Dave Cigna said:
Dave,
It is a form of negative feedback. (We are not talking about small anti-parasitic resistors.) When the anode swing goes low, the screen will draw more current, and thus also go down in voltage, and vice versa. It looks like UL performance in a sense but without the direct coupling with the output impedance and thus a form of loss. I have looked at this with a scope and the voltage appearing on G2 is more or less sinusoidal (with sine input), but I have never tried to measure effect on distortion (linearity). I would not do this but rather go to UL with its known characteristics. I have never seen graphs but they would be easily traceable with tube plotters. I still cannot imagine why with EL34 UL operation a 1K screen resistor is recommended specifically - but I believe the above is what is taking place.
Regards.
In a tetrode, a pentode or a beam tube, this is the g2 that attract the electrons, because the cathode see the g2, not the anode. The gain of the tube depend therefore much more on the voltage on the g2 as from the anode voltage.
The g2 is a very loss grid, so at the majority of the electrons will just pass thru it. After at they have passed thru the g2, they just continues to the anode.
The g2 cannot dissipate a lot of heat. Just a few watts for a power tube as an EL34 or 6L6. The problem is at, when the anode voltage is becoming lower as the g2 voltage, many electrons will be attracted by the g2, and you have to limit this current if you want to not destroy the tube.
To see how much current the g2 can take, look at page 8 and 9 at http://www.tubezone.net/pdf/807curv.pdf
Those curves show the variation of Ig2 for different Ua. It is a curve for different Ug1 and a serie for Ug2=250 et Ug2=300V
The 807 have almost the same electrical charachteristic as the 6L6, but with higher Ua.
On the curve of page 9, you have a point at Ug2=300V and Ig2=300mA. It implies Pg2=90watts. An hifi amp will never go to this point. Even a class B guitar amp will never reach to this point because to get at this point, you need to drive the tube with a power that no one driver will put on the command grid. (page 10 and 11 for Ig1=f(Ua)
But it is still points that we will get without screen grid resistor that can damage the tube.
The g2 is a very loss grid, so at the majority of the electrons will just pass thru it. After at they have passed thru the g2, they just continues to the anode.
The g2 cannot dissipate a lot of heat. Just a few watts for a power tube as an EL34 or 6L6. The problem is at, when the anode voltage is becoming lower as the g2 voltage, many electrons will be attracted by the g2, and you have to limit this current if you want to not destroy the tube.
To see how much current the g2 can take, look at page 8 and 9 at http://www.tubezone.net/pdf/807curv.pdf
Those curves show the variation of Ig2 for different Ua. It is a curve for different Ug1 and a serie for Ug2=250 et Ug2=300V
The 807 have almost the same electrical charachteristic as the 6L6, but with higher Ua.
On the curve of page 9, you have a point at Ug2=300V and Ig2=300mA. It implies Pg2=90watts. An hifi amp will never go to this point. Even a class B guitar amp will never reach to this point because to get at this point, you need to drive the tube with a power that no one driver will put on the command grid. (page 10 and 11 for Ig1=f(Ua)
But it is still points that we will get without screen grid resistor that can damage the tube.
Take a look at the fat bulb 6CA7 made by General Electric, this is a beam power tetrode and has very different internal construction to both other brands of 6CA7 which were pentodes and European made EL34 which I believe were/are all pentodes. I used to have some of these and if I find one I will try to post the picture, they are different - I used to see them in a lot of old dynaco st-70 amps which had been retubed in the 1960's or later. Seen also in guitar amps where they are considered desirable..
Kevin
Kevin
I understand that. In fact I said it here yesterday. But, Dennis Grimwood, the author of the web page from which your quote comes, claims that we want to keep the screen at a high enough voltage to accelerate electrons, but avoid having the electrons go through the screen circuit (instead of the plate circuit) because that wastes power. Specifically, he says
"Clearly there will be a particular value of Screen Grid Stopper Resistor that will provide optimum balance between the conventional "short-circuited" Screen Grid configuration and an arrangement whereby the Screen Grids are suitably loaded."
Well, it's not clear to me that there is an optimum balance (what exactly is being optimized?) or even that the screen should be loaded at all. I understand that applying negative feedback will reduce distortion, blah blah blah, but I don't think it 'clearly' is desirable to do so. More to the point, it's not clear what it even means that a screen be "suitably loaded."
It seems to me that it's just a design priority of his (minimize screen current in order to maximize plate current) and inserting series resistance is his method of choice.
Perhaps I'm a little predisposed: I've read the page in the past and what I've found impossible to forget is Grimwood's claim (at least it seems like he's claiming) that as the screen voltage is raised, when it reaches and exceeds the plate voltage it will rapidly start conducting more current. I just don't believe that it's true. Of course the screen conducts more as its supply voltage is increased, but nothing interesting happens at or around the plate voltage.
We can verify this from published plate curves of any tetrode/pentode/beam tube. If anything interesting happens when we hold Va constant and vary Vg2, then it should also be apparent when we hold Vg2 constant and vary Va. It would appear as a knee in the plate curves at Vg2. (At any given condition there's a certain number of electrons coming of the cathode. If more of them go to the screen then fewer of them will get to the plate.) But that knee isn't there in any of the pentode curves I've seen.
Anyway, I'm not entirely sure that Grimwood understands things as well as he appears to at first. I'm sure that he and I don't agree on our design priorities. Since minimizing screen current seems to be a design priority of his I'm not sure I place much value in his conclusions about svreen resistors.
As I said previously, the gain of a pentode or beam tube depend much more on Ug2 as on Ua. That implies at the best solution is to have a stablisation for Ug2 and no srceen resistor, that in order to have a stable gain on the whole range of output levels.
A screen resistor will limit the gain of the circuit, and this limitation will be proportional to Ig2. It will introduce distrortion too. It is a form of negative feedback, but its effect is not constant with the variations of output level.
I agree with Dennis Grimwood when he said at the grid screen is a protection device that limit the current in the g2, but I prefer to use a design where I can remove this resistor and wire the screen directly to the power supply, and that even if that implie at I must decrease Ug2.
If I look at class B amplifier as used in guitar and bass amps, I see at many (all?) manufacturers use fixed Ug2 and don't have such screen grid resistor, and that even if we get much more screen current with a such amp as with a hifi amp. A guitar amp is made to work in clipping, and it is in clipping at Ig2 become very important.
If you look at the Williamson or the Quad II, they doesn't have any grid screen resistor.
A screen resistor will limit the gain of the circuit, and this limitation will be proportional to Ig2. It will introduce distrortion too. It is a form of negative feedback, but its effect is not constant with the variations of output level.
I agree with Dennis Grimwood when he said at the grid screen is a protection device that limit the current in the g2, but I prefer to use a design where I can remove this resistor and wire the screen directly to the power supply, and that even if that implie at I must decrease Ug2.
If I look at class B amplifier as used in guitar and bass amps, I see at many (all?) manufacturers use fixed Ug2 and don't have such screen grid resistor, and that even if we get much more screen current with a such amp as with a hifi amp. A guitar amp is made to work in clipping, and it is in clipping at Ig2 become very important.
If you look at the Williamson or the Quad II, they doesn't have any grid screen resistor.
Dave Cigna said:
If you are referring to my post #21, I did not quote from anybody (cannot recall that I ever read Grimwood's article), just answered your question from basics. Also apologies that I did not think of your earlier post at the time; things get somewhat hazy here late at night... (it was 2:00 in the a.m.)
As said, I agree with you in that I cannot see the advantage of this practice. Has everybody else overlooked it - I believe not. I would also disagree with a sudden change in affairs when the G2 voltage goes higher than the anode if I understand correctly (not having read the article). Then one should have severe effects every time the anode goes on a negative swing. (But there still remains the EL34 findings with a 1K screen resistor.)
D_f, if you are talking of the original Williamson, that is a triode output topology. If you are talking of the later universally used UL, the G2 does of course get a signal on it which changes things (i.e. it does go down partly with the anode), as is the case with the Quad II (also UL).
Kevinkr, I look forward to your pictures, I have not seen that one. I am still asking: Did the characteristics remain the same from pentode to beam tube, and do we know whether the change included grids alignment?
Dave Cigna said:
"Clearly" is what I call a "fluff-word" - it means that if you say it fast enough and in the right manner, nobody asks awkward questions. Thus, it means the exact opposite of what it says...
By Child's law (3/2 power law) what goes on at the screen grid is non-linear. If we apply voltage from a low impedance source (the 20% or 43% taps on an output transformer) we can impose whatever voltage we want. If we apply that voltage via a resistor, then the non-linear current drawn will change the voltage at the screen grid. Perhaps that's what the resistors are tinkering with? Perhaps there is a null where two non-linear effects cancel?
I took a look at http://www.mif.pg.gda.pl/homepages/tom/files/blues_jr.gif
It is a fender schema using EL34 and 100 ohms screen grid resistor. If you look at the power supply. it is a 2k2 resistor between the +B for the anodes and the +X for the g2.
It is other schema on the same website, as exemple a marshall with 4 EL34, 1k g2 resistors and a serial coil to do the voltage for the g2.
The problem with such topology is at the DC voltage on the g2 is almost the same as the DC voltage on the anodes. That implies at the screens will conduct during a bigger part of the alternance where the anode is conducting. More, when the tube goes in clipping, the big amount of current in the power supply and the main transformer will do at not only the anode DC voltage will go down, but the DC voltage on the g2 will go down. As the gain of this pp stage is mainly determined by the dc voltage on the g2, the gain will decrease.
Another problem is at all those variation of the working point of the pp stage can cause high frequency oscillations. The output transformer is highly non-linear with the frequency, especialy at high frequency, and it can cause the circuit to oscillate.
A much better approch would be to use a higher DC anode voltage and a lower DC screen voltage. It have at least 2 advantages.
First, we have a bigger difference between the DC anode and screen voltage, so the part of the alternance where the g2 will take a bigger amount of current will be smaller.
Second, it will be easier to obtain a good regulation of the DC screen voltage in the power supply.
With a such topology, it become possible to use a high voltage cap between the 2 anodes of the pp (in parallel with the transformer) to avoid high frequency oscillations.
A valve as the EL34 is much better suited for a such approch as the 6L6.
In the siemens EL34 data sheet, we can read at with Ug2=375V and Ua=375V in class AB, we can reach 35 watts at the output with 5% distortion, and with Ug2=400V and Ua=800V in class B, we can reach 100 watts at the output with the same 5% distortion.
A good compromise would be to have 800V <= Ua <= 750V with 350V <= Ug2 <= 300V. Both philips and siemens recommand the use of a common screen resistor and they give the values. Wath we can see in those data sheet is at it is high variations of Va and Vg2 with the variations of the power output. So I still believe at it would be safe to have no grid screen resistor if Vg2 is stable.
Special car must be on the g2 in a guitar amp, because all the data sheets have the following remarq for the EL34 (it is also true with other tubes): "When using a sinusoidal input signal care should be takent not to exced the maximum admissble Wg2" http://www.mif.pg.gda.pl/homepages/frank/sheets/030/e/EL34.pdf
A guitar have not a sinusoidal signal, but it IS SINUSOIDAL when we get in larsen. So in fact a sinusoidal generator is a very good instrument to test the limit of a guitar amp. For you safety and the safety of the parts of the output stage, it is good to have a fuse in the cathode circuit.
I have done a few guitar amps with 807s, the first one was very simple, the last one have a regulated supply for Vg2, and no one have screen grid resistor. The first one have a Va of 530V at full output for 300V Vg2, and the last one have Va=600V and Vg2=250V for the same full output. The difference is at the sound of the last one have a much higher dynamic as the sound of the first one, and that even in clipping.
The 807 is a 6L6 with military quality and higher absolute rating. It can work with 600V on the anode at full output, and with intermittent service with 750V at full output.
It is no indication about a screen grid resistor in the 807 data sheets. But the RCA data sheet say at Vg2 must be made from an independant power supply or a voltage divider from the anode voltage (in order to have a good regulation).
I know at the EL34 is not the same tube, but I know at, if I was buying an amp with a such tube, I would try to use a good power supply for the g2 and no screen grid resistor. Only the practice can determine if it will work or not, and the gain in dynamic we get in all cases with a good power supply for Vg2 is worth the try.
It is a fender schema using EL34 and 100 ohms screen grid resistor. If you look at the power supply. it is a 2k2 resistor between the +B for the anodes and the +X for the g2.
It is other schema on the same website, as exemple a marshall with 4 EL34, 1k g2 resistors and a serial coil to do the voltage for the g2.
The problem with such topology is at the DC voltage on the g2 is almost the same as the DC voltage on the anodes. That implies at the screens will conduct during a bigger part of the alternance where the anode is conducting. More, when the tube goes in clipping, the big amount of current in the power supply and the main transformer will do at not only the anode DC voltage will go down, but the DC voltage on the g2 will go down. As the gain of this pp stage is mainly determined by the dc voltage on the g2, the gain will decrease.
Another problem is at all those variation of the working point of the pp stage can cause high frequency oscillations. The output transformer is highly non-linear with the frequency, especialy at high frequency, and it can cause the circuit to oscillate.
A much better approch would be to use a higher DC anode voltage and a lower DC screen voltage. It have at least 2 advantages.
First, we have a bigger difference between the DC anode and screen voltage, so the part of the alternance where the g2 will take a bigger amount of current will be smaller.
Second, it will be easier to obtain a good regulation of the DC screen voltage in the power supply.
With a such topology, it become possible to use a high voltage cap between the 2 anodes of the pp (in parallel with the transformer) to avoid high frequency oscillations.
A valve as the EL34 is much better suited for a such approch as the 6L6.
In the siemens EL34 data sheet, we can read at with Ug2=375V and Ua=375V in class AB, we can reach 35 watts at the output with 5% distortion, and with Ug2=400V and Ua=800V in class B, we can reach 100 watts at the output with the same 5% distortion.
A good compromise would be to have 800V <= Ua <= 750V with 350V <= Ug2 <= 300V. Both philips and siemens recommand the use of a common screen resistor and they give the values. Wath we can see in those data sheet is at it is high variations of Va and Vg2 with the variations of the power output. So I still believe at it would be safe to have no grid screen resistor if Vg2 is stable.
Special car must be on the g2 in a guitar amp, because all the data sheets have the following remarq for the EL34 (it is also true with other tubes): "When using a sinusoidal input signal care should be takent not to exced the maximum admissble Wg2" http://www.mif.pg.gda.pl/homepages/frank/sheets/030/e/EL34.pdf
A guitar have not a sinusoidal signal, but it IS SINUSOIDAL when we get in larsen. So in fact a sinusoidal generator is a very good instrument to test the limit of a guitar amp. For you safety and the safety of the parts of the output stage, it is good to have a fuse in the cathode circuit.
I have done a few guitar amps with 807s, the first one was very simple, the last one have a regulated supply for Vg2, and no one have screen grid resistor. The first one have a Va of 530V at full output for 300V Vg2, and the last one have Va=600V and Vg2=250V for the same full output. The difference is at the sound of the last one have a much higher dynamic as the sound of the first one, and that even in clipping.
The 807 is a 6L6 with military quality and higher absolute rating. It can work with 600V on the anode at full output, and with intermittent service with 750V at full output.
It is no indication about a screen grid resistor in the 807 data sheets. But the RCA data sheet say at Vg2 must be made from an independant power supply or a voltage divider from the anode voltage (in order to have a good regulation).
I know at the EL34 is not the same tube, but I know at, if I was buying an amp with a such tube, I would try to use a good power supply for the g2 and no screen grid resistor. Only the practice can determine if it will work or not, and the gain in dynamic we get in all cases with a good power supply for Vg2 is worth the try.
EL34 Screen Resistors
From experience:
EL34 Triode Mode - use 150R minimum
EL34 Ultralinear - Use 1K minimum
EL34 Pentode Mode - use 150R minimum
In Ultralinear Mode the grid stopper action of the resistor is more important than the peak power limiting function - so get that resistor body hard up against the screen pin of the tube socket.
I have a parallel push pull EL34 amp using VDV2100 Toroidal Output Trannies. With 150R screen resistors and Ultralinear Mode it had HF bursts. Driving a sine wave in and looking on an oscilloscope, bursts of RF were clearly (sorry EC8010) seen just after the peak. These bursts were NOT evident in Triode Mode or Pentode Mode.
In Ultralinear Mode 1K was used - this cured the RF bursts.
Cheers,
Ian
From experience:
EL34 Triode Mode - use 150R minimum
EL34 Ultralinear - Use 1K minimum
EL34 Pentode Mode - use 150R minimum
In Ultralinear Mode the grid stopper action of the resistor is more important than the peak power limiting function - so get that resistor body hard up against the screen pin of the tube socket.
I have a parallel push pull EL34 amp using VDV2100 Toroidal Output Trannies. With 150R screen resistors and Ultralinear Mode it had HF bursts. Driving a sine wave in and looking on an oscilloscope, bursts of RF were clearly (sorry EC8010) seen just after the peak. These bursts were NOT evident in Triode Mode or Pentode Mode.
In Ultralinear Mode 1K was used - this cured the RF bursts.
Cheers,
Ian
Gingertube,
Most interesting.
Can you recall the frequency of the bursts approximately? Can one localise this to the UL operation, or, if global NFB was used, could this have been feedback instability which was (perchance) cured by the 1Ks but which could also have been cured by some other feedback phase corrective measure?
I am asking because Mullard (I think it was) gave graphs of distortion for EL34s with and without the 1Ks; the latter would have been difficult if your experience was related to local instability (unless they did this mathematically, which I doubt - it was before the days of CAD). Also, I have used EL34s in UL without resistors before but, like with other tubes, had to put an RC between anode and G2 each side because of the output tranny's characteristics. GEC also does this occasionally in UL circuits using KT66 and KT88 in their book "An Approach to Audio Amplifier Design" (1994). There they mention that the values are transformer-oriented, as I have also experienced.
Not questioning your findings, just looking at other possible explanations.
Regards.
Most interesting.
Can you recall the frequency of the bursts approximately? Can one localise this to the UL operation, or, if global NFB was used, could this have been feedback instability which was (perchance) cured by the 1Ks but which could also have been cured by some other feedback phase corrective measure?
I am asking because Mullard (I think it was) gave graphs of distortion for EL34s with and without the 1Ks; the latter would have been difficult if your experience was related to local instability (unless they did this mathematically, which I doubt - it was before the days of CAD). Also, I have used EL34s in UL without resistors before but, like with other tubes, had to put an RC between anode and G2 each side because of the output tranny's characteristics. GEC also does this occasionally in UL circuits using KT66 and KT88 in their book "An Approach to Audio Amplifier Design" (1994). There they mention that the values are transformer-oriented, as I have also experienced.
Not questioning your findings, just looking at other possible explanations.
Regards.
Johan,
I did'nt actually measure the frequency. Was probably up toward 1MHz. The oscillation was definitely due to the feedback via the Ultralinear connection. There was NO global feedback being used. It had bursts of oscillation with 150R screen resistors BUT not with 1K. I did'nt try values in between.
Remember though, I'm using VDV2100 Toroidal Output Transformers which have huge bandwidth. The problem may not exist with lesser bandwidth output trannies.
Menno Vandervenne confirmed the problem and my fix (increase screen resistors from 150R to 1K) on both the VDV70/100 and UL40-S2 Amps in Ultralinear Mode and there is a note on his website labelled "Important Screen Grid Modification" (dated 2004) to this effect.
Cheers,
Ian
I did'nt actually measure the frequency. Was probably up toward 1MHz. The oscillation was definitely due to the feedback via the Ultralinear connection. There was NO global feedback being used. It had bursts of oscillation with 150R screen resistors BUT not with 1K. I did'nt try values in between.
Remember though, I'm using VDV2100 Toroidal Output Transformers which have huge bandwidth. The problem may not exist with lesser bandwidth output trannies.
Menno Vandervenne confirmed the problem and my fix (increase screen resistors from 150R to 1K) on both the VDV70/100 and UL40-S2 Amps in Ultralinear Mode and there is a note on his website labelled "Important Screen Grid Modification" (dated 2004) to this effect.
Cheers,
Ian
Sorry to dig out such an old thread. At least I'm searching before asking... 
Fantastic discussion here. I especially like Gingertube's recommendations for EL34 tubes in various topologies. Would anyone care to make a similar set of recommendations for the KT88 (especially in UL configuration)?
I'm rebuilding a set of Dynaco Mk III blocks. They originally had no resistors on the screens, but I'm thinking it might be a good idea to add them. KT88 aren't cheap, and it seems like screen resistors help tubes live longer lives.
Fantastic discussion here. I especially like Gingertube's recommendations for EL34 tubes in various topologies. Would anyone care to make a similar set of recommendations for the KT88 (especially in UL configuration)?
I'm rebuilding a set of Dynaco Mk III blocks. They originally had no resistors on the screens, but I'm thinking it might be a good idea to add them. KT88 aren't cheap, and it seems like screen resistors help tubes live longer lives.
Hi Ty,
Beam tubes like KT66, KT88 are generally less prone to instability than pentodes (EL34) in my experience. The "shaded" construction of the G2 has a lot to do with it. I have not used KT88s extensively, but have quite some literature on it and the general value of Rg2 used is 100 - 220 ohm - this also in circuits by the manufacturers. I have grown up with the principle that manufacturers would not willingly publish less than adequate designs.
I would therefore suggest that you use those values. I have used a lot of KT66s and even more 6L6s (similar construction as KT88) in a wide variaty of designs and never had tendency toward instability. As said by previous contributors much depends on the characteristics of the output transformer, but ill effects of such are mostly better cured by some phase correcting network over the output transformer. GEC often uses 1 - 2nF between A-G2 in their UL designs; occasionally a resistor of 1 - 1,5K is inserted in series with the capacitor. It is not known what output transformers they used.
If you are fortunate enough to have a scope and signal generator, you could look at the response, which will tell you easily enough about impending supersonic peaks/oscillation (such tests without NFB to begin with).
I am not sure whether any practical value of Rg2 will extent tube life noticably as such. Emission mainly depends on the magnitude of cathode current drawn unless other electrodes work at rediculous voltages/dissipation.
Beam tubes like KT66, KT88 are generally less prone to instability than pentodes (EL34) in my experience. The "shaded" construction of the G2 has a lot to do with it. I have not used KT88s extensively, but have quite some literature on it and the general value of Rg2 used is 100 - 220 ohm - this also in circuits by the manufacturers. I have grown up with the principle that manufacturers would not willingly publish less than adequate designs.
I would therefore suggest that you use those values. I have used a lot of KT66s and even more 6L6s (similar construction as KT88) in a wide variaty of designs and never had tendency toward instability. As said by previous contributors much depends on the characteristics of the output transformer, but ill effects of such are mostly better cured by some phase correcting network over the output transformer. GEC often uses 1 - 2nF between A-G2 in their UL designs; occasionally a resistor of 1 - 1,5K is inserted in series with the capacitor. It is not known what output transformers they used.
If you are fortunate enough to have a scope and signal generator, you could look at the response, which will tell you easily enough about impending supersonic peaks/oscillation (such tests without NFB to begin with).
I am not sure whether any practical value of Rg2 will extent tube life noticably as such. Emission mainly depends on the magnitude of cathode current drawn unless other electrodes work at rediculous voltages/dissipation.
Some EL 34 are in fact beam tetrode: wikipedia.org: Beam_tetrode.
One that made an amplifier must understand that with any kind of tube (6L6, KT66, KT88, EL34, whatever) it is many differnet brand and that each of those brands of the same tube can have different data. So one must have the data sheet for the brand he is using. If it is not possible to find this data sheet, the only reliable way is to measure the tubes (with respect for the max rating).
Another thing is that I found a very good reading on the beam tetrode by O.H.Shade from RCA (the gui that made the 6L6)
http://www.clarisonus.com/Archives/TubeTheory/Schade 1938 Beam Power Tubes.pdf]
This paper talk about the internal of the triode, pentode and beam tetrode.
About the differences of a pentode and a beam tetrode, one is very important: a beam tetrode is made to work well at relatively low anode voltage (250-350 V) but am equivalent pentode will work well at higher anode voltage. That implies that it is not possible to take a circuit made for a few KT88, replace them by EL34 and expect the same result.
Another difference is that in contrast with standard practice with pentodes, the plate load for a beam tetrode is not selected at minimum total distortion, but for a minimum of higher order distortion, because the design of the tube allows nearly maximum power output to be obtained with such loads.
Also, a practical way to determine if a circuit need screen grid resistors is to remove one (I assume 2 tubes). If the sonic characteristic of the amplifier doesn't change, you can try to remove the second resistor. And again, if the sonic characteristic of the amplifier doesn't change, your amplifier don't need them.
One that made an amplifier must understand that with any kind of tube (6L6, KT66, KT88, EL34, whatever) it is many differnet brand and that each of those brands of the same tube can have different data. So one must have the data sheet for the brand he is using. If it is not possible to find this data sheet, the only reliable way is to measure the tubes (with respect for the max rating).
Another thing is that I found a very good reading on the beam tetrode by O.H.Shade from RCA (the gui that made the 6L6)
http://www.clarisonus.com/Archives/TubeTheory/Schade 1938 Beam Power Tubes.pdf]
This paper talk about the internal of the triode, pentode and beam tetrode.
About the differences of a pentode and a beam tetrode, one is very important: a beam tetrode is made to work well at relatively low anode voltage (250-350 V) but am equivalent pentode will work well at higher anode voltage. That implies that it is not possible to take a circuit made for a few KT88, replace them by EL34 and expect the same result.
Another difference is that in contrast with standard practice with pentodes, the plate load for a beam tetrode is not selected at minimum total distortion, but for a minimum of higher order distortion, because the design of the tube allows nearly maximum power output to be obtained with such loads.
Also, a practical way to determine if a circuit need screen grid resistors is to remove one (I assume 2 tubes). If the sonic characteristic of the amplifier doesn't change, you can try to remove the second resistor. And again, if the sonic characteristic of the amplifier doesn't change, your amplifier don't need them.
Dominique_free,
Yes, on other threads this strange EL34 was mentioned (I think by Kevinkr) but I cannot recall the particulars. I have never seen it and neither had many others, so it appears to be rare. I would certainly frown on such a design - what are the characteristics in the end - if indeed it is a true beam tube?
In the same vein I would generally frown on different makes of the same tube (number) having different characteristics, if I understood you correctly. I would accept that such can exist; anything is possible these days. But an EL34 should be an EL34 otherwise it must be called something else. (It sounds a little like calling asperin paracetemol simply because both relieve pain.) I have experienced that the Russians can be quite free with their designations.
Then thanks for bringing the Scade article to light. It has the ring-of-truth and authority of an original, and though vintage I would consider it essential reading for everyone serious about power tube design. Quite a few urban legends are exposed therein. Also, doubt is raised about the degree to which some "modern" beam tubes (flat anodes etc.) still conform to lowest distortion design.
And then, let us note that the 6L6 design is 70 years old next month!
I appreciate your intentions in your last paragraph, but must point out that G2 oscillation can also be spurious (e.g. on peaks of cycles only). In such a case it may not be noticable immediately, only causing listener fatigue later, depending on the volume and type of music played. It will be advisable to rather check for this with an oscilloscope, but which of course may not be available to many.
But good post by you.
Yes, on other threads this strange EL34 was mentioned (I think by Kevinkr) but I cannot recall the particulars. I have never seen it and neither had many others, so it appears to be rare. I would certainly frown on such a design - what are the characteristics in the end - if indeed it is a true beam tube?
In the same vein I would generally frown on different makes of the same tube (number) having different characteristics, if I understood you correctly. I would accept that such can exist; anything is possible these days. But an EL34 should be an EL34 otherwise it must be called something else. (It sounds a little like calling asperin paracetemol simply because both relieve pain.) I have experienced that the Russians can be quite free with their designations.
Then thanks for bringing the Scade article to light. It has the ring-of-truth and authority of an original, and though vintage I would consider it essential reading for everyone serious about power tube design. Quite a few urban legends are exposed therein. Also, doubt is raised about the degree to which some "modern" beam tubes (flat anodes etc.) still conform to lowest distortion design.
And then, let us note that the 6L6 design is 70 years old next month!
I appreciate your intentions in your last paragraph, but must point out that G2 oscillation can also be spurious (e.g. on peaks of cycles only). In such a case it may not be noticable immediately, only causing listener fatigue later, depending on the volume and type of music played. It will be advisable to rather check for this with an oscilloscope, but which of course may not be available to many.
But good post by you.
The 807 datasheet specifically mentions the need for anode stoppers otherwise high frequency oscillation can be expected. I would think that inexperienced design and layout, combined with the natural tendency of some pentodes to oscillate may be at the root of the listener fatigue experienced by some. Most builds will never have the benefit of a scope which would show up these faults in an instant. In the end many of these builds get stripped down without ever getting to the root of the problem.
Look at most transmitter radio builds and you will usually find the top cap/anode well terminated with at least a 10ohm resistor with 10 turns of wire wound around it.
Anything that applies to the anode is probably doubly so for the g2. Anode and screen stoppers seem like a basic precaution against these issues.
Shoog
Look at most transmitter radio builds and you will usually find the top cap/anode well terminated with at least a 10ohm resistor with 10 turns of wire wound around it.
Anything that applies to the anode is probably doubly so for the g2. Anode and screen stoppers seem like a basic precaution against these issues.
Shoog
Quite a bit of misinformation going on in this thread... and even though it's old, it's worth trying to clean up.
EL34s are pentode tubes with a true third grid wire. They are NEVER beam types. There are EL34 electrically equivalent beam tetrode tubes such as the 6CA7 and KT77, but they are not actually EL34s, even though you can drop them right in to the tube socket and go.
And while Wikipedia is a great thing, you cannot simply trust everything on there as anyone can change anything there. I could go on their right now and say I invented the vacuum tube, then someone could read it and post it here, but it wouldn't make it so. Would it?
As for the screen grid resistor, it serves to reduce screen current, reduce distortion at high output and to prevent oscillations. This is clearly covered in the Mullard amp book...Now, you can build a great amp without any screen grid resistors. So, it depends on the circuit and its use as to whether you need these and their value.
As tubelab said about the guitar amps, this is the most common place to see screed grid resistors because guitar amps are run for hours at 110%... you need to make sure in this case that the screen grid dissipation is not exceeded. That is exactly what Marshall is doing with their 1k resistors. No guitarist wants a tube to fail when their rocking out Voodoo Chile on stage. You know?
If you are using true pentode output (connecting screens to B+), I would recommend using a screen grid resistor quite highly. As for UL outputs, it is probably far less critical and certainly doesn't require as large of a resistor as the screen follows the plate via the transformer winding... and triode connected, again, yes... you should use a resistor between the plate and the screen grid. As for exact values, that all depends on the circuit and it's use... and that part is up to the amp designer builder via theory or testing.
You make it sound so cool!
To me this implies that the beam tubes have higher screen dissipation/voltages than an EL34. However, this is NOT true. All the common beam type tubes used in audio have lower screen dissipation and voltage ratings than the EL34. Strictly speaking, if you are having problems with your screen grid circuit, moving to a beam tube should make it worse, not better. I think some time spent troubleshooting the problem would be better than swapping tube types. Fix the problem, then switch tubes if that is indeed what you desire to do.Johan Potgieter said:
Dominique_free said:
EL34s are pentode tubes with a true third grid wire. They are NEVER beam types. There are EL34 electrically equivalent beam tetrode tubes such as the 6CA7 and KT77, but they are not actually EL34s, even though you can drop them right in to the tube socket and go.
And while Wikipedia is a great thing, you cannot simply trust everything on there as anyone can change anything there. I could go on their right now and say I invented the vacuum tube, then someone could read it and post it here, but it wouldn't make it so. Would it?
As for the screen grid resistor, it serves to reduce screen current, reduce distortion at high output and to prevent oscillations. This is clearly covered in the Mullard amp book...Now, you can build a great amp without any screen grid resistors. So, it depends on the circuit and its use as to whether you need these and their value.
As tubelab said about the guitar amps, this is the most common place to see screed grid resistors because guitar amps are run for hours at 110%... you need to make sure in this case that the screen grid dissipation is not exceeded. That is exactly what Marshall is doing with their 1k resistors. No guitarist wants a tube to fail when their rocking out Voodoo Chile on stage. You know?
If you are using true pentode output (connecting screens to B+), I would recommend using a screen grid resistor quite highly. As for UL outputs, it is probably far less critical and certainly doesn't require as large of a resistor as the screen follows the plate via the transformer winding... and triode connected, again, yes... you should use a resistor between the plate and the screen grid. As for exact values, that all depends on the circuit and it's use... and that part is up to the amp designer builder via theory or testing.
You make it sound so cool!
A friend that is making tubes amplifier in Geneva was having a few of them it was 20 years ago. I can ask him if he remember which brand that was.Johan Potgieter said:
According to him, they was exactly the same. And they was true beam tubes.Johan Potgieter said:
Johan Potgieter said:
An EL34 remain an EL34, but if you compare the datasheet for a Mullard EL34 and a philips EL34, you will find some differences. As example, the Mullard have a mutual conductance of 11 mA/V when the philips have a mutual conductance of 12.5 mA/V.
You can take any kind of tubes, and you will find some differences between manufacturers. It is also some other differences between 2 brands of the same manufacturer. The machines making the valves was the same in most cases, but it was always some small differences with the settings of those machines. A difference of a few 1/1000 of millimeters in the settings of the control grid and the characteristics will vary.
And that apply for any kind of components. I mesurd tubes and transistors, but I never get exactly the same curves as in the data sheets, and that even with very good equipments.
Yes, it is a vintage reading, but quite essential. If you look at modern beam tubes, you can find them in power of a lot of kilowatts and anode voltage of a few kilovolts. They are made for high power high frequency transmitter. So, why a little 25 W 800V rated beam tube like the EL34 would be impossible?Johan Potgieter said:
In theory, it would even be possible to use such high power tubes to make an audio amplifier. It will work like hell (the biggest problem would be the power supply), but the sound will not be good because they are not intended for audio work at all.
You are absolutely right. I was writing too fast. It was in fact a statement in Shade's article.Johan Potgieter said:
Thank you.Johan Potgieter said:
I agree. It depend on the circuit in use.Cycline3 said:
I know that, I am playing guitar myself.Cycline3 said:
I made a guitar amplifier with 2 pieces of 807 and an EL84 in triode as driver (class A with transformer). I made a lot of experiment with it. I never used a screen grid resistor with that amp, but a separated supply.
My last try was to use a regulated supply for the G2 of the output stage. The dynamic was outstanding, but the G2 was too hot because the control grid voltage was not falling enough at full output. So, I will try to improve the G1 circuit if I want to be able to keep this regulated supply. But this is another and difficult matter.
The same montage (regulated G2 supply) will not be a problem on a hifi amplifier because such amplifier do have much more conservative settings in regards to maximum rating of the tubes as a guitar amp. It is just costing much more as a screen grid resistor, but the sonic result is a far better dynamic. In fact, a regulated G2 supply for an hifi amp would be an overkill because the G2 current will never come as high as in a guitar amp. A good separated supply would be enough (can even be the same as the driver supply) in most cases.
The supply of the G2 will affect directly the dynamic of the sound. Good regulation implies good dynamic, not so good regulation implies not so good dynamic. It is why I will prefer a circuit that doesn't need screen grid resistors.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.