|
|
Hits: 6225 Replies: 2
1V6 Curve Traces and Gammatron Action
|
|||||||||
Joe Sousa
30.Apr.13 |
1
Fellow Radiophiles: The 1V6 filamentary Triode-Pentode frequency converter makes intentional use of "Gammatron" control to couple the local oscillation in the Triode half to the Pentode mixer half. The oscillation injection is done from the oscillating triode grid to the space charge under the pentode grid on the far side of the filament that is shared by the triode and pentode. The Gammatron configuration triode is one where there is a cathode, which is preferably filamentary, surrounded by two plates. One plate serves as the conventional triode plate, while the other serves as control element in an analogous fashion to a control grid. Gammatron was a trade name by Heintz and Kaufman, who commercialized a line of transmitting Gammatron structured triodes before World War II. However, the plate-cathode-plate structure was already included in Lee DeForest's triode Patent from 1908. Don Black started a recent thread in the Tube Collector's Association newsgroup about the coupling mechanism from the local oscillator in the triode half to the mixing pentode half. A simple inspection of the electrode disposition makes it clear that Gammatron action is very likely between the triode grid on one side of the very thin cathode filament and the pentode on the other side of the filament. The curve traces below prove the Gammatron action. The electrode sequence and pinout in the 1V6 are as follows: 1-Pentode plate 4-Pentode suppressor grid (tied to filamentary cathode minus) 2-Pentode screen grid 3-Pentode control grid 4-7-Filamentary cathode 5-Triode control grid 6-Triode plate The Raytheon data sheet makes no reference of direct control of the triode half over the pentode half, except for stating a conversion transconductance of 200uS and the usual application circuit in frequency converter configuration has no explicit external coupling between the local oscillator triode and the mixing pentode. The extensive curve sweeps on the 1V6 give a sense of it's unusual internal operation. The following curves include standard triode and pentode sweeps as well as sweeps which account for the interaction of voltages between the triode and pentode halves. Standard Triode and Pentode curves without interaction. As shown in the following first two plots, the triode-connected Pentode appears to be more efficient than the Triode, with higher gm and mu in triode connection. But some of the Pentode gm is lost to it's screen current. The two curves on the right compare Pentode operation at 90V and 45V.
Triode and Triode-connected Pentode curves with interaction from other half Now things get more interesting as the influence between triode and pentode is accounted for. Over 80% of this cross-influence is accounted for by the the two control grids G1T and G1P. The remaining electrodes in the triode and pentode are further decoupled by their μ factors. This influence across the filamentary cathode is analogous to the operation of a Gamatron triode with anode plate on one side of the filamentary cathode and the controlling Gamma plate on the other side. The first plot on the left shows the effect of the pentode control grid G1P on the Triode, while the triode grid is grounded. This effect is surprisingly high, to the point that G1P has nearly the same effect on the triode current as G1T. Compare the left-most plot below to the left-most plot above with nearly identical gm and μ. The second plot shows the combined effect on of G1T+G1P on triode current, thus doubling gm and μ. The two plots on the right show a similar effect by the triode control grid on the triode-connected pentode. The effect of G1T on the pentode is sligtly lower with gm=300uS, but the combined effect of G1P+G1T on the pentode is the same as on the triode, with a gm=1000uS.
Various Gammatron configurations at High voltage (45V) and low voltage (<10V) The first plot below shows the effect of the triode control grid G1T on the pentode in the configuration that is recommended by the data sheet with G2P=45V as a frequency converter. The triode grid needs to present an oscillation greater than 4Vp-p in order to bring the pentode conduction to 10% of it's maximum value. The bunching of curves on the bottom left indicates a remote cutoff effect of G1T on the pentode. This remote cutoff effect recalls the behaviour of the Russian tripple control rod pentode 1zh37b. The 1zh37b also has some of this Gammatron effect where the space charge at the filament is directly controlled by more than one rod electrode. The last three plots explore the possibility of operating the 1V6 below 10V. Note the horizontal scale only spans 10V for the last three plots, instead of 100V for all other plots. The second plot sweeps the triode plate and control grid wired together as a Gammatron anode, while controlling the current with the pentode control grid G1P. The third plot explores the operation of the Triode half arranged as a Tetrode with G1T=+6VDC and G1P serves as control grid of this Tetrode-configured Triode half. The last plot shows the converse operation, where the Pentode is wired as a low voltage Tetrode with G1P=+6V, G2P is wired to the Pentode plate and G1T serves as the control grid for this low voltage tetrode-configured Pentode half.
ConclusionsAfter taking the curves, some of the insights helped explain the operation of conventional tubes, if the two half grids on either side of the filamentary cathode are thought of as two halves of a conventional grid. It became clear that the effect of one side of a conventional cylindrically wound grid extends beyond the immediate space near the filament, to the space beyond the filament. In the case of the triode in the 1V6, it's control grid and the pentode control grid have about the same transconductance (gm) to the triode plate current. Either grid has about gm=500uS, and their combined transconductane to the triode plate is 1000uS. Compare the 1st, 5th and 6ths curve traces above. One significant operational difference between the action of a grid that is located between plate and cathode and one where the cathode is located between the anode and grid is that the latter has remote cutoff characteristic. It seems to be hard to completely pinch off the space charge on the far side of the filamentary cathode. The gm and μ of the gammatron effect were comparable to the gm and μ in standard triode configuration. Much of this is owed to the very thin filament and the tight proximity from the filament to the grids. The last three curve traces also suggest that this tube could have been used for low voltage service while still drawing several mA of plate current. The last two plots also show that a Tetrode configuration, with well over 1mA of anode current, is possible with as little as 6V at the anode and G1 serving as the screen grid. The two plots show that either triode or pentode half can be configure as a low voltage Tetrode. The configuration is not directly equivalent to a space charge grid Tetrode because G1 in the active side still serves the screening purpose of Miller capacitance reduction from the output plate to the other controlling G1 on the far side of the filament. The other distinguishing feature with the use of G1 as a screen grid from a space charge grid is that this screen grid draws less current than the anode. The only passing similarity of G1 used as a screen grid to a conventional space charge grid is that G1 is adjacent to the cathode. The remaining operation of the low voltage Tetrode configuratiion is like that of a conventional Tetrode. Best regards, -Joe
|
||||||||
Joe Sousa
05.May.13 |
2
Fellow Radiophiles: One last measurement to help illustrate the effect of both control grids around the filamentary cathode. This curve trace family shows the pentode half and triode half in parallel. The vertical scale is increased to 1mA/div. The net parallel pentode/triode gm is 1500uS. Note how relatively little gm was contributed when compared to the 8th plot above, where the pentode is swept in triode connection with both control grids in use and gm=1000uS.
The net parallel pentode/triode mu=μ=11 is approximately the average of the independent triode mu=8 and pentode mu=15. This stands to reason as mu depends primarily on linear field attenuation between anode and control grids. (James C. Maxwell used mu to describe how much electric field could penetrate a metalic mesh. When a grounded metalic plane mesh is placed between two metalic planes, one of which is also grounded, the electrostatic field between the two original plates is attenuated by mu. This expains why mu is a universal tube paramenter. Can't locate an external reference for this at the moment.)
The largest increase from the parallel triode/pentode connection was in anode current, which is more than the sum of the individual triode and pentode connected currents. At 100V and grounded control grids, this plot shows 9.5mA, while the triode alone peaks at 4.5mA in plot #6 and the triode connected pentode peaks at 3mA in plot #8. It is apparent that the positive anodes reinforce each other's currents. It is as if the two control grids had been made more positive. To a first order, and to the degree that mu is a fixed quantity, the application of anode voltage can be duplicated with the application of control grid voltage that is mu times smaller. Mu varies somewhat because the shape and location of the space charge varies under the influence of the electric field. Regards, -Joe |
||||||||
Joe Sousa
12.Jun.14 |
3
Fellow Radiophiles: The 1V6 construction lends itself, with two separate control grid halves on either side of the shared monofilar filament, to the study of tube behaviour that is not observable with a conventionally constructed single control grid around the filament. In the following series of measurements, I attempt to answer the question why there was never a dual power pentode with a shared single filament for service in Push-Pull audio output stages of battery powered radios. The reasoning for the question is that, the total cathode current in a Class A audio power push-pull stage is approximately constant, thus making one wonder if a single cathode might not be usefully shared by the two halves of a dual pentode constructed similarly to the 1V6. If this worked, it would save the power to light up one filament. In the case of Class AB push-pull operation, the total cathode current is highest at the signal peaks while one of the cathodes is also cut-off completely at the signal peaks. This too would be a reason to try to share a single cathode between the two halves of a dual pentode in push-pull audio Class AB power amplifier. As my measurements will show, the push-pull operation of the 1V6 yields less output power than simply wiring the two halves of the 1V6 in parallel. The fundamental reason for this unexpected result is the so-called Gammatron effect that was measured above, where each of the two control grids controls not only the current for it's side of the 1V6, but also has a significant control on the other half. This controlling effect on the opposite half of the 1V6 has a degenerative effect that prevents full power from being delivered. In push-pull configuration of the 1V6, as one control grid turns off one side, it also diminishes the gain of the grid of the other side that is trying to get turned on more, so that push-pull operation still occurs, but with diminished overall power output. A related question, is why is a single filament never shared by the two anodes in a full wave filamentary rectifer. If this worked well, it would save the power to light up two separate filament sections. The 1V6 shows that sharing the single filamentary cathode makes poor rectifiers out of the two control grid halves. Individual Control Grid Sweeps and Anode Currents The IV curve describing the effect of the control grid of one half of the 1V6 on the other half has two regions as shown in the following measurement. These two plots show simultaneously in each graph, anode current for the triode half and for the pentode half (triode connected A=G2) of the 1V6.
The solid lines show the Pentode anode currents and the dashed lines show the Triode anode currents.
In the left plot, the pentode half control grid is kept grounded, while the triode half control grid is swept from 0V to -25V. In the right plot the control grid roles are swapped: the triode half control grid is kept grounded, while the pentode half control grid is swept from 0V to -25V. The three sets of curves in Red, Blue and Black are for different anode loads. Red is the anode load resistor voltage drop with 10k loads to 45V on both anodes. Blue is anode current with a fixed 45V at the anodes. Black shows load resistor drops with 36k at the triode anode and 60k at the anode+screen of the Pentode. Both loads to 90V. The reason for the 36k/60k case is to equalize the voltage gains between the two halves of the 1V6. In order to compare the shapes of these three different anode loading configurations, all curves are scaled to one unit with 0V control grid bias. The legend within the plot shows the paramentric value that is equivalent to 1 unit at zero volts. I was a little surprised to see that there is only a mild effect on the IV shape from the anode loading variations. I expected that if the Anode can move a lot under a high impedance load, it might have a great effect on the shape of IV characteristic as compared to the case where the anode voltage is kept fixed at 45V and anode current is measured. The steeper curves near the vertical axis show the classic gm characteristic in both plots for a grid controlling the current on it's own half of the 1V6. Now for the interesting shape in the IV curves note how flat the top of the curve of the opposing half of the tube is. On the left plot, sweeping the triode grid from 0V to -4v changes the current only about 10%. Beyond -4V at the triode grid, the pentode current starts to finally drop in a conventional gm shape. Note there is even a slight increase in Pentode current on the solid-blue anode current curve as the triode voltage is swept from 0V to -2V. The sets of curves towards the center of the plots show the two kinds of influence of the opposing grid in it's Gammatron-style operation. The first effect when the grid goes negative on one side is to push cathode charge to the other side to the point of producing a slight anode current increase in the solid blue curve. The second effect is for the field of this grid to reach around the space charge and actually start reducing the current of the opposing half, as happens on the left plot beyond -2.5V. Both effects are simultaneously present, but the space charge gets reshaped and shrinks wih more negative grid voltages, thus making the second reach-around effect the dominant one beyond -4V. Anode Current Differences For our purposes of push-pull operation, the desirable portion of the curves is when a control grid has a strong control on it's half, but little effect on the other half. In the first plot this the region between 0V and -5V.
The second plot has an even smaller useful region for push-pull operation, with less flattening of the IV curve of the opposing half.
The useful output current for push-pull operation is thus the difference between the two anode currents as recalculated from the data of the plots above and plotted in the two following plots as a difference of anode currents.
Note how the push-pull transfer function is non-monotonic. It becomes necessary to limit the swing of each grid to -4V on the left plot and to -2.5V on the right plot. The reason for two different voltage values is that the mu of the pentode is about twice the mu of the triode. Keep in mind that in each of these two plots one grid is kept grounded while the other grid is swept. This is not what would happen with the differential drive of a push pull stage where both grids move in opposite directions. If both grids move in opposite directions, the curve peak would move out as the grid that goes positive prevents current reduction on it's side, as the negative-going grid on the other side tries to reduce current on the opposing side.
Poor Push-Pull Operation
The following plot compares differential anode currents as the two control grids are driven differentially with different bias voltages (Vcm=common mode). A conventional single ended current with both anodes tied together and both grids driven simultaneously is also plotted in the brown trace near the vertical axis. The voltage that was applied to the control grid on the pentode half was attenuated by half to equalize the mu of triode and pentode. The anode voltages were kept at a fixed 45V, so they are in the same conditions as the blue curves above.
The single ended brown curve shows a p-p current swing of 2800uA with a 0-2V drive.
The optimal bias is -2V at the blue curve, yielding a p-p current swing of only 2200uA with a 8Vp-p differential grid drive. While this demonstrates Push-Pull operation of the 1V6, it also proves how much poorer push-pull operation is than simple single ended operation, with reduced output as well as reduced input sensitivity.
Perhaps some clever dual grid design would have widened the range where one grid does not disrupt the operation of the other half, as hinted in the flat topped portions of the first plots, but simply splitting a conventional control grid in two, as was done inside the 1V6, does not produce a useful push-pull tube.
Poor Full Wave Rectification
One last aspect of the Gammatron effect is the degradation of voltage drop in full-wave rectifier application. This can be verified with the 1V6 by forward biasing one control grid while reverse biasing the other control grid.
The black traces in the following plots show the forward voltage drop on one grid that is biased with about 1uA while the other grid is reverse biased. Note how -15V reverse bias on the left plot increased the voltage drop on the other grid from 500mV to 2200mV. This is a very undesirable characteristic in a full wave rectifier and serves to explain why all filamentary full wave rectifiers were made the two separate filament sections for each of two anodes. The filament sections in conventional full wave filamentary rectifiers were usually connected in series as is convenient for full wave rectification.
Conclusion
The so-called Gammatron effect, where one element on one side of a filamentary cathode controls the operation of a similar element on the other side of filament, is useful for local oscillator injection as was the design goal of the 1V6. The measurements of the previous posts also show how useful Gammatron operation is, to draw large anode currents with very low anode voltages, thus making low anode voltage operation practical. But the Gammatron effect would also disrupt the operation of a push-pull pair that tried to share the same filament and it would also greatly degrade the forward drop in full-wave rectifier service.
Best regards,
-Joe
|
Data Compliance | More Information |