Home |
Search |
Today's Posts |
#1
|
|||
|
|||
Quantifying Miller effect in multi-grid tubes
About any introductory material on vacuum tubes says that tetrodes, beam
tetrodes, and pentodes have reduced Miller effect, but nowhere have I been able to find a formula or even an estimate of how much the reduction is. How then, can I determine an appropriate minimum output impedance of the grid driving circuit for a large tetrode (4X150A, well, it's small in it's category of transmitter tubes, but it's large compared to most tubes used in audio) that I plan to use in a plate follower configuration? I've also sent this question to Broskie of tubecad.com, but I was hoping to get more than one opinion. |
#2
|
|||
|
|||
Prune wrote: About any introductory material on vacuum tubes says that tetrodes, beam tetrodes, and pentodes have reduced Miller effect, but nowhere have I been able to find a formula or even an estimate of how much the reduction is. Well, the reduction is virtually complete reduction. A garden variety of pentode or beam tube when connected triode may have say 10pF measured between anode and grid when no signal is present. When the gain is say 4, as it may be with a KT88 in triode, Miller says Miller C in will be 4 x 10pF = 40 pF, which limits the effectiveness of triodes for RF amplification. But when beam connection is used, the static Cin remains at say 10 pF, and virtually does not change bcause the screen screens off the effect of the anode voltage changes. There is more about it in the old books such as RDH4, Terman's Radio Engineering etc. How then, can I determine an appropriate minimum output impedance of the grid driving circuit for a large tetrode (4X150A, well, it's small in it's category of transmitter tubes, but it's large compared to most tubes used in audio) that I plan to use in a plate follower configuration? Back to basics and the data for the answers. I've also sent this question to Broskie of tubecad.com, but I was hoping to get more than one opinion. There will be only one true answer though. You could always set up a tube and **measure** what the Cin was..... Then you'll know for sure. Patrick Turner. |
#3
|
|||
|
|||
Prune wrote:
About any introductory material on vacuum tubes says that tetrodes, beam tetrodes, and pentodes have reduced Miller effect, but nowhere have I been able to find a formula or even an estimate of how much the reduction is. How then, can I determine an appropriate minimum output impedance of the grid driving circuit for a large tetrode (4X150A, well, it's small in it's category of transmitter tubes, but it's large compared to most tubes used in audio) that I plan to use in a plate follower configuration? I've also sent this question to Broskie of tubecad.com, but I was hoping to get more than one opinion. The Miller C of the 4X150A is only 0.06 pf as given in the RCA Transmitting Tube Manual TT-4. Compare that to a common power triode such as the 300B where you have 15 pf! Even a 6L6GC has a Miller C of 0.6 pf. Input C for the 4X150A is 15.7 pf & I would think that in any ordinary circuit including what you propose that would be the primary concern so far as the driver is concerned. Beyond that you will have an additional capacity due to wiring & the output C of whatever driver you will use. Also, do you ontend to run into G1 current (Class AB2)? If so, you will need a low Z driver anyway. The output C of the 4X150A is given as 4.3 pf. Cheers, John Stewart |
#4
|
|||
|
|||
Thanks to both Patrick and John (is that your real name, or are you just a
fan of The Daily Show)? The datasheet that some Eimac engineer faxed me has another number called Feedback Capacitance. I don't quite understand what it is. The number is very low, 0.05 pF, so I'm not sure if it matters here. Does this refer to plate-to-control grid feedback? How much of an effect does it have? BTW, the air system sockets I have, SK630, have built in screen grid capacitors, which are part of the socket structure and almost impossible to remove (I guess built in this way to remove inductances of the leads of a separate capacitor). Since they are designed in the socket for high frequency operation in mind, can I assume that I can just ignore them and they won't give me a problem at audio frequencies? |
#5
|
|||
|
|||
Prune wrote:
Thanks to both Patrick and John (is that your real name, or are you just a fan of The Daily Show)? That is my real name. Not sure about The Daily Show. I don't watch the tube much. The graphics are usually much better outside!! The datasheet that some Eimac engineer faxed me has another number called Feedback Capacitance. I don't quite understand what it is. The number is very low, 0.05 pF, so I'm not sure if it matters here. Does this refer to plate-to-control grid feedback? Yes. And there really is feedback. Can be negative or positive, depending on the kind of impedances in the grid & plate circuits. They in turn are frequency dependent. The Feedback Capacitance of 0.05 pf referenced on your Eimac datasheet is the same as what I called the Miller C in error on the earlier post. In fact the Miller C is really (Gain + 1) * (grid-plate capacitance). See also http://frank.pocnet.net/sheets/088/4/4X150A.pdf On the downloaded Eimac data sheet they show 0.03 pf as the grid-plate capacity. In any case, the Miller effect in the audio application of this tube is insignificant. Of more concern is the input C of 15.5 pf & stray wiring capacities. But even those will not be much of a problem. The Miller Effect was a real problem for early designers of RF & IF amplifiers, especially when triodes were used. The screen grid provides an electrostatic shield between the input & output circuits in a grounded (RF) cathode amplifier. The high internal impedance of a tetrode or pentode allowed large gain in amplifiers using them for IF & RF. Those characteristics were not possible with triodes. As a result, special circuits such as the Neutrodyne were needed to prevent self oscillation when triodes were used. However, the triode could be an effective, low noise RF amplifier if it were connected as a grounded grid amp. The control grid then forms an electrostatic shield between the input & output circuits. But the gain is not nearly as high as with pentodes. A way around that is the Cascode amp using two triodes. At very high frequencies the cathode lead impedance also becomes a problem. You will see two cathode connexions brought out. One is connected to the input circuit while the other goes to the output circuit. That way feedback thru the cathode circuit is kept to a minimum. If you check the various listings for grid-to-plate capacity you will find those designed for RF applications in general have very low numbers. But those directed to audio work are quite a bit higher since plate to grid feedback at audio is much less of a problem. How much of an effect does it have? BTW, the air system sockets I have, SK630, have built in screen grid capacitors, which are part of the socket structure and almost impossible to remove (I guess built in this way to remove inductances of the leads of a separate capacitor). Since they are designed in the socket for high frequency operation in mind, can I assume that I can just ignore them and they won't give me a problem at audio frequencies? If you intend to use the tubes in ordinary pentode connexion, no problem. If you want to go ultralinear (UL), that capacity may get in your way. It would be interesting to know its size. See 4X150A/4000 Eimac SKT 4X150A-4000.pdf (263707 bytes) For the socket Cheers, John Stewart |
#6
|
|||
|
|||
The screen bypass capacitor is pretty large at 1100 pF. I'm not using an
output transformer (not for dynamic speakers), and I'll probably have a separate supply for the screen grid. Still, I was considering modulating that supply to improve linearity, which is what an ultralinear connection does anyway, but I guess that would be pretty hard with such a large capacitance. And, that type of modulation makes the curves more triode- like, which is not what I'm after as I'm using the tube as a transconductance stage (voltage controlled current sink). I did find this graph of mutual characteristics, http://www.webace.com.au/~electron/tubes/FIG23.jpg showing them for several screen grid voltages. In the single ended use I have for this, it's very nonlinear around any reasonable operating point (non-positive control grid and screen grid under 400 V). I doubt cathode degeneration would make it much better. The only way to linearize before addition of NFB seems predistortion, such as by a input-signal parallel diode tube (it has curves with almost inverse nonlinearity, and can be adjusted by a ring magnet for radially structured diode tubes). I was looking at Hawksford-type error correction, but that only seems to work for stages with gain close to 1, and I don't know if it can be adapted to a high gain stage (I've seen an attempt at diyaudio, but another member argued that then it becomes equivalent to just more NFB). John Stewart wrote in : Prune wrote: Thanks to both Patrick and John (is that your real name, or are you just a fan of The Daily Show)? That is my real name. Not sure about The Daily Show. I don't watch the tube much. The graphics are usually much better outside!! The datasheet that some Eimac engineer faxed me has another number called Feedback Capacitance. I don't quite understand what it is. The number is very low, 0.05 pF, so I'm not sure if it matters here. Does this refer to plate-to-control grid feedback? Yes. And there really is feedback. Can be negative or positive, depending on the kind of impedances in the grid & plate circuits. They in turn are frequency dependent. The Feedback Capacitance of 0.05 pf referenced on your Eimac datasheet is the same as what I called the Miller C in error on the earlier post. In fact the Miller C is really (Gain + 1) * (grid-plate capacitance). See also http://frank.pocnet.net/sheets/088/4/4X150A.pdf On the downloaded Eimac data sheet they show 0.03 pf as the grid-plate capacity. In any case, the Miller effect in the audio application of this tube is insignificant. Of more concern is the input C of 15.5 pf & stray wiring capacities. But even those will not be much of a problem. The Miller Effect was a real problem for early designers of RF & IF amplifiers, especially when triodes were used. The screen grid provides an electrostatic shield between the input & output circuits in a grounded (RF) cathode amplifier. The high internal impedance of a tetrode or pentode allowed large gain in amplifiers using them for IF & RF. Those characteristics were not possible with triodes. As a result, special circuits such as the Neutrodyne were needed to prevent self oscillation when triodes were used. However, the triode could be an effective, low noise RF amplifier if it were connected as a grounded grid amp. The control grid then forms an electrostatic shield between the input & output circuits. But the gain is not nearly as high as with pentodes. A way around that is the Cascode amp using two triodes. At very high frequencies the cathode lead impedance also becomes a problem. You will see two cathode connexions brought out. One is connected to the input circuit while the other goes to the output circuit. That way feedback thru the cathode circuit is kept to a minimum. If you check the various listings for grid-to-plate capacity you will find those designed for RF applications in general have very low numbers. But those directed to audio work are quite a bit higher since plate to grid feedback at audio is much less of a problem. How much of an effect does it have? BTW, the air system sockets I have, SK630, have built in screen grid capacitors, which are part of the socket structure and almost impossible to remove (I guess built in this way to remove inductances of the leads of a separate capacitor). Since they are designed in the socket for high frequency operation in mind, can I assume that I can just ignore them and they won't give me a problem at audio frequencies? If you intend to use the tubes in ordinary pentode connexion, no problem. If you want to go ultralinear (UL), that capacity may get in your way. It would be interesting to know its size. See 4X150A/4000 Eimac SKT 4X150A-4000.pdf (263707 bytes) For the socket Cheers, John Stewart |
#7
|
|||
|
|||
Prune wrote: Thanks to both Patrick and John (is that your real name, or are you just a fan of The Daily Show)? The datasheet that some Eimac engineer faxed me has another number called Feedback Capacitance. I don't quite understand what it is. The number is very low, 0.05 pF, so I'm not sure if it matters here. Does this refer to plate-to-control grid feedback? How much of an effect does it have? Internal feedback in most amplifying devices is an inherent part of each one of them. But let us confine our thoughts to vacuum tubes. If you consider a tetrode or pentode with 0.05pF between anode and grid, then consider if you have a voltage gain of say 200 in a give circuit. Thus for a +ve going 0.25 v at the grid, there will be a -ve going voltage of -50v at the anode. So if we had a sensitive meter to measure the input capacitance to this tube we would find that there is capacitance between the grid and cathode, and to the screen, in this case held at a fixed voltage. and the the anode, whose voltage changes. The effect of the gain of 200 in the tube causes the capacitance of 0.05pF to be effectively equal to 200 x 0.05 pF = 10pF. The total of the other C measured with no signal voltages will stay the same when signal is applied, and if that other C was say 5pF, then Cin = 15 pF total If you had a 500k volume pot before the pentode with gain = 200, then the maximum series resistance in front of the tube assuming we have a low impedance CD player hooked up to the tube is 125kohms. So you have in effect a low pass filter with 125k and 15 pF, and this has a -3dB point = 159,000 / ( 125,000 x 15/0.000015 ) = 84 kHz. Thus there is a pole caused by series R and Cin. If the pot was 50k, then the maximum series R would reduce to 12.5k, so the pole would be at 840 kHz. At other points along a typical amplifier circuit there are other R&C interactions causing poles where roll off of the HF occurs. Suppose we connect the screen to the anode, but leave the same RL. Things become radically different. The anode is then allowed to indirectly affect its own electron flow. With the screen taken to a fixed voltage, the tube may only need +0.25v at grid1 to swing the anode -50v, with an increase in anode current. But when the change at the anode is the same as that at the screen, and this screen voltage change then affects the electron flow very like the control grid, so a -ve going anode voltage causes a -ve going screen voltage and this tries to reduce the electron current flow, in direct opposition to what the control g1 is doing. So the G1 signal has to be a lot more to cause the same -50v anode swing, due to having to counter what the screen is trying to do. The effect of the screen is to introduce electrostatic feedback. A typical pentode above might have a gain of only 20 when the screen is connected to the anode. The other big difference with a triode connected pentode is that that the C between grid and anode increases to perhaps 4pF, and since we have a circuit gain of say 20, the miller capacitance will be 80pF, and the other stray C will still be say 5pF, so with a triode we have a total input C = 85 pF. The use of a 500k volume pot set to the -6db position still gives 125k of series R, so the pole will be 15 kHz, and this ight be ok in a cheap old radio, but not ok for hi-fi. But a 50k pot would move the pole up to 150 kHz, and that would be fine. So you can see that cpacitance will definately affect the workings of a tube at HF, and at 10Mhz, 15pF is an impedance of only 1,060 ohms, and so some considerable current must be supplied to the input of the tube to charge and discharge this capacitance at HF. A triode with 85pF Cin would be useless at 10Mhz. unless used as a grounded grid stage, and then there is a price, we must have a very low impedance source to supply an input to the cathode. Your questions raise many more other questions than answers I have time for. BTW, the air system sockets I have, SK630, have built in screen grid capacitors, which are part of the socket structure and almost impossible to remove (I guess built in this way to remove inductances of the leads of a separate capacitor). Since they are designed in the socket for high frequency operation in mind, can I assume that I can just ignore them and they won't give me a problem at audio frequencies? It depends on the value of C and your circuit. Without working out all the circuit workings taking into account gains, stray C. and all the resistances involved, you can't assume anything. You should never assume anything except that you know nothing until you prove beyond doubt that you know something. This attitude is basis of objective scientific thought; any other attitude soon leads to circles and riddles of ignorance. You cannot learn electronic knowledge unless you take your tendency to assume things and trample all over it in hob nail boots shouting " I won't assume, I won't assume". Kick yourself in your arse each morning shouting "only learn the truth which I must find out !!" Your questions are about very basic electronic behaviour, and its great to see you asking them, but please do go read a few books about all this and do some experiments with an old tube. If you don't look, you won't know. Patrick Turner. |
Reply |
Thread Tools | |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Forum | |||
Lots Of Great Tubes For Sale | Marketplace | |||
Lots Of Great Tubes For Sale | Marketplace | |||
211 Ultra Linear PP output stage?? | Vacuum Tubes | |||
Grid current in 6C33C | Vacuum Tubes | |||
Bogen amp, 8417 tubes blow up, want my 6550 mo 100's? | Vacuum Tubes |