e. make a useful triode out of a pentode.
Looking at the Svetlana application note by B. Danielak, from the curves one can see a gain of about 6.6 into a 2.4 kohm load using G2 drive. Since the resulting curves are sort-of half way between triode and pentode, Rp is not too low at roughly 5 k ohms.
Yes, that was originally my first thought until you look up the rather low maximum G2 voltage. As an example, PL519, which is the biggest sweep tube available here, has a G2 Vmax = 275V. Tie this to the plate and you are limited to a B+ of roughly half that in PP, which is quite low, so a huge plate Vmax really does nothing for you in this case, unless you run G2 over spec (wich, given the robustness of the tube and a cold G2 max of 700V should not be a big issue, but still, there's a reason for the spec). There is considerable allure in this approach anyway - the tubes are (relatively) low cost, and the low Rp may mean a low cost mains toroidal transformer as an OPT in a PP configuartion.
Still, I suspect the use of more commonly available OPTs or simply higher B+ voltages to get more power, is why G2 drive is applied to these tubes.
PRR, thanks once again for clarifying the issue in a straightforward manner! This sort of discussion is exactly what I've been looking for in ths thread.
The idea behind it is a reduction of G2 current.
Yes, not much of an improvement but in some cases enough - enough to use a driver tube you already have handy
I'm not really looking for revolutionary discoveries here Again some good insight. Still, looking at examples of sweep and similar tube G2 drive (which is mostly what I'm getting at anway), the G2 current stays more or less within sane values.
how do you make a sort of triode out of a pentode
Of course, that would be the obvious solution. The problem is that it is VERY diffcult to get them at sane proces here. So, as I said, one has to figure out how to make use of what one has, i.e. make a useful triode out of a pentode.
Looking at the Svetlana application note by B. Danielak, from the curves one can see a gain of about 6.6 into a 2.4 kohm load using G2 drive. Since the resulting curves are sort-of half way between triode and pentode, Rp is not too low at roughly 5 k ohms.
Yes, that was originally my first thought until you look up the rather low maximum G2 voltage. As an example, PL519, which is the biggest sweep tube available here, has a G2 Vmax = 275V. Tie this to the plate and you are limited to a B+ of roughly half that in PP, which is quite low, so a huge plate Vmax really does nothing for you in this case, unless you run G2 over spec (wich, given the robustness of the tube and a cold G2 max of 700V should not be a big issue, but still, there's a reason for the spec). There is considerable allure in this approach anyway - the tubes are (relatively) low cost, and the low Rp may mean a low cost mains toroidal transformer as an OPT in a PP configuartion.
Still, I suspect the use of more commonly available OPTs or simply higher B+ voltages to get more power, is why G2 drive is applied to these tubes.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.