I have found a way to make an excellent pentode from some real "junk" tubes. Look at the plate curves for any of the dual control pentode tubes (6LE8, 9KC6, 6BV11, 6MK8,6BU8,....)and they are very rounded and high distortion due to screen current siphoning off plate current at low plate voltage. These tubes have a close-meshed G3 (suppressor grid) instead of the very open spaced G3 usally found in normal pentodes. (Usually used for control/modulation purposes in the dual control case) The screen grid G2 wires normally cause electrons passing by to be deflected slightly off course due to the screen grid potential. The open mesh G3 of normal pentodes at zero volts still lets most of these get by to the plate, but the close mesh G3 of the dual control tubes deflects a lot of these back to the G2 screen grid causing the high screen grid currents. G3 is normally operated at zero volts (relative to cathode) for convenience and to repel secondary emission electrons from the plate back to the plate. Most normal pentodes have G3 internally connected to the cathode as well. If one puts about +12V to +15V on G3 (relative to the cathode) of the dual control pentodes (which have an isolated pin for G3), then the screen deflected electrons still get by to the plate. Secondary emission suppression is hardly affected by 15V less difference between plate and G3. The plate curves square up into the best beam pentode curves you have ever seen. Screen current drops significantly (compared to 0V on G3) at low plate voltage. Current into G3 is miniscule. The plate resistance of these tubes is very high due to the screen grid and the close meshed G3 both acting as shields. They should make excellent high impedance current sources. These dual control tubes often go for $.50 when on sale.
Don
Don
Re: A better pentode for less!
That's an interesting post smoking-amp. I looked up a data sheet, and see where you're coming from. I then wondered where one could use one of these - given that the output capacitance is a bit higher than the competition. The ideal use would be as a CCS for a pair of balanced cathode followers. If the drive to the cathode followers is balanced, then the sum of the audio currents is balanced, so the cathode current in the dual anode CCS valve would be constant. The fact that the CCS's output capacitance is a bit high wouldn't matter, because of the low impedance at the cathode of those cathode followers, and that capacitance would be swamped by the load capacitance anyway.
Good find! I may have to do some experimenting.
That's an interesting post smoking-amp. I looked up a data sheet, and see where you're coming from. I then wondered where one could use one of these - given that the output capacitance is a bit higher than the competition. The ideal use would be as a CCS for a pair of balanced cathode followers. If the drive to the cathode followers is balanced, then the sum of the audio currents is balanced, so the cathode current in the dual anode CCS valve would be constant. The fact that the CCS's output capacitance is a bit high wouldn't matter, because of the low impedance at the cathode of those cathode followers, and that capacitance would be swamped by the load capacitance anyway.
Good find! I may have to do some experimenting.
Some more dual control tubes: 7 pin: 6GY6, 6GX6, 6HZ6, 6AS6, 6DB6, 6DT6, 5725, 6187, 6954
9 pin: 6LE8, 6BU8, 6KF8, 6MK8, 6GS8, 6HS8 (these have dual plates too)
octal: 6888
compactron: 6BV11,12BV11 (two dual control units per tube)
compactron mixed: 6AD10, 6AL11, 6BF11, 6BY11, 6G11, 6T10, 6Y10?, 9BJ11 ,12AE10, 13V10, 18AJ10 (one dual control unit and one power pentode per tube)
The 9 pin tubes are dual control and dual plate. I have been using the 6LE8 with the plates tied together to act as one tube,
but as EC8010 suggests, they can be used as dual current sources separately, and balanced signals fit nicely.
I have been looking at the specs on output plate capacitance of the dual control tubes to compare with other tubes. Just from the presence of more wires in the G3 grid I would expect more output capacitance to plate. Surprisingly though, it depends on what you compare with. For example, the 9KC6 has identical structure to a 12HL7 tube except for G3, the 9KC6 has 3.0 pf and the 12HL7 has 6.0 pf. This is probably the case for beam pentodes which have a lot of sheet metal in their beam forming G3. The 6LE8 has 4.2 pf per plate section for a total of 8.4 pf, while a 6HB6 (nearly identical cathode and grid 1, a beam pentode) has 8.0 pf. However, the dual control tubes are derated about 50% in power due to the high screen grid current they usually draw. With G3 at +12V to +15V, the screen current drops significantly, so some of this derating can probably be rolled back.
If one compares the dual control tubes at their given watt rating with similarly rated tubes though, then output capacitance is more like 2x. Since the dual control tubes were used at high frequencies as demodulators typically, I would be surprised if output capacitance is really that much of an issue, especially at audio frequencies. But in any design, output capacitance, stray wire capacitance and next tube grid input capacitance must be considered when calculating max slew rate.
Don
9 pin: 6LE8, 6BU8, 6KF8, 6MK8, 6GS8, 6HS8 (these have dual plates too)
octal: 6888
compactron: 6BV11,12BV11 (two dual control units per tube)
compactron mixed: 6AD10, 6AL11, 6BF11, 6BY11, 6G11, 6T10, 6Y10?, 9BJ11 ,12AE10, 13V10, 18AJ10 (one dual control unit and one power pentode per tube)
The 9 pin tubes are dual control and dual plate. I have been using the 6LE8 with the plates tied together to act as one tube,
but as EC8010 suggests, they can be used as dual current sources separately, and balanced signals fit nicely.
I have been looking at the specs on output plate capacitance of the dual control tubes to compare with other tubes. Just from the presence of more wires in the G3 grid I would expect more output capacitance to plate. Surprisingly though, it depends on what you compare with. For example, the 9KC6 has identical structure to a 12HL7 tube except for G3, the 9KC6 has 3.0 pf and the 12HL7 has 6.0 pf. This is probably the case for beam pentodes which have a lot of sheet metal in their beam forming G3. The 6LE8 has 4.2 pf per plate section for a total of 8.4 pf, while a 6HB6 (nearly identical cathode and grid 1, a beam pentode) has 8.0 pf. However, the dual control tubes are derated about 50% in power due to the high screen grid current they usually draw. With G3 at +12V to +15V, the screen current drops significantly, so some of this derating can probably be rolled back.
If one compares the dual control tubes at their given watt rating with similarly rated tubes though, then output capacitance is more like 2x. Since the dual control tubes were used at high frequencies as demodulators typically, I would be surprised if output capacitance is really that much of an issue, especially at audio frequencies. But in any design, output capacitance, stray wire capacitance and next tube grid input capacitance must be considered when calculating max slew rate.
Don
I haven't seen any schematics using these oddball tubes. But a few guidelines can be used to convert them to conventional tube versions. First, you have to find the optimum g3 voltage that minimizes g2 current at the operating point chosen. Typically +12V to +32V works on g3. (too much + g3 and the current starts going to g3 instead, NG) This is low current at g3, so a resistive divider can be used from B+ to derive this, or a zener setup.
A primitive curve tracer setup using a scope can be very illuminating while adjusting g3. The plate curves will square right up like a normal beam pentode when g3 bias is optimum.
Next, measure the new g2 current and plate current at the chosen operating point with g3 bias applied. You will find that g2 current has dropped considerably from the datasheet figure (given at g3 = 0). And this means that more current is arriving at the plate than before. (the + g3 bias prevents electron returns back to g2).
This effects two things. The previous large g2 power dissipation will drop, and this is usually what was limiting for the design sheet data. The 9KC6 for example is datasheet rated at 7 Watts plate dissipation, and the similar 12HL7 and 12GN7 is rated for 10 Watts. You should still observe the 1 watt (12HL7) limit for the screen grid when exceeding the 7 Watt limit.
Second, the g1 transconductance increases with g3 pos. bias since more current is making it to the plate instead of being intercepted by rebound at g2. The 9KC6 should increase it's g1 transconductance from its 24000 datasheet figure to more like 36000 (12GN7 territory). Remember, that gm scales as plate current to the .666 power (practical, not theoretical) when comparing different operating points or different tubes. Rp varies as -.666 power with current. I don't think the g3 bias will much affect the rp figure at the same current like it did for the g1 gm though.
So net effect is the 9KC6 becomes close to a 12GN7. A 6LE8 becomes close to a 6HB6 (although the 6LE8 does not have the same large plate structure as the 6HB6, so I would stick to the same sum: 2+2+2 = 6 watts for total screen grid + plates dissipation). Similar upgrade adaptations can be figured for the other dual control tubes.
Some further tweeks are available with these tubes. Normally a pentode has an expanding 3/2 power rule for current versus g1 voltage up until saturation. By lowering the + g3 DC bias from its optimum, this expansionist rule can be bucked some by diverting some current back to the g2 grid. (Keep an eye on g2 dissipation though). So some linearization (really counter distortion) can be produced much like in UL with a high g2 voltage. A distortion analyzer is called for to look for an optimum.
For g2 drive applications (not common for these small watt tubes though) one can try connecting the g3 grid to the g2 to get more g2 gm effectively.
Don
A primitive curve tracer setup using a scope can be very illuminating while adjusting g3. The plate curves will square right up like a normal beam pentode when g3 bias is optimum.
Next, measure the new g2 current and plate current at the chosen operating point with g3 bias applied. You will find that g2 current has dropped considerably from the datasheet figure (given at g3 = 0). And this means that more current is arriving at the plate than before. (the + g3 bias prevents electron returns back to g2).
This effects two things. The previous large g2 power dissipation will drop, and this is usually what was limiting for the design sheet data. The 9KC6 for example is datasheet rated at 7 Watts plate dissipation, and the similar 12HL7 and 12GN7 is rated for 10 Watts. You should still observe the 1 watt (12HL7) limit for the screen grid when exceeding the 7 Watt limit.
Second, the g1 transconductance increases with g3 pos. bias since more current is making it to the plate instead of being intercepted by rebound at g2. The 9KC6 should increase it's g1 transconductance from its 24000 datasheet figure to more like 36000 (12GN7 territory). Remember, that gm scales as plate current to the .666 power (practical, not theoretical) when comparing different operating points or different tubes. Rp varies as -.666 power with current. I don't think the g3 bias will much affect the rp figure at the same current like it did for the g1 gm though.
So net effect is the 9KC6 becomes close to a 12GN7. A 6LE8 becomes close to a 6HB6 (although the 6LE8 does not have the same large plate structure as the 6HB6, so I would stick to the same sum: 2+2+2 = 6 watts for total screen grid + plates dissipation). Similar upgrade adaptations can be figured for the other dual control tubes.
Some further tweeks are available with these tubes. Normally a pentode has an expanding 3/2 power rule for current versus g1 voltage up until saturation. By lowering the + g3 DC bias from its optimum, this expansionist rule can be bucked some by diverting some current back to the g2 grid. (Keep an eye on g2 dissipation though). So some linearization (really counter distortion) can be produced much like in UL with a high g2 voltage. A distortion analyzer is called for to look for an optimum.
For g2 drive applications (not common for these small watt tubes though) one can try connecting the g3 grid to the g2 to get more g2 gm effectively.
Don
Member Pete Millett published an amplifier design in the AudioXpress magazine recently. It used a dual control pentode with positive bias applied to G3. The article can be found here:
http://www.audioxpress.com/magsdirx/ax/addenda/media/millet2903.pdf
The transconductance of the beam forming plates is fairly low in beam pentodes I've seen. But it seems to be enough to prevent the "snivets" or Barkhausen oscillations with +30 V or so for TV deflection use.
A lot of the horiz. output tubes have "Barkhausen plates" in addition on the inside of the plate structure. They skew the electron arrival times at the plate so as to "snuff out" the VHF oscillations. Usually can tell by a bulge in the central plate structure near the mounting rails externally.
For example, 6KV6, 6HB5 and 21JV6 have big Bark... plates. The 6HJ5, 6HD5 have minute ones. 6JN6, 6GE5 may not have any. (haven't dissected one yet, but no visible bulge.)
No idea what effect Bark... plates have on audio sound quality. Probably not much, unless one tries to do triode mode with the plate as actual plate rather than the screen grid + plate as plate. The Bark... plates would skew the physical Mu factor over a small range for the plate alone.
Don
A lot of the horiz. output tubes have "Barkhausen plates" in addition on the inside of the plate structure. They skew the electron arrival times at the plate so as to "snuff out" the VHF oscillations. Usually can tell by a bulge in the central plate structure near the mounting rails externally.
For example, 6KV6, 6HB5 and 21JV6 have big Bark... plates. The 6HJ5, 6HD5 have minute ones. 6JN6, 6GE5 may not have any. (haven't dissected one yet, but no visible bulge.)
No idea what effect Bark... plates have on audio sound quality. Probably not much, unless one tries to do triode mode with the plate as actual plate rather than the screen grid + plate as plate. The Bark... plates would skew the physical Mu factor over a small range for the plate alone.
Don
Are you referring to the fins mounted on the inside of the plate that run perpendicular to the plate? I dissected a 6LW6 that rolled off of my workbench and became one with the atmosphere. It was one of the big boys that have heat radiating fins welded to the outside of the plate. I had no clue why there were fins (2 of them) on the inside of the plate. These fins are welded to, and therefore connected to the plate.
The 6LW6 does work very well in triode mode (G2 connected to plate, and G3 grounded). That pesky 275 volt screen grid rating is not an issue I have been to 500 volts and 100 mA without even a sweat. Some 6LW6's do not have the extra fins on the outside of the plate. 45 to 55 watts is the limit for these guys. The big boys with the fins don't start to glow until about 100 watts!
tubelab:
"Are you referring to the fins mounted on the inside of the plate that run perpendicular to the plate? "
Yes, that's them. I would guess that most of the really big horiz. output tubes have them.
With g2 connected to the plate, the g2 grid should be defining the triode Mu factor, and I would expect good results.
Don
"Are you referring to the fins mounted on the inside of the plate that run perpendicular to the plate? "
Yes, that's them. I would guess that most of the really big horiz. output tubes have them.
With g2 connected to the plate, the g2 grid should be defining the triode Mu factor, and I would expect good results.
Don
I'm not familiar with the Bottlehead upgrade. But biasing g3 positive on beam pentodes is really a little different animal from biasing g3 positive on the dual control tubes. Certainly the "snivets" problem per se has been known for quite some time in TV design. Relevance to audio is less clear to me anyway.
The beam pentode's g3 (beam plate) design would presumably be optimised already for minimum g2 current. And they have a very low g3 transconductance besides.
The dual control tubes, on the other hand, have a far overeffective g3 that bounces back nearly half the plate current to the g2 (at g3 = 0 V), so that g3 modulation has some headroom to increase or decrease current to the plate. Putting some positive volts on the dual control g3 makes it more transparent (less of a barrier) to the electron stream, so brings operation back toward normal beam pentode design.
But clearly the two schemes are related.
Don
The beam pentode's g3 (beam plate) design would presumably be optimised already for minimum g2 current. And they have a very low g3 transconductance besides.
The dual control tubes, on the other hand, have a far overeffective g3 that bounces back nearly half the plate current to the g2 (at g3 = 0 V), so that g3 modulation has some headroom to increase or decrease current to the plate. Putting some positive volts on the dual control g3 makes it more transparent (less of a barrier) to the electron stream, so brings operation back toward normal beam pentode design.
But clearly the two schemes are related.
Don
If you all don't mind my resurrecting an ancient thread, this seemed the place to put this...
I'm still looking into squeezing the most out of a 12T10 and after trying triode strapping went back to testing the dual control pentode as a pentode, but with the suppressor biased positive. It hadn't worked well for me before, but I didn't know what I was doing either...
With more care I found something really significant. The curves square up better and better up to about +6V on the "suppressor", then stabilize. Beyond this up to about 20V the curves look identical to me. At 6V it was pulling 2ma to 0.5 ma depending on the current in the plate circuit.
This is hugely significant for my application as a LTP because it means that this grid has become insensitive and will not react to the signal coming in on the cathode. I thought I was going to have to design around this, but the problem doesn't even exist.
It also means there's no need to regulate, or for that matter bypass, the supply, assuming I'm right.
Has anyone else observed this?
Bonus question - for a LTP with the cathode at, say, 60V, I'm assuming that this means I'll need to set the voltage on the "suppressor" to 60V + ~10V = 70V, right?
I'm still looking into squeezing the most out of a 12T10 and after trying triode strapping went back to testing the dual control pentode as a pentode, but with the suppressor biased positive. It hadn't worked well for me before, but I didn't know what I was doing either...
With more care I found something really significant. The curves square up better and better up to about +6V on the "suppressor", then stabilize. Beyond this up to about 20V the curves look identical to me. At 6V it was pulling 2ma to 0.5 ma depending on the current in the plate circuit.
This is hugely significant for my application as a LTP because it means that this grid has become insensitive and will not react to the signal coming in on the cathode. I thought I was going to have to design around this, but the problem doesn't even exist.
It also means there's no need to regulate, or for that matter bypass, the supply, assuming I'm right.
Has anyone else observed this?
Bonus question - for a LTP with the cathode at, say, 60V, I'm assuming that this means I'll need to set the voltage on the "suppressor" to 60V + ~10V = 70V, right?
Last edited:
Here are some curves for the 6LE8 dual control tube, with various positive voltages on g3:
http://www.diyaudio.com/forums/tubes-valves/160240-suppresor-grid-used-feedback-6.html#post2083661
See also the 6BV11 curves above those.
When the g3 voltage goes even more positive than the optimum level for squared up "knees" on these dual control tubes, the curves start to get sharply shrunken down knees below 50V on the plate, much like tetrode kink curves. If you stay out of the 50V or below plate region you should be OK. What is happening with excessive positive g3 and low plate V, is the g3 starts to absorb the current before the plate can get it. Absorbing secondary emission too.
The g3 positive voltage is with respect to the cathode. You could just use a Zener up from the cathode and a resistor to something positive to set it.
http://www.diyaudio.com/forums/tubes-valves/160240-suppresor-grid-used-feedback-6.html#post2083661
See also the 6BV11 curves above those.
When the g3 voltage goes even more positive than the optimum level for squared up "knees" on these dual control tubes, the curves start to get sharply shrunken down knees below 50V on the plate, much like tetrode kink curves. If you stay out of the 50V or below plate region you should be OK. What is happening with excessive positive g3 and low plate V, is the g3 starts to absorb the current before the plate can get it. Absorbing secondary emission too.
The g3 positive voltage is with respect to the cathode. You could just use a Zener up from the cathode and a resistor to something positive to set it.
Last edited: