Tetrodes and many pentodes show a kink or dip in the anode curves (anode current vs anode voltage) due to secondary emission at around 50 to 70V. This is always most obvious at low anode currents.
Why is it most obvious at low currents?
In beam tetrodes (eg 6V6, 6AQ5) etc, the beaming causes a space charge after the screen dense enough to repel secondary emission electrons back to the anode - that is said to be the idea of beam tetrodes. So it is reasonable to assume that at low anode currents there isn't sufficient post-screen space charge to make this work.
But you see this behaviour in ordinary non-power non-beam pentodes eg 6BH6 (a small signal RF pentode) as well. I would have thought any variation in post-screen space charge would be insignificant compared to the field established by the suppressor grid, even with a pretty open suppressor grid pitch.
Why is it most obvious at low currents?
In beam tetrodes (eg 6V6, 6AQ5) etc, the beaming causes a space charge after the screen dense enough to repel secondary emission electrons back to the anode - that is said to be the idea of beam tetrodes. So it is reasonable to assume that at low anode currents there isn't sufficient post-screen space charge to make this work.
But you see this behaviour in ordinary non-power non-beam pentodes eg 6BH6 (a small signal RF pentode) as well. I would have thought any variation in post-screen space charge would be insignificant compared to the field established by the suppressor grid, even with a pretty open suppressor grid pitch.
Thanks Duncan, but this page is about inducing secondary emission under fairly high current conditions by disabling the suppressor grid by tying it to the screen. He was thus operating a pentode as a tetrode. It does not therefore show any hint or clue on my question.
I checked the associated page at https://www.frostburg.edu/personal/latta/ee/2ndemission/secondary.htm but this is just a common description of tube operation. It is also evident that the author doesn't distinguish between electron reflection (which occurs at all voltages) and secondary emission (which increases rapidly with voltage).
Keit
I checked the associated page at https://www.frostburg.edu/personal/latta/ee/2ndemission/secondary.htm but this is just a common description of tube operation. It is also evident that the author doesn't distinguish between electron reflection (which occurs at all voltages) and secondary emission (which increases rapidly with voltage).
Keit
Several effects are probably in action there. The screen grid will start eating a larger proportion of the cathode current at low plate V (electrons having less momentum to get by the screen).
The suppressor grid works both ways. It not only repels secondaries back to the plate but repels primaries back to the screen grid. Portions of the primary stream get deflected off direct course when passing near g1 and g2 grid wires, so there is some spread in velocity toward the plate. Weakly directed ones get sent back to positive grid 2. Low plate V would send a lot back to grid 2.
Dual control tubes like 9KC6 can show interesting effects on a curve tracer by varying the Voltage positive or negative on g3. +12V on grid 3 turns the tube into a nice square knee'd tube with little sign of kinks anywhere. (with like 40,000 gm, a D3a replacement for $3 ) Negative V on g3 rounds the plate curves off terribly (as shown on the data sheet), by sending large portions of the cathode current back to grid 2. (by the way, positive +12V on g3 reduces heating of grid 2 greatly, so the tube can be up-rated near to a 12HL7. Every Watt reduced on grid 2 typically allows 4 Watts increased spec for the plate diss. )
The suppressor grid works both ways. It not only repels secondaries back to the plate but repels primaries back to the screen grid. Portions of the primary stream get deflected off direct course when passing near g1 and g2 grid wires, so there is some spread in velocity toward the plate. Weakly directed ones get sent back to positive grid 2. Low plate V would send a lot back to grid 2.
Dual control tubes like 9KC6 can show interesting effects on a curve tracer by varying the Voltage positive or negative on g3. +12V on grid 3 turns the tube into a nice square knee'd tube with little sign of kinks anywhere. (with like 40,000 gm, a D3a replacement for $3 ) Negative V on g3 rounds the plate curves off terribly (as shown on the data sheet), by sending large portions of the cathode current back to grid 2. (by the way, positive +12V on g3 reduces heating of grid 2 greatly, so the tube can be up-rated near to a 12HL7. Every Watt reduced on grid 2 typically allows 4 Watts increased spec for the plate diss. )
Last edited:
Yes, true. You can use a single pentode to amplify without phase inversion by using the suppressor as a control grid and taking the output from the screen (anode held at constant voltage), which clearly demonstrates repulsion of forward electrons. Practically nobody uses a tube that way in amplification because the gain is really low and the distortion usually pretty bad, but it has its uses in certain oscillators.
However, I am having trouble visualizing why repulsion of forward electrons should produce a more obvious "tetrode kink" at low anode currents - the anode voltage range is the same and therefore the electric field gradients from screen to anode at each anode current is the same (the different currents being caused by changing the control grid bias).
Actually, repulsion of forward electrons would enhance a space charge between screen and suppressor - this would I think make the effect more obvious at high anode currents - the opposite of what happens. (at low currents, there would be few electrons to build up a space charge)
I did some experiments with CCS-loaded power beam tubes and true pentodes. I was using them at low currents (compared to what they were rated for).
I noticed I'd get a big rise in distortion if the plate went lower than 75 volts or so because of the kinky behavior down there. Beam tubes were worse than pentodes. Lowering the screen voltage helped a lot. I didn't play with suppressor voltage (kicking myself for not doing so).
With 10mA CCS-loaded EL34, I was getting ridiculously-low distortion at massive voltage swings with plate idle voltage at 500V or more and screen set to 35V.
I don't have a curve tracer to confirm graphically but I think lowering screen voltage is the secret to fixing this. If you are going to drive into the low-current, low-voltage part of the curves, you don't need a high screen voltage anyway.
I noticed I'd get a big rise in distortion if the plate went lower than 75 volts or so because of the kinky behavior down there. Beam tubes were worse than pentodes. Lowering the screen voltage helped a lot. I didn't play with suppressor voltage (kicking myself for not doing so).
With 10mA CCS-loaded EL34, I was getting ridiculously-low distortion at massive voltage swings with plate idle voltage at 500V or more and screen set to 35V.
I don't have a curve tracer to confirm graphically but I think lowering screen voltage is the secret to fixing this. If you are going to drive into the low-current, low-voltage part of the curves, you don't need a high screen voltage anyway.
The 9KC6 when g3 is tuned ( +12V on g3) to let maximum electrons get past, does not have any kinks in the low current range. So it seems to me that those other pentodes with kinks must have too coarse a winding pitch on g3 to act as a screen, but instead are just acting as a spiral beam former. (losing the space charge effect at low current) However, the 12GN7 beamer doesn't have much kinking either (at least for modest screen V). The 9KC6 looks like it is a 12GN7 with the beam plates replaced by a dense grid 3.
Another mysterious case is Crazy Drive, where g2 voltage (and g1 + voltage) is proportional to plate current. Rather kinky TV Sweeps loose the kinks completely in this mode. There must be some factor of reflected current (and/or grid 1 wire focusing) that vanishes when the screen V is proportional to current (hence low screen V at low current). Or maybe it is just because grid 1 does not go negative (see below).
Screen drive similarly looses the kinks, unless g1 is negative. (g2 drive) So maybe some grid 1 wire beam focusing is in play in all this. (to avoid hitting grid 2 wires)
Another mysterious case is Crazy Drive, where g2 voltage (and g1 + voltage) is proportional to plate current. Rather kinky TV Sweeps loose the kinks completely in this mode. There must be some factor of reflected current (and/or grid 1 wire focusing) that vanishes when the screen V is proportional to current (hence low screen V at low current). Or maybe it is just because grid 1 does not go negative (see below).
Screen drive similarly looses the kinks, unless g1 is negative. (g2 drive) So maybe some grid 1 wire beam focusing is in play in all this. (to avoid hitting grid 2 wires)
Last edited:
Oh, I know how to avoid kinking being a problem - lower the screen voltage as you say, restrict signal amplitude so the kink region isn't entered, or select a pentode that doesn't kink. Some types do and some don't.
What makes it interesting is that some types, and the 6BH6 is an example, show kinking if made by one manufacturer (Tung Sol in this case) but show no kinking if made by someone else (eg Sylvania in this case).
Also, some tubes are made as pentodes by some, and tetrodes by others (eg 3V4), but both the pentode and the tetrode show minimal kinking.
But I want to understand why, when kinking does occur in pentodes, it is always most obvious at low anode currents resulting from control grid biasing.
The 9KC6 was designed as a TV video amplifier - as such it is a sort of power tube, operating with considerable current. TV sweep tubes even more so.
As I said in my original post, one can understand there is enough current in power tubes to produce a beam tetrode-like post screen grid space charge - this implies that at low currents this might not be the case.
But I was talking about small signal RF pentodes - the sort that would be employed in radio RF & IF stages. I had been thinking that these never have enough current to make a significant space charge, so the suppressor must be doing its job. Yet many of them still show kinking at (for them) low currents.
Keit I waited to see if somebody had come up with an answer that satisfies you ,but not so far.
Did you discount my post ?
I have a reason for stating it and that is the modern standard of the metallic components inside a radio valve don't come up to the quality of the big valve factories that used to exist in the UK .
In the UK ( unlike the USA ) domestic washing machines have glass doors , in conjunction with a UK Professor on the same website I was on who specialized in chemical composition of materials we proved that a lot of the glass used was low grade and incorrectly manufactured ,resulting in exploding glass .
You don't need me to tell you where those so called well known in the UK & the States glass was manufactured and yes I checked into 100,s of them .
Conversely the metallic composition of anodes/cathodes/ grids etc --IMO-- is ( in some cases ) sub standard .
We are talking of the quality of TUNGSTEN ALLOYS -NI-W--NI-CA--NI-SR etc .
I have been hampered by the low grade of searches on my three search engines not precisely dealing with purely engineering matters but ( thank god ) America has come to the rescue and I managed to find a USA scientific search engine which goes into great detail about the metallic properties of that type of metal.
Now what I think might not be right but it cant be discounted , the tube I mentioned on my previous post was exactly the type you are talking about .
Did you discount my post ?
I have a reason for stating it and that is the modern standard of the metallic components inside a radio valve don't come up to the quality of the big valve factories that used to exist in the UK .
In the UK ( unlike the USA ) domestic washing machines have glass doors , in conjunction with a UK Professor on the same website I was on who specialized in chemical composition of materials we proved that a lot of the glass used was low grade and incorrectly manufactured ,resulting in exploding glass .
You don't need me to tell you where those so called well known in the UK & the States glass was manufactured and yes I checked into 100,s of them .
Conversely the metallic composition of anodes/cathodes/ grids etc --IMO-- is ( in some cases ) sub standard .
We are talking of the quality of TUNGSTEN ALLOYS -NI-W--NI-CA--NI-SR etc .
I have been hampered by the low grade of searches on my three search engines not precisely dealing with purely engineering matters but ( thank god ) America has come to the rescue and I managed to find a USA scientific search engine which goes into great detail about the metallic properties of that type of metal.
Now what I think might not be right but it cant be discounted , the tube I mentioned on my previous post was exactly the type you are talking about .
Duncan,
I didn't comment on your post about inferior metals, because use of any inferior metals does not bear upon my question.
Pentodes of certain types have always shown kinking in anode curves - and not only has that been shown in data published by Sylvania, Philips/Mullard, RCA, etc, I have verified it on NOS American and Mullard examples.
I think you are right about inferior metals used in modern production - other things like glass, too. When tube production was in full swing in the UK, Holland, Germany, USA, Australia, etc, it supported a huge specialised industry supplying special extreme purity metals, etc. While the guitar amplifier market and the audiophile market is sufficient to sustain tube production in Russia, China, and even 300B production in the USA, the quantities are very very tiny compared to production in western countries before the semiconductor age.
I suspect current manufacturing in Russia and China just has to make do with what materials they can get. Undoubtedly a lot of knowhow no longer exists. Manufacturers like Mullard, RCA, had some very clued up physicists and engineers working for them. And substantial quality control infrastructure.
Personally, I avoid Chinese tubes completely and only use Russian tubes when I have no alternative. Mostly I seek out NOS and carry out tube testing when purchases arrive to verify as new condition.
Now, back to just how kinking in pentodes comes about at low anode currents ....
I didn't comment on your post about inferior metals, because use of any inferior metals does not bear upon my question.
Pentodes of certain types have always shown kinking in anode curves - and not only has that been shown in data published by Sylvania, Philips/Mullard, RCA, etc, I have verified it on NOS American and Mullard examples.
I think you are right about inferior metals used in modern production - other things like glass, too. When tube production was in full swing in the UK, Holland, Germany, USA, Australia, etc, it supported a huge specialised industry supplying special extreme purity metals, etc. While the guitar amplifier market and the audiophile market is sufficient to sustain tube production in Russia, China, and even 300B production in the USA, the quantities are very very tiny compared to production in western countries before the semiconductor age.
I suspect current manufacturing in Russia and China just has to make do with what materials they can get. Undoubtedly a lot of knowhow no longer exists. Manufacturers like Mullard, RCA, had some very clued up physicists and engineers working for them. And substantial quality control infrastructure.
Personally, I avoid Chinese tubes completely and only use Russian tubes when I have no alternative. Mostly I seek out NOS and carry out tube testing when purchases arrive to verify as new condition.
Now, back to just how kinking in pentodes comes about at low anode currents ....
I thought "work function" referred to the thermal EMF developed across dissimilar metal joints. This -could- effect grid 1 bias at the high temps occurring in tubes.
The 6BN11 and 6J11 are identical tubes except for an innate voltage difference in the pentode grid 1 biasing. (explanation from JEDEC data sheets). Apparently this was requested by some TV manufacturer. Seeing as how neg. V bias on grid 1 sometimes seems to be related to kinking, maybe.... but more probably allowed the manufacturer to skip some $0.01 biasing resistors.
I have seen occasionally (26LW6 comes to mind) some offset of grid 1 between tubes that needed to be taken into account for Crazy Drive configuration. (a small offset V required for the grid 1 to cathode resistor to get proper constant gm curves)
--------------------------------------
Some 6JC6 tubes marked "JAPAN" that I got on sale had very sharp knees with some hysteresis in the plate curve knees (knee different versus directionality of the plate V scan). This turned out to be fixable with some positive V on grid 3 (beam plate actually). Surprisingly, these tubes turned out to have very good triode curves compared to the regular "good" pentodes. (no effect on triode mode curve quality from g3 biasing, just a small offset toward more conduction) (6JC6 is a small frame grid 1 IF tube, no grid alignment, no kinks)
I've also seen some other small RF beamers that apparently use too small a slot in the g3 beam plate. I'm guessing this reduces the capacitance from plate back to grid 1, and the user was expected to provide a small +V on the beam plate to fix the curves. ( knee fix similar to the 6JC6) Maybe some small RF pentodes did similarly with an overly fine g3 pitch.
Another observation is that bigger video frame grid 1 tubes like 12HL7 and 12GN7/12HG7 seem to have reduced kinking. The frame grid 1 does not have any allignment with the non frame grid 2 wires.
Seeing as how the E130L has frame grids for both g1 and g2 (presumably aligned, 80 ma versus 2.2 mA) , its pentode curves might be of interest for this low current kinking issue. The datasheet does show some kinking at low current, but not severe.
The 6BN11 and 6J11 are identical tubes except for an innate voltage difference in the pentode grid 1 biasing. (explanation from JEDEC data sheets). Apparently this was requested by some TV manufacturer. Seeing as how neg. V bias on grid 1 sometimes seems to be related to kinking, maybe.... but more probably allowed the manufacturer to skip some $0.01 biasing resistors.
I have seen occasionally (26LW6 comes to mind) some offset of grid 1 between tubes that needed to be taken into account for Crazy Drive configuration. (a small offset V required for the grid 1 to cathode resistor to get proper constant gm curves)
--------------------------------------
Some 6JC6 tubes marked "JAPAN" that I got on sale had very sharp knees with some hysteresis in the plate curve knees (knee different versus directionality of the plate V scan). This turned out to be fixable with some positive V on grid 3 (beam plate actually). Surprisingly, these tubes turned out to have very good triode curves compared to the regular "good" pentodes. (no effect on triode mode curve quality from g3 biasing, just a small offset toward more conduction) (6JC6 is a small frame grid 1 IF tube, no grid alignment, no kinks)
I've also seen some other small RF beamers that apparently use too small a slot in the g3 beam plate. I'm guessing this reduces the capacitance from plate back to grid 1, and the user was expected to provide a small +V on the beam plate to fix the curves. ( knee fix similar to the 6JC6) Maybe some small RF pentodes did similarly with an overly fine g3 pitch.
Another observation is that bigger video frame grid 1 tubes like 12HL7 and 12GN7/12HG7 seem to have reduced kinking. The frame grid 1 does not have any allignment with the non frame grid 2 wires.
Seeing as how the E130L has frame grids for both g1 and g2 (presumably aligned, 80 ma versus 2.2 mA) , its pentode curves might be of interest for this low current kinking issue. The datasheet does show some kinking at low current, but not severe.
Last edited:
correction: I think it was 6AF11 and 6AS11 tubes with the different pentode biasing V, not 6BN11 and 6J11.
Hmmm, I have a 6LB6 TV Sweep tube that has perfect -anti- alignment between grid 1 and grid 2 (visually obvious). As I recall dimmly, it has different plate curves from a good 6LB6. It also gets a hot grid 2 in seconds. I'll try to find that tube to trace it again. (may take a while to find, moved...) Might have some bearing on the low current kink issue.
Hmmm, I have a 6LB6 TV Sweep tube that has perfect -anti- alignment between grid 1 and grid 2 (visually obvious). As I recall dimmly, it has different plate curves from a good 6LB6. It also gets a hot grid 2 in seconds. I'll try to find that tube to trace it again. (may take a while to find, moved...) Might have some bearing on the low current kink issue.
Last edited:
Its thermal electron emission in tube metal that can occur not just at very high temperatures but in lower ones depending on the applied conditions to the other parts of the tube .
I found a "Paper " on it , I should have bookmarked it or stuck a link in .
That was my point that the standard logic when delved into showed that "secondary emissions " on a small scale not related to standard theory can occur due to the composition of the metal involved and the test was on an RF radio tube .
But Keit says it doesn't apply so further study for me ---but its interesting keeps my mind active.
I found a "Paper " on it , I should have bookmarked it or stuck a link in .
That was my point that the standard logic when delved into showed that "secondary emissions " on a small scale not related to standard theory can occur due to the composition of the metal involved and the test was on an RF radio tube .
But Keit says it doesn't apply so further study for me ---but its interesting keeps my mind active.
I think you either mis-read or is-remembered that paper. Any emission from grids would show up as grid current - that only occurs when there is mica contamination or cathode oxides have got on to the control grid. Or the tube is gassy. Or control grid driven positive.
Electron emission increases super sharply with temperature.
Re departure from expected due to variation in metal alloys:-
Just to be clear, I didn't say it didn't apply. I said it has no bearing on my question because I was asking about something that has always happened in some pentode types as indicated in original data book graphs.
Last edited:
In writing my last post, I just realised something. Any electrons from the cathode striking the screen grid can cause secondary emission from the screen. These will be too slow to get to or past the control grid but a fraction could go to the anode, given the coarse pitch used for suppressor grids. (the remainder would go back to the screen.) At normal anode currents, the increase in anode current from this cause would be quite negligible.
The fraction of screen secondary emission going to the anode would rise sharply with anode voltage - hence for low anode currents as set by control grid bias, it may not be negligible at moderate to high anode voltages. But only if the screen current falls off more slowly with control grid bias than does the anode current.
If so, it implies it is still tending towards a negligible addition to anode current at low anode voltages. Ahah! a cause of the kink or dip that has nothing to do with anode secondary emission. Never seen mention of this in any textbook though. All the textbooks I've ever seen just blame the kink on secondary emission at the anode. So I can't see this being right, but I've never seen plots of the screen current verse anode current for various control grid bias values - and I've never plotted it either.
The fraction of screen secondary emission going to the anode would rise sharply with anode voltage - hence for low anode currents as set by control grid bias, it may not be negligible at moderate to high anode voltages. But only if the screen current falls off more slowly with control grid bias than does the anode current.
If so, it implies it is still tending towards a negligible addition to anode current at low anode voltages. Ahah! a cause of the kink or dip that has nothing to do with anode secondary emission. Never seen mention of this in any textbook though. All the textbooks I've ever seen just blame the kink on secondary emission at the anode. So I can't see this being right, but I've never seen plots of the screen current verse anode current for various control grid bias values - and I've never plotted it either.
Last edited:
Secondary emission from the screen grid would require cathode current to produce it, which is low under the stated conditions. And how would that produce a drop in plate current?
Seems to me that screen grid absorption of the reduced cathode current is more likely. With low plate voltage, the electrons are more easily diverted to the constant screen grid voltage. Explaining why low screen V fixes the problem. Mis-aligned grids could also cause/agravate that, or de-focusing of the grid 1 shielding/focusing by increased negative grid 1 voltage beyond optimum. (too much neg. V on grid 1 could push the electron streamlets over to the next grid 2 wires.)
I'm pretty sure the mis-aligned 6LB6 I had had much worse kinks.
Seems to me that screen grid absorption of the reduced cathode current is more likely. With low plate voltage, the electrons are more easily diverted to the constant screen grid voltage. Explaining why low screen V fixes the problem. Mis-aligned grids could also cause/agravate that, or de-focusing of the grid 1 shielding/focusing by increased negative grid 1 voltage beyond optimum. (too much neg. V on grid 1 could push the electron streamlets over to the next grid 2 wires.)
I'm pretty sure the mis-aligned 6LB6 I had had much worse kinks.
Last edited:
Errr... Apparent hysteresis is a classic symptom of parasitic oscillation that can occur in some tube testers especially with high gm frame grid tubes like the 6JC6 and especially tubes that the tube tester manufacturer could not have checked when evaluating pre-production prototypes of his tester.
There isn't really hysteresis - it's just that oscillation may be provoked more during the upsweep or the downsweep or vice versa. The oscillation results in some degree of clipping or rectification, and thus shifting the curves.
If you are using a home-made tube tester all bets are off, unless you do what I do - add large ceramic capacitors with very short leads from each pin to cathode, and slug that tube down so hard it has to oscillate so high up in the UHF or microwave range that transit time won't let it.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.