Russian Subminiature Tubes

ID: 200277
Russian Subminiature Tubes 
19.Sep.09 07:14

Joe Sousa (USA)
Articles: 664
Count of Thanks: 95
Joe Sousa

The following article was originally published in the June 2009 issue Vol 11 No. 3 of "Tube Collector", the bi-monthly publication of the Tube Collectors Association - TCA.

TCA is a non-profit organization and offers memberships which include a subscription to Tube Collector. TCA also maintains an active discussion in Yahoo-Groups.

Send an email to to join the Yahoo group.

I wrote the article and TC editor Ludwell Sibley edited it for publication. Lud also added an introductory article with historic background to complement my measurement oriented article.

As far as I know, these tubes were used exclusively in Soviet military applications. The RM database appears to confirm this supposition with a complete absence of commercial Soviet radios using these tubes.




Russian Subminiature Tubes

Russian Subminiature Tubes are constructed entirely differently from other subminiature tubes. The internal structure of conventional subminiature Tubes shown in figure 1 is easily recognized as a miniaturized version of the classic receiving tube structure, with a filamentary or a unipotential cathode, one or more thin wire ladders as grids, perhaps beam forming plates and, lastly, a classic anode plate.

Fig. 1 Conventional pencil tubes. SN891B-triode 5678-pentode

Fig. 1 Conventional Subminiature tubes. SN891B-triode 5678-pentode

Fig. 2 Russian Subminiature Tubes. 1ZH29B 1P24B 1ZH18B.  All pentodes.

Fig. 3 1ZH29B without glass envelope.

The Russian Subminiature Tube [2] design shown in figure 2, is hardly recognizable as a triode, tetrode or pentode. The cathode is still a conventional filament, but the remaining elements are all constructed from similar gauge metal rods. Figure 3 shows three micas; 2 at the ends and one in the center of the tube to hold the rods in place, with the central mica assuring low microphonics, and consistent element alignment.

The rods are arranged to control the electron path from the filament to the a diametrically opposed pair of flattened rods that serve as the anode. One of the anode rods is the widest visible  od left of center in Figure 3.  The anode connection is always via the top wire.

Figure 6 below, shows the path taken by electrons leaving the two cathode filaments at the center. The electrons form two pairs of longitudinal sheet beams as they are attracted by two diametrically opposed rods that serve as the Anode (the plate in a conventional tube). Three beam forming flattened rods are  on either side of the two filaments to assist the formation of the two pairs of sheet beams flowing in opposite directions to their respective Anode rods. These three flattened rods also serve as control grid G1. Two additional sets of 3 round rods are aligned with the control grid G1 rods and attract the sheet beam to implement the screen grid G2. Four pairs of round rods between G2 and the Anode are somewhat pulled away from the electron path serve to suppress secondary emission as G3. Some tube types, such as the 1ZH29B on figure 6, have additional rods at the periphery that serve as grounded shields.

Fig. 4 1ZH29B base detail. Fig. 5 1ZH29B top detail with getter cup removed. Fig. 6 1ZH29B Top detail showing electron path from filament to Anode.

F1, F2, F3: Center tapped filament.

G1: Control grid has 3 rods adjacent to 2 filaments, no grid wires.

G2: Screen grid with 6 rods. G3: Suppressor grid with 8 rods.

A: Anode made from 2 flattened rods.

S: are shielding rods at the periphery that may be grounded along with G3.

The electrical operation of a Russian subminiature pentode is recognizable as such, and the schematic symbol is still the conventional pentode. The data sheets for Russian subminiature tubes usually specify the intended frequency range as 60MHz, but some are specified for operation at 120MHz [2a].

As filamentary cathode subminiature tubes usually have ratings for battery operation, so do these tubes, like 1.2V at the filament drawing between 12mA and 60mA, and  60V at the Anode with a current draw of a few mA or less.

Fig. 7 1ZH37B Pentode operation. Fig. 8 1ZH37B One control grid rod grounded. Pentode operation. Fig. 9 1ZH37B G2,G3 tied to Anode. Triode operation.

Figures 7-11 show curve families for the 1ZH37B [1]. The curves were photographed on a Tektronix 575 transistor curve tracer. This tube has a single filament, and two control grid rods that are brought out separately. The intended application for this tube is as mixer/oscillator. Each of two rods contributes about half the 1mS transconductance, as seen in Figure 8. Perhaps each of the grids could be dedicated to oscillator and RF input functions. The control grid characteristic is sharp cutoff, as might be expected with a constant spacing between filament and control grid. Perhaps a remote cutoff characteristic could have been obtained with grid rods that slanted away from the filament, with the wider portion of the gap having a much higher cutoff voltage.

Fig. 10 1ZH37B Effect of G3 on pentode operation. Fig. 11 1ZH37B G3 tied to plate. Tetrode operation.

The 15V Anode knee of this pentode is quite low as seen in Figure 7. Figure 9 shows a very conventional set of triode-connected curves. Figure 10 explores the possibility for control with the suppressor grid G3. As the curves show, control by G3 is most effective with 20V at the plate. The region of high control at G3, when the Anode voltage is 20V, can be used as mixer input for the local oscillator. The plot shows that 3Vp-p would produce effective mixing with 20V at the anode. But keep in mind that plate impedance is low in this region, which would limit gain and selectivity of a loading IF tank circuit. This region may also have been used for AGC.

Figure 11 shows tetrode operation with a very moderate tetrode kink. The low voltage negative resistance seen between 10V and 15V would not cause instability for external Anode load resistances less than 30kOhms with worst case 0V bias at the control grid. The very flat characteristic of the tetrode curves above 20V shows a higher impedance than in the pentode curves, in excess of 500kOhms at 1.5mA. This 500KOhm Anode resistance combines with the 1mS transconductance to yield an tetrode mu of 500; impressive for a battery tube running off 45V. This high Anode impedance is particularly suitable for very high gain IF stages with high impedance IF transformer loads running at 100kHz to 1MHz.

DC Voltage transfer function sweeps.

As I contemplated this tube structure, I wondered how independent, v ltage gain (mu) would be, from Voltage and Current. There are no grids as such, with unwanted grid wire pitch variation and it's effect on voltage gain over a wide range of plate voltages and currents. There is no "inselbildung" over the cathode as when a conventional ladder grid pinches off cathode regions directly under the grid wires. In the original HF application for these tubes, a very constant voltage gain would not be a primary concern, but the unusual rod structure sharpened my curiosity about mu, just the same.

Fig. 12 1ZH37B G3, G2 tied to plate. Triode Voltage transfer G1 to Anode.
Fixed Cathode Currents. (1uA, 2uA, 5uA, 10uA, 25uA, 50uA, 100uA, 200uA, 400uA, 800uA, 1600uA, 3200uA)

Fig. 13 1ZH37B. G3, G2 tied to plate. Triode mV transfer errors G1 to Anode.Fixed Cathode current. Fixed Cathode current.

Figures 12 and 13 show measurements of voltage gain in triode connection with a fixed cathode current. Fixed cathode currents ranging from 1uA to 3200uA in nearly exponential doubling steps were convenient to force in my measurement setup. The Anode was allowed to swing up to 80Vp-p, and was driven by the output of an opamp, while the cathode current was monitored and forced to be constant by the same opamp. This eliminated any loading errors in Anode voltage measurements because it was driven with the low impedance of the opamp output. As long as the Grid current is negligible, the Anode current equals the Cathode current.

Figure 12 shows curves with very straight, therefore linear, sections in the triode voltage transfer function. Figure 13 shows the difference between the input and the resistor attenuated version of the output. The resistor network attenuation is adjusted to reduce the output voltage swing until it matches the input as closely as possible. Any linear difference between the input and the attenuated version of the output, while ignoring fixed DC bias voltages, is a gain error.

The lower traces of Figure 12 are for voltage sweeps conducted at the lowest fixed currents, starting at 1uA. These sweeps show a marked gain deviation, or non-linearity, at low plate voltages, which could be explained by the grid current becoming significant. The Voltage sweeps conducted with 200uA of cathode current, no longer show any grid current errors because the plate voltage is sufficiently high to attract electrons the grid might conduct. Note that the 100uA sweep shows a marked bend, while the 200uA sweep is flat. While it is helpful that plate current is high compared to possible grid currents, the key mechanism is the starving effect that plate voltage has on grid current.

Fig. 14 1ZH37B. G3, G2 tied to plate. Triode %p-p errors for 5Vp-p input swings at G1. Fixed Cathode current.

Fig. 15 1ZH37B. G3, G2 tied to plate. Triode gain for various fixed Cathode currents. G1=5Vp-p

Figure 14 shows peak-to-peak linearity errors for all the Cathode current steps from 1uA to 3200uA, as a percentage of a 5Vp-p swing at the input. The 5Vp-p swing was biased to exclude the low  plate voltage region that causes grid current conduction. The voltage gain (mu) seen in Figure 15 varies from 11.5 at 1uA to 15 at 3200uA, thus yielding output swings from 57.5Vp-p with 1uA fixed current to 75Vp-p with 3200uA fixed current. The p-p linearity errors are less than an impressive 0.2% for currents between 10uA and 100uA. At the highest currents, one source of linearity error is additional self heating of the filamentary cathode.


Fig. 16 1ZH37B. G3, G2 tied to plate. Triode mV transfer errors G1 to Anode. Fixed Cathode current.


Figure 16 is a magnified version of Figure 13 showing mV errors of the most linear sections of the voltage transfer characteristic in greater detail. A 10mVp-p linearity from a 5V swing translates into a 0.2%p-p linearity error

Pure square law transconductance

For completeness sake in the assessment of DC operation, I also made precise measurements of the classic triode-connected plate I/V curve family over a sequence of grid voltages. Here, the interesting result was the nearly perfect square law, or parabola shape, of the I/V plate transconductance( also commonly known as mutual conductance ). One plot shows all the curves shifted to the left until they are super-imposed. The consistency of the parabolic shape is remarkable.

Fig. 17 1ZH37B. G3, G2 tied to plate as Triode. Anode Current vs Voltage. Fixed G1 Voltage steps.

Fig. 18 1ZH37B. G3, G2 tied to plate as Triode. Anode Current vs Voltage. Fixed G1 Voltage steps.
Superimposed parabolas.

The only deviation from the parabolic shape occurs at low plate voltages, when the most negative end of the 1.2V filament starts to turn off. Figure 19 shows the I/V curves from figure 18 in log scale, to highlight the deviation from a pure parabola, at low currents. This low plate voltage behavior is classic for a filamentary tube. Theoretically, the start of this deviation should start at an Anode voltage that is mu=12.5 times the 1.2V filament voltage or, in this case, below  12.5x1.2V=15V at the Anode.


Fig. 19 1ZH37B. G3, G2 tied to plate as Triode. Anode Current vs Voltage. Fixed G1 Voltage steps. Superimposed parabolas. Log scale.


Finally, if the tube is applied in a classic triode connection with a grounded cathode and a resistive load, the resulting transfer characteristic is a blend of the pure square-law parabolic characteristic for very low impedance loads to a pure linear voltage gain for very high impedance loads. The resulting distortion should be entirely dominated by second harmonic.

High Frequency Performance

Nearly all the data sheets for this line of Russian pentodes indicate High Frequency up to 60MHz as the intended application, but I have not been able to ascertain further HF performance specifics from the data sheet. I made some step response measurements that give an idea of transit time [4] and high frequency transconductance. Transit time establishes the fundamental limit of the tube for use as a mixer or RF amplifier. Small signal electrode capacitances are of lesser importance as they can be tuned out.


Fig. 20 1ZH17B.  60V at Anode and G2.
G3 and the shield are grounded.
The first trace is inverted and shows G1.
The second trace is the Anode.


Figure 20 shows the step response of the 1ZH17B pentode. This pentode has a single filament and is rated for 60V at the Anode. The figure shows an output step response of 100mV from a 2V input step. This computes to a 1mS transconductance. The input trace to the control grid G1 was inverted at the scope for easier measurement of the 1.85ns delay time from G1 to Anode. This delay is equal for both transitions. The rising and falling edges also have similar durations of approximately 1ns, and appear to be shorter than the delay and faster than the input grid transitions.

Other tubes I have measured, such as the 6AU6, 6CB6 and 6AG5 RF pentodes, have dissimilar transitions and turn off faster in 2.5ns than they turn on, in 3ns. This effect is even more pronounced in the slower 6EW6 video amplifier pentode with a 3ns turn-off plus rise time and 4ns turn-on delay plus fall time, from the 50% of the input grid voltage to 90% of the plate voltage transition.

My Tektronix PG502 pulse generator tops out at 250MHz. I was able to squeeze it to 270MHz, and there was no reduction in transconductance. This confirms the fast internal electron transit time. But the 50 Ohm load shorts out the small capacitance at the Anode. The 50 Ohm load serves as an effective current sense resistor, and it is not matched to the Anode impedance. The Anode needs a much higher load impedance, say 3kOhm, for a 10dB gain with the 1mS transconductance. In a real RF application the anode capacitance would be part of the loading tank. This 10dB gain seems quite feasable at 100MHz in view of the measured transition times and transconductance.


Fig. 21 6CY5.  80V at Plate and G2.
The first trace is inverted and shows G1.
The second trace is the Anode.


Figure 21 shows the step response of the 6CY5 VHF shadow-grid Tetrode for comparison to the 1ZH17B in Figure 20. The 6CY5 showed faster delays around 1.3ns, but slower rising and falling edges. The 6CY5 was one of the fastest VHF front end TV tuner tubes. It's 8mS of transconductance and shadowed screen grid, made it easy to match impedance to the antenna circuit with a minimum of noise degradation.


Fig. 22 - 5678 Shielded RF pentode by Raytheon.


Figure 22 shows the step response of the 5678 RF filamentary subminiature tube pentode by Raytheon. The 5678 draws 50mA from 1.2V at the fiament, and 1mA from the Anode at 60V. The DC characteristics and power drain of the 5678 are very similar to the 1ZH17B, yet the 5678 is so much slower, with 3ns delays.

Concluding the High Frequency evaluation, it is clear that the 1ZH17B pentode is faster than the 6AU6, 6CB6 and 6AG5 RF heater type pentodes; and much faster than the western counterpart, the 5678.

The 1ZH17Bis on par with the fastest VHF receiving tubes designed for commercial TV service, albeit at a lower transconductance of 1mS. But the truly remarkable feat is that the 1ZH17B is a low power filamentary tube designed for battery service, dissipating only 100mW on it's plate, and 120mW on it's filament, while the 6CY5 VHF tetrode dissipates 80V*11mA=0.88W on it's plate, and 1.2W in it's heater.

Final notes

It appears that these Russian subminiature tubes were used exclusively by the Soviet military for telecommunications gear from 1959 to the 1980's [3]. Most of these radios operated below 50MHz. Model r326, by Aleksandrov Radio Works in Russia, uses the 1ZH37B dual grid pentode in mixer service. One web site [2a] lists the operating frequency of the high frequency Russian subminiature tubes as 60Mhz. This is apparently extracted from the original Russian data sheet.

Fig. 23 Superhet Reflex by Tooru Kawabata uses  1ZH24B, 1ZH29B


Tooru Kawabata has a spectacular web site dedicated to his contemporary home brew designs of pocket tube radios. The following link points to his Superhet Reflex design using 1ZH24B 1ZH29BRussian tubes.


Translation from Cyrillic characters often results in different roman character versions. For example, the 1ZH24B may be listed as 1J24B [2].

References and acknowledgements

[1] The unusual 1ZH37B dual control grid pentode was a gift from TCA member Eric Tauechio. Tube Historian Ludwell Sibley provided encouragement to write this article.

[2] Sources of Russians tubes, data sheets and Cyrillic-roman tube type translation:   Russian tube store contains extensive listing of Russian subminiature tube types with summarized characteristics in English.
Russian site for  "Klausmobile Russian Tube Directory " with an English section, listing scans of original  Russian tube data sheets in the original Russian. Russian tube manual from 1962. - Jon M Townsend of Pierceton, IN 46562 runs an ebay shop specializing in subminiature tubes, including Russian types.

[3] Examples of Russian military radios using subminiature tubes: - example of Russian military  SW radio 1-20MHz with Russian subminiature tubes 1961-1978. is a personal page for Zenonas Langaitis. He collects Russian Military radios.

[4] Electron Transit time is the most fundamental characteristic that establishes the highest operating frequency of a tube. The delay time and transition time measurements I made are directly limited by the Electron Transit time for the tube, but are not the same as Electron Transit time.

To thank the Author because you find the post helpful or well done.

Tube Structure Photos and Dimensions 
20.Sep.09 06:41

Joe Sousa (USA)
Articles: 664
Count of Thanks: 90
Joe Sousa

Dear Radiophiles, thank you all for the very warm reception that this article has received.

I have assembled a few photos to show the remarkable internal structure of these tubes. Most remarkable of all, are the tiny dimensions. For example, in the case of the 1Zh37B the distance from the 0.07mm (70micron!) thick coated filament to the two grid rods on either side, is only 0.12mm (120micron!).

These tiny dimensions in the elegant rod structure constrain the very high performance sheet beam that gives these tubes their unique speed and linearity characteristics. I can't help but compare these dimensions to those inside modern integrated circuits. The point of the comparison is not so much that these tubes are very small, but that the small dimensions, as is the case with modern IC technology, are part of the reason for the high performance of the tubes.

These two figures show the rod structure rotated 90o from each other. I cut the filament rod on the first figure to show the Anode-grid sequence. I cut one of the two Anode rods on the second figure to show the filamentary coated cathode.

View of the top mica shows relative rod placement. Two sheet beams radiate from the filament between the G1 rods to the Anode rods.

The following three photos show three views of the top mica. The anode rods were removed, leaving behind the two holes on the mica.



1Zh37B Structure above the top mica, before I removed the getter cup and anode rods. This photo shows the anode rods already cut.


Two additional photos of the same 1Zh29B tube shown in the original article, but with dimensions added.

Note the two very thin filaments in the 1Zh29B  are visible on the photo  to the left. Surprisingly, this  0.3mm (300micron) filament is substantially thicker than the  0.07mm (70micron) filament in the 1Zh37B shown above.

 Emilio Ciardiello talks about a distance of only 15microns=0.015mm=0.0006inches from grid to cathode in the 416A Western Electric planar triode in his post about the construction of planar grids


1Zh29B view fom one anode side. The anode rod is prominent in the center.


1Zh29B view showing the various grid rods. The control grid is at the center and the anode rods are at top and bottom.

Best regards.



To thank the Author because you find the post helpful or well done.

1Zh37B in Gammatron / Gridless-Audion connection 
24.Sep.09 08:18

Joe Sousa (USA)
Articles: 664
Count of Thanks: 76
Joe Sousa

The 1Zh37B has separate connections for each of it's two control grid rods. Each of these rods sits a mere 0.12mm on opposite sides of the 0.07mm filamentary cathode.

This Rod-Filament-Rod sequence is akin to the Plate-Filament-Plate sequence found in Lee de Forest's gridless Audion that was included in the original triode US patent 879532 from 1908.

The most successful implementation of this approach was in the Gammatron line of Transmitter triodes from Heintz and Kaufman, starting in the early 1930's. These triodes have a low mu of 3, but very high current at low plate voltage. The reduced plate voltage makes up for the low mu to keep the grid swing comparable to the lower current conventional triodes. The plate current is so much higher that a much lower plate voltage can deliver the same power. Another important advantage of this electrode configuration is that the capacitance from the output plate to the input plate (i.e. control grid) can be much lower than for a comparable conventional triode, thus simplifying Miller capacitance neutralisation. US Patent 2071630 illustrates a dual triode variant for push-pull operation. US patent 2022212 illustrates a Gammatron triode that was optimized for very low input plate to output plate capacitance.

1Zh37B in Gammatron / Gridless-Audion Connection

With these precedents in mind, I decided to examine the performance of the 1Zh37B tube in Gammatron/Gridless-Audio configuration.

The fundamental operation of the 1Zh37B in this mode is characterized by driving one grid rod as a control grid, while collecting current from the other grid rod as a plate. Minor variants of this configuration are possible, depending on where the remaining rods are connected. Their connection affects operation slightly.

The attractiveness of Gammatron/Gridless-Audion operation has always been the reduced plate voltage requirement while running at high current. So is the case with the 1Zh37B, with a 10mA plate current drawn from just 9V. The transconductance is around 600uS, while the intrinsic voltage gain from one rod to the other is between 0.4 and 0.3, depending on the configuration of the remaining rods.

Connecting the extra rods to the control grid rod, gives the highest transconductance of 0.75uS with 9V and 9.6mA at the plate. The intrinsic voltage gain is mu=0.4 at this operating point.


Connecting the extra rods to the control plate rod, gives the lower transconductance of 0.6uS with 9V at the plate, but a higher 10.4mA current. The intrinsic voltage gain is mu=0.35 at this operating point.


Connecting the extra rods to the cathode-minus, gives the same transconductance of 0.6uS with 9V at the plate and the lowest 9.5mA current. The intrinsic voltage gain is mu=0.3 at this operating point.


Connecting the extra rods to the cathode-plus, gives the same transconductance of 0.6uS with 9V at the plate and 9.6mA current. The intrinsic voltage gain is mu= 0.35 at this operating point.


 This last photo is a composite of two sets of curve traces to show operation at the plate above and below ground, when the positive end of the filamentary cathode is grounded. Note that the vertical and horizontal scales are zoomed-in for greater detail and the zero plate voltage line is at the center. The most noteworthy feature of this plot is that there is a 450uA plate current with zero volts at the plate. The transconductance at 0V is down to a still useable 40uS and the intrinsic voltage gain mu is 0.25.

 This operating region has been used to design an AM transmitter that operates down to just 1.5V


Running the 1Zh37B in Gammatron/Gridless-Audion configuration in steady state much above 9V will tend to overheat the filamentary cathode, as shown by the looping delay caused by the filamentary self heating current.

The nominal filament current is only 60mA.

However this high current region may still be useful in class C operation, which has a low duty cycle with high peak currents and low average current.

Peak currents reach 16mA at 18V with 1.1mS transconductance and an intrinsic voltage gain mu=1.2. This increased mu, and the curve bending, suggest the onset of thermally limited emission


Occasionally, some thought is given by experimenters to running a low mu=2 triode, such as the 6AS7, in reverse operation, with the output taken from from the grid and the plate is the input controlling element. As expected, this gives a lower mu, which is approximately the inverse of the rated mu, but at much higher output current, albeit at reduced transconductance. This reverse triode mode of operation is similar to Gammatron/Gridless-Audion tube operation, for DC currents and voltages, but the inter-electrode capacitances are very different.



Wolfgang Holtman as posted a very interesting overview of the Dual Plate Triode, which is also known as the Gammatron, Gridless Audion or Gridless triode. (added Jan 18th 2010)



To thank the Author because you find the post helpful or well done.

1Zh37B - Sharp, or Extraordinarily Remote Cutoff 
03.Oct.09 06:47

Joe Sousa (USA)
Articles: 664
Count of Thanks: 74
Joe Sousa

The two rods that comprise the control grid of the 1Zh37B  shown schematicially to the right, give it special properties beyond other Russian Subminiature Tubes with filamentary cathodes and rod construction.

As illustrated in the earlier posts, the control grid rods, when driven together as one control grid, give a very pure square law control of Anode current by control grid voltage.

When one of the control grid rods is biased at a fixed DC voltage, the other rod can be used for transconductance control over a spectacular range exceeding 7 decades!

A DC sweep of this control rod from 0V to -20V, will change the Anode current from 1mA to 100pA in a very smooth logarithmic progression. The screen voltage for this sweep was 30V and the plate voltage was wired together with the suppressor grid in a tetrode configuration and biased at 10V.

The measured RED curve in the following plot illustrates the Extraordinarily Remote Cutoff that is possible with one control grid rod of the 1Zh37B at -20V. But it is just as extraordinary that this very same tube can present a very sharp cutoff to 1uA with just -3V driven simultaneously to both control grid rods.

The seven decade Remote Cufoff range represents the limit of my current measurements at 100pA, and not necessarily the lower limit of this tube.

The BLACK and MAGENTA curves represent calculated transconductance from the measured RED and BLUE curves.

Two additional thin straight blue lines show the very good log conformity below a few microamps. This might have been a handy feature in analog computer days or ,as MIT EE student Dimitri Turbiner suggested, to build a very wide range RF log detector.

The following two plots from the original 1Zh37B are unclear about the grid voltage in the horizontal axis, but it seems now that the first plot shows the remote cutoff transfer function with one congrol grid rod, and the other plot shows the sharp cutoff transfer function with both control grid rods driven.

The third plot shows my data in linear scale for comparison.

Another fundamental aspect of the remote cutoff behaviour is that the classic curve family retains it's shape over a wide range. I was able to trace the following two curve families:

The usuefulness of the remote cutoff characteristic is that the linearity remain good for the largest signals. The transconductance of these two curve families differs by 26dB, yet either curve family could amplify a 2Vp-p sinewave with low distortion.

These curves also illustrate a critical advantage over control with the suppressor grid that was shown on the first post, and that is that the plate impedance remains high under all control conditions, so that the bandwidth of any tuned load circuits is not impacted. This is one reason why the 1Zh37B  ddata sheet recommends multiplying mixer applications with local oscillator injection into one of the two control grid rods.

It should be apparent from these various curves that AGC control could be constructed by AC-Coupling the RF signal to both control grids rods, while maintaining one rod at a low fixed bias, and  applying the AGC control voltage to the other control grid rod. Perhaps a slightly more sophisticated AGC implrementation would apply a small level of AGC to the "fixed" rod, and the full AGC voltage to the other rod. Thiis would allow for running at highest possible transconductance under weak signal conditions, while shifting the bias slightly on the "fixed" rod to accomodate a larger RF signal, up to 2Vp-p.

The AGC design techniques for this logarithmic behaviour of transconductance control , could draw lessons from bipolar transistor design, which have a transconductance that also responds logarithmically to the control signal.



To thank the Author because you find the post helpful or well done.

1Zh37B-1J37b Data Sheet in English 
03.Oct.09 07:24

Joe Sousa (USA)
Articles: 664
Count of Thanks: 79
Joe Sousa

MIT Electrical Engineering Student Dimitri Turbiner has kindly translated the 1Zh37B data sheet.

The translation quality is excellent and we are all very grateful for Dimitri's contribution to the understanding of the Russian Subminiature Tubes with rod construction.

Thank you Dimitri.



To thank the Author because you find the post helpful or well done.

1Zh37B-1J37b Differential Mode Rejection 
03.Oct.09 21:08

Joe Sousa (USA)
Articles: 664
Count of Thanks: 98
Joe Sousa

Another parameter that is particular to the dual control grid rods of the 1Zh37B is the Differential Mode Rejection Ratio DMRR to differential signals presented to the control grid rods.

The following measurement shows anode current as a function of  applied differential voltage when the bias voltage is -702mV. I only show results for negative differential voltages, but positive differential voltages should yield a very similar result, reflected about the vertical axis at 0V.

This plot also includes driving one of the grids positive, above the -0.7V bias, to the left of the vertical -1.4V dashed line.

Note that the vertical scale is zoomed into the 500uA to 640uA range where significant Differential Mode respose occurs.

The Differential Mode Transconductance - DMT - for absolute voltages greated than zero, or to the right of the -1.4V line, can be calculated for p-p values as 38uA/1.4V=27uS.

The Differential Mode Rejection Ratio - DMRR - is the ratio between DMT and the transconductance with both rods driven, which is the usual transconductane and could be called in this context, a Common Mode Transconductance CMT.  DMRR=DMT/CMT=388uS/27uS=14.3, or 23dB. This is just one p-p result for a 1.4V differential swing on a 0.7V bias. Different vlues will be obtained for different amplitudes and bias values.

Ideally, CMRT should be zero for a perfectly symmetric linear system. In this case, the good symmetry is shown by the flatness of the plate current near zero volts, so the  CMRT arises from the non-linearity of the grid rod transfer function.

An additional application for the the dual control rods of the 1Zh37B could be in frequency doubling. A large differential swing at the control grid rods results in a full-wave rectified Anode current at twice the frequency.

This opens the door to the design of aditive mixers that allow the Local oscillator to run at half the frequency, while the RF signal is applied to both grids. One such topology may look like this:

One advantage that could be reaped from running the local oscillator at half frequency, is the reduction of oscillator drift sensitivity to stray capacitance, up to a factor of 4, if the same inductance is used when compared to an oscillator running at full frequency.

Perhaps a battery version of the very elegant reflex FM stereo multiplex stereo decoder by Grundig could be adapted to take advantage of the dual input grids to double the pilot frequency from 19kHz to 38kHz. At the very least, the dual grids could double the 19kHz to 38kHz in a more conventional multitube stereo decoder.



To thank the Author because you find the post helpful or well done.

1Zh37B Capacitance Measurements 
10.Oct.09 08:40

Joe Sousa (USA)
Articles: 664
Count of Thanks: 71
Joe Sousa

Capacitance measurement method

A conventional capacitance meter is designed to measure the capacitance between two terminals.

But the most important capacitance in in a tube, which is usually the capacitance from the Anode to the Control grid, and it depends on three terminals:



3-all other terminals shorted to ground as one terminal.

A different method of capacitance measurement can be used in the home lab, that uses a square wave generator and an osciloscope to determine the capacitance between two terminals when a grounded third terminal is involved.

The basic principle of measurement is to connect the unknown capacitance between a large square wave voltage and a known load capacitance that is monitored by an oscilloscope.

An improvised home-lab implementation

In the implementation shown here in schematic form, I used a Wavetek 142 function generator to drive 10Vp-p square wave at 1MHz into the first terminal of the unknown tube capacitance, which the example shows as the Anode

Then all other terminals that would have been at AC ground potential are grounded.

The second terminal of the unknown tube capacitance, which is the two control grid rods, drives a known 100pF load capacitance. This load capacitance is the sum of 20pF scope input capacitance, 65pF 1X probe capacitance, and an additional external 15pF trim capacitance to bring the total to a convenient 100pF value for calculations.

The 100k load resistance prevents Hum pickup by forming a 16kHz low cut filter with the 100pF load capacitor.

The Scope monitors output voltage such that 100mVp-p represents 1pF of capacitance from input node to output node.

Tube shield and Measurement shield for 3 terminal capacitances

The actual value of capacitance that was measured in the example was 0.008pF=8fF from Anode to Control Grid rods G1A+G1B, with all other electrodes grounded and with an external grounded shield covering the length of the tube. This grounded foil shield hugging the glass envelope is required for minimum capacitance. Western subminiature tubes used conductive paint over the glass envelope, that also helped trap sray capacitive coupling from Anode to Control grid of Pentodes.

This particular measurement was the most difficult. It only produced 0.8mV p-p of signal on a 2mV/div graticule, and made it necessary to place a grounded copper sheet separating the input Anode side of the measurement from the Grid output side.

The following photos show the improvised mechanincal construction for the measurements. Note that the red wire seen in the third photo has 10Vp-p from the generator, and must, therefor be completely on the anode side of the shield.

The fourth photo shows the anode terminal coming out the top as a single wire, the two control grid rods are grabbed by the 1X scope probe, and the remaining terminals and shield are bundled and held by the grounded aligator clip.

Click to enlarge photos or schematic

The copper shield board is a recycled pcboard with a convenient set of holes.

Note the grounded copper foil shield over the tube. This shield is necessary to trap stray fields from the anode that would reach around the other electrode rods and leak additional capacitance to the control grid rods.

Capacitance results

The Capacitance from Anode to G1A+G1B control grid rods with the grounded copper foil shield over the tube is 0.008pf, if the foil copper shield over the tube is removed, the capacitance increases to 0.05pF. This much additional capacitance is enough to cause instability in an IF/RF amplifier.

The individual 2-terminal capacitances from each electrode to all others was in the range from 3pF to 6pF.

The ouput capacitance from the various possible output nodes A, G3, G2  to the input control grid rods G1A+G1B with a shielded tube were as follows:

A to G1A+G1B= 0.008pF in Pentode connection with all other electrodes AC grounded. Removing the external shield increases this capacitance to 0.05pF. This increase was enough to cause oscilations in my current IF 455kHz amplifier design.

A+G3 to G1A+G1B=0.17pF in Tetrode Connection with all other electrodes AC grounded. Removing the external shield increases this capacitance to 0.24pFpF.

G2 to G1A+G1B=2.7pF which is nearly the same as triode connection, but with all other terminals AC ground.

A+G3+G2 to G1A+G1B=2.8pF in Triode Connection. Removing the external shield increases this capacitance to 3.0pF.

These measurement results supplement values in the translated 1Zh37B data sheet and will be useful in my RF/IF circuit desings with this tube.



October 16th 2009. Added capacitance measurements in triode configuration, and additional measurements with the external shield removed.

To thank the Author because you find the post helpful or well done.

Background from Barbour and Heider 
13.Oct.09 03:49

Joe Sousa (USA)
Articles: 664
Count of Thanks: 70
Joe Sousa

In a recent email exchange with Tube engineer/businesman/editor Eric Barbour, he had this to say when he worked as applications engineer at Svetlana.
The following is a quote from Eric's email:
At Svetlana, I had a chance to see some of those subminis, because (so I was told) they were designed by Svetlana's engineering staff, mostly in the 1950s. Very ingenious designs. I begged them to make more, and they said "all the tooling was sent to Novosibirsk factory, where most of them were made. Nobody wants that stuff anymore. It was used in military equipment only, like avionics."
The 1ZH37B, which I also found fascinating, was "very difficult to make". How much of the old leftover stocks remain? "I don't know, it was all thrown away".


I notice that the Novosibirsk factory is still in business, and even has a website. Making ceramic transmitting tubes, and "a range of coaxial-waveguide modules of microwave frequency on the basis of active vacuum elements".


Hans Jürgen Heider has a nice list of Soviet trademarks.

Thank you for the contributions.


To thank the Author because you find the post helpful or well done.

1p24b data sheet in English  
08.Nov.09 21:33

Joe Sousa (USA)
Articles: 664
Count of Thanks: 79
Joe Sousa

Gammatron operation with reduced filament voltage 
09.Feb.10 08:03
2885 from 164820

Joe Sousa (USA)
Articles: 664
Count of Thanks: 75
Joe Sousa

Fellow Radiophiles,

Wolfgang Holtman has demonstrated a higher voltage gain of 2 in Gammatron operation, if the filament voltage is reduced. He demonstrated this elevated gain in his very interesting Review of Dual Plate Triodes in English, and in the original German.

Wolfgang attributes the increase in voltage gain to the smaller electron cloud that forms at the filamentary cathode with reduced temperature from lower filament voltages. His insight about electron cloud size was not only helpful to understand the operation of this special tube, but also the nature of the space charge around a cathode.

I added a few reflections to Wolfgang's post about my understanding of his results.


The following curve traces show Gammatron operation with g1a as the control rod and all other rods tied together as the anode (plate).

Each plot shows a decrease of 100mV in filament voltage, except where noted.


These first six curve families show very little change as filament voltage is decreased from 1.5V to 1.0V. The last of these six plots starts to show a widening of spacing of the first curves starting from the left-most curve, and a correspondign increase in mu from 0.4 to 0.7.

The widening of the curve spacing becomes more pronounced for the first negative grid steps, and below Vfilament=0.8V the anode current starts to drop substantially. The reduced current comes with a reduced transconductance, while the intrinsic voltage gain mu keeps increasing, reaching 1.05V/V when Vfilament=0.7V.

The next three plots were all taken at the same 0.6V filament voltage.

Below Vfilament=0.7V, the current drops rapidly.  The first plot shows a 50% current drop, as Vfilament was droped from 700mV in the previous plot to 600mV.

Rescaling the Vertical axis in the middle plot shows a disproportionate widening of the first few grid steps, and a corresponding increase in intrinsic voltage gain mu to 1.8V/V at the same +10V anode voltage used until now.

The third plot extends the horizontal axis to cover 50V. Now the thermally limited current flow is apparent in the more horizontal slope of the upper curves at higher anode voltages.

The highest voltage gain mu=3V/V, with Vfilament=0.6V, occurs just before thermally limited current flow dominates the behaviour at higher anode voltages.

The following three plots show space charge limited current flow to the left of the curve knee, and thermally limited current flow to the right of the curve knee. The reduction in slope at higher current levels, marks the thermally limited current flow.

The anode current keeps dropping substantially with each reduction in filament voltage. Note the shrinking vertical scale.

The highest voltage gain mu=4.5V/V was realized with Vfilament=0.4V, but at the price of greatly reduced gm=9uS with the Anode biased at 20V and drawing 40uA.

The relatively even spacing along a horizontal line at the center of the curve families suggest that a wide voltage swing of 100V is possible under current source load conditions.

The last plot shows that the filament has no useful emission left with Vfilament=0.3V.

Thank you Wolfgang, for taking to initiative to experiment and show us a new class behaviour in these tubes.


To thank the Author because you find the post helpful or well done.

1zh24b vs 1zh37b filament efficiency. 
27.Nov.10 03:42
6828 from 164820

Joe Sousa (USA)
Articles: 664
Count of Thanks: 86
Joe Sousa

Fellow Radiophiles,

Our recently joined member Dmitri Faguet has brought to my attention the amazing power efficiency of the 1zh24b rod pentode. The 1zh24b filament draws 15mW, while the 1zh37b filament draws 70mW. The following table compares the essential low frequency characteristics of these two tubes.

1zh24b vs 1zh37b
  Filament draw (1.2V) Anode current G2=45V G1=0V gm1
1zh24b 12mA/15mW 0.8mA 600uS
1zh37b 60mA/70mW 2mA 1000uS

I made a few measurements to get a sense of how the filaments work in these tubes.

Summary of Filament measurements
  Volts mA Resistance (Ω) Rhot/Rcold Temperature (oK) Power(mW)
1zh24b 1.204 12.6 95.3 3.51 1053 15.2
1zh37b 1.202 60.4 19.9 3.85 1157 72.5

The absolute operating temperature of the filaments can be estimated from the resistance increase. In other words, the tungsten filament serves as an approximate temperature sensor. I used the simple rule that the tungsten resistance is directly proportional to absolute temperature. I neglected the slight curvature that links resistance to temperature, but that is negligible to compare these two tubes. The room temperature (27oC=300K) measurements were done with less than 40mV across the filaments to avoid any spurious increase in temperature.

 The following plots illustrate the filament operation as voltage is increased from 0V to operating levels up to 1.5V.

The length and supporting structure for the two filaments is virtually identical. A few simple conclusions can be drawn from these plots:

-The 1zh37b runs approximately 100oK (oK=oC+273oC) hotter than the 1zh24b. This is the opposite of what might be expected for higher emission efficiency.

-While it only takes 15.2mW to bring the 1zh24b to 1053o, it takes about 55mW to bring its filament to the same temperature.

-In terms of emission efficiency, the 1zh24b needs 18.8mW/mA, while the 1zh37b needs 35mW/mA. The 1zh24b appears to be twice as efficient in converting filament heating power to thermionic electron emission.

-The efficiency is even higher in terms of expended heat to achieve a certain transconductance. 25mW/mS vs 70mW/mS.

Perhaps the only construction difference between the two filaments is wire thickness. If this so, it is remarkable how this simple difference has such a strong effect on filament efficiency. Perhaps the close proximity of the control grid G1 rods is a strong factor in filament emission efficiency. The disproportionate increase in transconductance efficiency suggests this.

A further comparison could be made with the DF96 and DF92, which were also measured a while back for  filament efficiency. Note how the DF96 is more efficient than the DF92, yet runs it's filament at a higher temperature. This is the opposite circumstance in the comparison between the 1zh24b and 1zh37b.



To thank the Author because you find the post helpful or well done.

1zh24b IF gain with reduced filament voltage. 
27.Dec.10 02:35
8262 from 164820

Joe Sousa (USA)
Articles: 664
Count of Thanks: 63
Joe Sousa

Fellow Radiophiles,

Dmitri Faguet and I have been collaborating on experiments with the 1zh24b to see how far the power efficiency of this tube can be taken. The following  report was written by Dmitri Faguet:

In the following experimental study the 1j24b Rod Tube was tested in order to determine its performance in a resonant amplifier circuit with a high Q tank, when tube filament is operating below its specified “standard” temperature levels.
The 1j24b is a universal ultra low power miniature HF pentode designed in the mid 1950’s as part of the series of so-called Rod Tubes, invented by Russian engineer and academic Valentin Avdeev, who worked at a special vacuum tube research and manufacturing plant code named “No.617”, located in Novosibirsk, Russia during and after WWII. For many years, Rod Tubes became the backbone of Soviet military and aerospace electronics, with more than 200 million Rod Tubes manufactured without any modifications for nearly four decades from 1950’s to 1990’s. All Rod Tubes were low power battery operated pentodes with one or two thin 1.2 V filaments, and with other rod-type electrodes rigidly located at very small distances from each other.
Typical application of 1j24b pentode was for amplifiers, oscillators and mixers of communication receivers and battery operated field radios, such as a small size R-126 and R-131 tactical field radios, a backpack radio R-105M and its modifications, a special purpose 19-tube communications receiver R-326, a long distance 24-tube GRU control and communications receiver R-309, a 60-tube (!) commander tank radio R-130, etc.

Typical specifications of 1j24b pentode are summarized in the following Table 1.


Table 1 Rod Tube 1j24b technical data




Uf [V]


Directly heated filament cathode; voltage range 0.95 to 1.4 V

If [mA]


+/- 2 mA, depending on actual Uf value

Ua [V]


Typical value. Normal anode operating voltages from 30 to 120 V

Ia [mA]


At Uf = 1.2 V, Ua = 60 V, Ug2 = 45 V, Ug1 = 0 V. Max Pa = 120 mW

S [mA/V]


Same conditions

Rin [k]

> 100

Control grid impedance, measured at F = 60 MHz

Ca [pF]


Anode (output) capacitance

Cag1 [pF]

< 0.008

Control grid to anode capacitance with external grounded shield.

Rg1 [k]


Max value of external resistor at control grid

T [hrs]


Guaranteed service life at normal operating conditions, hours


D = 8.5 mm, H = 42 mm, weight = 4 g. Linear acceleration: 100 g

To carry out the 1j24b tests a special high Q tank was designed and made, and a one stage resonant amplifier test circuit was constructed, as shown here:

Photo 1 Resonant amplifier stage connected to signal generator and oscilloscope

 Photo 1 Resonant amplifier stage connected to signal generator and oscilloscope

The resonant LC circuit used in these experiments has the following parameters:


Table 2 LC circuit data




L [uH]



Ferrite core, D=18 mm, H=10 mm; 70 turns of “Micron” cable UAA3401 with 3 strands

C [pF]


Sum of 110 pF (parallel C), 12 pF (x10 Tek probe), 3 pF (Ca, 1j24b), 7 pF (coil C)

Fres [kHz]


Formula: Fres = 159 / sqrt (L * C)



Unloaded Q factor, measured at resonant frequency

R(LC) [M]


Tank parallel impedance: R = Q * sqrt (L / C)



When connected to x10 Tektronix THS710A scope probe, 10 M



Equivalent impedance, with the above probe

The choice of this particular load for resonant amplifier test was justified by two opposite factors that a designer should take into account in this particular case: a) maximum gain factor; b) stability without special neutralization measures. To measure actual tube plate impedance in dynamic mode (which is one of our goals here) and to maximize gain, one should use LC circuits with as high value of parallel impedance [at resonant frequency] as possible. However, at high R(LC) the condition of theoretical stability of a resonant amplifier can cause problems, which may require the use of tapped coils. Indeed, the vacuum tube resonant amplifier stability condition is expressed by the following well known approximate formula [This formula was published in a Russian 1957 "Ham Reference Book"]:

RLC - (kΩ) resonance

Fres - (MHz) resonant frequency

Cag1 - (pF) Anode to g1 capacitance

S - (mA/V) transconductance

When P > 1, the stability criteria is met, and there is no need to use tapped coils in a resonant amplifier or capacitance based neutralization . It is easy to see that in our case this condition is violated, and test circuit used here can become unstable. Of course, when we connect a 50 ohm output of a signal generator to the control grid of the tube, the amplifier becomes stable, and we can ignore the aforementioned stability condition. At the same time, for practical purposes we should assume that the tank will be connected to the control grid of the next stage, with typical value of a resistor in AGC loop of about 1 M. As we shall see from the following tables, the above stability criteria can be met in such case, at least for low end Uf values for which tube S factor becomes smaller.
The test circuit diagram used for our measurements looks as follows: 

Exhibit 1. Test circuit schematic diagram

Our goal was to measure gain factor and plate impedance when 1j24b filament voltage was allowed to vary from 500 mV to 1.2 V in 100 mV steps. An interesting feature of 1j24b Rod Tube is its ability to work even at slightly positive control grid voltage, with no deterioration of its input impedance characteristics, and at a relatively high input signal levels. This feature, mentioned in Rod Tube reference books, was routinely used by Soviet engineers when they designed narrow FM field radios that did not require AGC loops and negative voltage injection into control grid of Rod Tubes, or front end stages of AM receivers. For example, in the following Photo 2 one can see part of R-326 actual circuit diagram, showing the second stage of the RF amplifier in the receiver, in which 1j24b tube is used, and its first mixer stage, with a special type 1j37b pentode with two separate control grid rods. As one can see, both tubes operate here at zero control grid voltage.

Photo 2. Part of front end R-326 communications receiver circuit diagram

The results of these experiments are summarized in two Tables. Here:
Uf – 1j24b filament voltage, measured with maximum 0.5% error;
If – filament current;
Ia – anode current measured in static mode, with no RF signal at control grid;
Pf – total power dissipated by tube filament;
Pa – total power dissipated by anode;
Gain – actual gain measured in “small signal” mode, when output RF voltage was close to 1 V, and when the LC circuit was shunted by an additional 1 M resistor to represent an equivalent load by control grid resistor of the following stage. The resulting anode load is then equal to 500 k, which is sufficient in terms of stability of the amplifier, and at the same time still guarantees a high gain factor;

Qx - an equivalent Q of LC circuit, where Fres = 163 kHz is the resonant frequency in test circuit, dF = 5 kHz is a frequency shift at which output RF voltage U is measured, and Ures is the output RF voltage at resonance



Rx – an equivalent parallel impedance of LC circuit in the “loaded” mode



Ra – an equivalent dynamic plate impedance of 1j24b tube, where RXLC=1M is taken from Table 2


Sx – an equivalent value of 1j24b S factor

Table 3. Ua = Ug2 = + 45 V

Uf [mV]

If [mA]

Ia [mA]

Pf [mW]

Pa [mW]

 Gain 500kΩ


Rx [k]

Ra [k]




















11 059










5 374










3 969










3 332


1 000








2 840


1 100








2 448


1 200








2 129



Table 4. Ua = Ug2 = + 30 V

Uf [mV]

If [mA]

Ia [mA]

Pf [mW]

Pa [mW]

Gain 500kΩ


Rx [k]

Ra [k]






























11 059










7 888










6 855


1 000








6 038


1 100








4 825


1 200








4 363


As one can see from these numbers, the 1j24b Rod Tube can be successfully used at only half of its 1.2 V filament voltage, and at relatively low anode voltage of about 20 to 30 V. At Uf = 0.7 V the power required for filament heating is equal to just 7 mW, while the anode power dissipation at Ua = + 30 V in this case is less than 2 mW. Gain factor of a two-stage resonant amplifier can therefore be expected to be in the range from at least 60 to 80 db, with total power required for such amplifier of less than 20 mW. In other words, when Rod Tubes were invented by Russians in mid 1950’s, they were as economical as germanium transistors that were available in those days, and at the same time Rod Tubes had much better electrical characteristics, in terms of their excellent high frequency performance and high input impedance, than Ge transistors. These factors determined an astonishingly long and triumphant life of Rod Tubes in Russian military electronics.

Dmitri Faguet

Thank you Dmitri for demonstrating the extraordinary efficiency of the 1zh24b.


To thank the Author because you find the post helpful or well done.

1zh24b DC operation at low voltage 
27.Dec.10 06:04
8284 from 164820

Joe Sousa (USA)
Articles: 664
Count of Thanks: 60
Joe Sousa

Fellow Radiophiles,

The following DC curve traces were prompted by Dmitri's work in the previous post, and put the bias levels of the IF stage in the previous thread in the wider context of the operating range of the 1zh24b.

These first two plots compare operation at full 1.25V filament voltage with the screen grid G2 biased at 43V and 30V. Note that the curves also include positive grid steps because grid current remains below 0.1uA until the the grid is +1V with respect to the negative end of the filamentary cathode.

This lack of control grid G1 conduction more than doubles the linear input voltage range, and also makes it possible to trade screen voltage for control grid voltage to some extent. If the screen grid G2 voltage is lowered from +43V to +30V,  the 450uA plate current that flows when G2=43V can be regained by increasing G1 bias to +700mV.

The next two plots illustrate the extraordinary finding by Dmitri in the previous thread that substantial gain and transconductance are still realizable when the very lean 1.2Vx13mA=15.6mW filament is reduced an even leaner 0.6Vx8mA=4.8mW . Note how helpful the lack of G1 conduction below +1V is in expanding the operation at low filament voltage. Most of the useful operating range of the control grid G1 with Vf=600mV is centered around +400mV of grid bias.

The following curves curves show the extraordinary lack of control grid G1 conduction below +1V.

Note that the (0,0) origin is on the left, and +1V is at the second division from the left. Each vertical divsion is 10uA, and the little tics are 2ua.

The first photo was taken with rated 1.25V filament voltage, and the second was taken with 600mV at the filament. Note how G1 does not reach 1uA with Vf= 600mV until +1.7V.

Control grid conduction on the second photo with Vf=600mV shows what appears to be thermally saturated emission, as all emission is split between the forward biased control grid G1 and the relatively high G2 voltage. This current sharing is controlled by μg1g2=20. The previous two plate sweep plots with Vf=600mV and Vf=552mV avoid thermally limited emission in part because there is no current spent at the control grid G1.


This plot is included here for comparison with the control grid G1 conduction characteristics of another Russian Subminiature Rod tube, the 1zh37b.

Note that the Voltage origin on this plot has moved to the center line to show conduction for negative voltages.

There is a heavy G1=35uA short circuit current with zero bias at G2, and still G1=5uA short circuit current, if G2 is brought up to 30V.

The position of the second curve in the plot suggests that bringing G2 to 60V would move the conduction knee to the right by another 500mV, thus eliminating most G1 short circuit current.


In Post#12 Dmitri made his measurements with the manufacturer recommended control grid bias of 0V. The DC curve traces suggest even greater efficiency if the grid bias is set at +400mV. For example, from Table4, the realized voltage gain with Vf=600mV is 37 with Vg1=0V. The third photo of this post has a sweep for G1=+400mV that suggests  an increase in transconductance from 77uS in Table4 to 200uS, which would triple the realized voltage gain of 37 to 111.

The earlier filamentary efficiency comparisons in post#11 and the lack of G1 conduction below +1V suggest a special efficiency in the design of the 1zh24b cathode/grid space. Perhaps this efficiency comes from the very fine filament diameter, and the resulting small space charge diameter, which would not reach the control grid, and would be more easily controlled by the electric field of the control grid G1.

Best regards,


To thank the Author because you find the post helpful or well done.

1zh24b IF gain with positive G1 bias 
10.Jan.11 05:54
8790 from 164820

Joe Sousa (USA)
Articles: 664
Count of Thanks: 58
Joe Sousa

Fellow Radiophiles,

Dmitri Faguet has updated his low power IF measurements to include operation with positive G1 bias. This takes advantage of the lack of grid current conduction until Vg1>+1V. The added positive bias increases transconductance.

Dmitri writes:

The 1j24b tube was "lean-powered" at Uf = 700 mV, and Ua = Ug2 = +30 V, Ug3 = 0 (in this case Ia = 0.05 mA at Ug1 = 0).

I used the same amplifier configuration as in the previous test, with 1 M shunting resistor for LC circuit (the resulting anode load is equal to 500 k), and with Ug1 now changing from +0.5 V down to -1.6 V in 100 mV steps.

The results are summarized in attached table and graphs. As you can see, the IF amplifier stage actually delivers at least 40 db gain, with 60 db AGC capability. This confirms your excellent suggestion that 1j24b tube can successfully operate at positive control grid voltage levels, without causing control grid (rods) leak (I used 1 M resistor at control grid).  It also confirms our previous assumption that a simple two-stage IF amplifier with 1j24b Rod Tubes operating in "micro power" mode, with total power consumption of less than 10 mW per tube, can deliver up to 80 db gain, with excellent AGC capability of about 60 db per stage when the control grid bias is allowed to change from plus 400 mV (critical positive bias level at which gain differential starts to decline) to minus 1.6 V (at which point 1j24b anode current goes down to negligible levels, something like 100 nA).

Also note an excellent linearity in the range of control grid bias varying from +0.3 V to -0.3 V. 

Uf [V] = 0.70
Ua [V] = 30.00
Ug2 [V] = 30.00
Ug3 [V] = 0.00


Ug1[V] Gain Gain, db

(500 k) (500 k)
-1.60 0.10 -20.00
-1.50 0.19 -14.42
-1.40 0.38 -8.40
-1.30 0.70 -3.10
-1.20 1.13 1.06
-1.10 1.90 5.58
-1.00 2.90 9.25
-0.90 5.00 13.98
-0.80 7.50 17.50
-0.70 11.50 21.21
-0.60 17.00 24.61
-0.50 23.00 27.23
-0.40 30.50 29.69
-0.30 40.00 32.04
-0.20 51.00 34.15
-0.10 62.00 35.85
- 72.00 37.15
0.10 83.00 38.38
0.20 93.00 39.37
0.30 104.00 40.34
0.40 112.00 40.98
0.50 117.00 41.36

Thank you Dmitri for continuing to expand the possibilities for the remarkable 1zh24b. 


The cutoff of the 1zh24b characteristics are roughly comparable to the DF96, with a g1 control range spanning 2V.

However care must be taken to avoid cutoff with large input signals, exceeding 100mVp-p at the control grid of the 1j24b or DF96.

Having two stages with  slightly staggered AGC bias, such that the second stage gets about 300mV more positive bias than the first stage should keep the IF signal away from cutoff.


Best regards,

-Joe Sousa


To thank the Author because you find the post helpful or well done.

1j24b more on Structure 
01.Oct.11 18:39
17398 from 164820

Michael Watterson (IRL)
Articles: 1006
Count of Thanks: 51

There are lots of good photos at the start of this article and also various diagrams of the rod layout. But perhaps not all the valves (tubes) are quite the same also to me the photographs don't lie but the correspondance with the various diagrams sometimes seem not 100%.

I accidently burnt a filament. It lit quite bright for a second with the 32V HT on it!

So I scored the base and wrapped the tube and tightened a bit in the vice slowly. The glass seems much thicker than the nomal Noval valves such as ECC82


Inside a 1j24b

I had thought initially a filament ran on outer edge, then maybe a filament support wire. Joe Sousa pointed out that this was most unlikely. So I thought again and took a photo of an undamaged unused but tested "OK" filament 1j24b (dated 1991!)

Closeup of the base of 1j24b

This closeup shot (quite hard to light correctly) suggests that there is no wire through the three eyelets on the mica spacers. The purpose though of these three eyelet isn't clear to me. The opposite diagonal perhaps has plain holes in the Mica.

The logical solution is that either the two flat plates at sides at right angles to the two outer Anode plates are feeding the top end of the filament, or there are two filaments supported by spring attached to the flat plates.

Unlike many valves the other screen isn't a cylindar around the entire Anode but just rods at the four corners as two upside down "U" joined by the getter dish. The getter dish acts as additional shielding between the "U" connecting the Anode Plates and the top wire.

Thus all the rod pentodes are like two pentodes in a common tube with a common cathode and the all the electrodes electrically connected. I think this helps to understand why the 1j30b and 1j37b behave like two paralleled pentodes with separated g1 connections and not at all like a hexode.


To thank the Author because you find the post helpful or well done.

Detail on 1j24b, 1p24b and 1j37b rod layout 
04.Oct.11 16:31
17951 from 164820

Michael Watterson (IRL)
Articles: 1006
Count of Thanks: 54

I had envisaged the g1 rods being between the filament and anode path on the Rod pentodes and that actually is clearly not the case.

The old Radio article in 1964 (Russian) and the close up photos reveal that there is unlike "normal" pentodes nothing at all between the heater/filamentary Cathode and the Anode plates.


original 1964 diagram rod pentode

Cross-section of rod  pentode
Redrawn from the 1964 Radio Article.

Misleading aspects of the diagram are:

  • The filamentary cathode is barely visible compared to rods.
  • The two horizontal flat "grid" rods are very much closer to the filament and a little less wide than the Anode plates (thickness looks similar on 1j24b from above).
  • The G3 rods may be much closer to anodes

Otherwise the relative placement and size is correct. There are two beams with no obstructions, one each from filament to two Anodes.

The two extremes of Rod Pentodes are:

1j24b which may have about 25mW Anode power and 11mA to 13mA Filament current (About 12.5mA 1.3V) and the 1p24b which is about 2.5W Anode power and 90mA to 110mA each on two filaments (total about 220mA @ 1.3V)

The 1j24b diagram drawn on top of actual photo in an earlier post

Part of glass and one of the three mica wafer outline is shown. For completeness the Shield /Screen rods are shown and part of the feed (with spring) to top end of filament. The size of the filament is greatly increased just for the diagram, otherwise it's approximately to scale. My confusion over the 3 holes (only one of which has an eyelet) on the 1j24b tube is because a stub of the top of the f+ wire is in the eyelet at the bottom of the three mica wafers (the extreme bottom left hole in diagram). This wire has feed without spring to the bottom of the filament. The Mica Wafers are symmetrical for ease of assembly thus have unused holes as shown above.

To compare here is the other extreme of the range of rod pentodes, the 1p24b-v which has 20x filament current in two filaments (so x10 per filament) and about x100 or more output power by x2 HT and thermal management in design.

1p24b-v construction

1p24b-v construction: Cross section and top filament feed arrangement.
(the wires are pretty straight from springs to filament, they are drawn with kink for clarity to see g1)

Unlike the other rod Pentodes the tube has four sunk ribs in the glass to locate the square mica wafers.

Again you can see there is no obstruction between filament and Anodes. There are two rods marked "sh!" The 1j29b also has a pair of such rods which are connected via straps to the adjacent actual Shield/Screening rods. These rods have a gap at the bottom of tube (not full length) and thus likely serve to ensure the G3 to ground capacitance is the same for "both" pentodes to match the capacitance of the filament feed rods or that the electrostatic focus is the same strength on both sides of the beam.

While the 1j37b has two grids, it's actually the simplest type in the family though only the 1p24b has more than its very high (for rod Pentode) 60mA current in a single filament. The 1j29b has about 56mA  at 1.35V, but that is in two filaments.

1j37b construction

The diagram of 1j37b is an  overaly from photograph on this thread

The spacing of g3 is a bit more than the other rod Pentodes I think and there are no shield/screen rods other than the one marked "f!" so that the right hand "pentode" g2 and g3 have same capacitance to filament/Earth.

To thank the Author because you find the post helpful or well done.

Basic studies to rod control 
07.Oct.11 11:33
18249 from 164820

Marc Gianella (CH)
Articles: 314
Count of Thanks: 55

The booklet "Radio-Röhren" (Radio-Tubes) by Herbert G. Mende Franzis-Verlag, Munich 1966 notes the fact of rod control as follows:

The two bars holding the control grid seal a partial area of the anode off the electron system through their negative potential, as resulting in the studies of tubes with willemiteanode (luminous coating anode). (Strutt, Moderne Mehrgitter-Elektronenröhren, Band 1, Berlin 1937)


The main effect of a control-grid built for instance of two rods was found by compressing the electron-bunch attaining from cathode to anode. It is possible to mount an auxiliary anode in front of the anode, catching a part of the electron current at few negative control rods respectively not beeing hit by electrons at strong negative control rods. Arbitrary anode-current characteristic can be achieved by means of suitable design of the auxiliary anode. For bigger transconductance, four or more control rods instead of two can be arranged around the cathode. Exponential characteristic for remote cut-off can be achieved, too.
(Jonker, Philips Research Reports 4/1949, 357)

Unfortunately I do not own these books. I hope I translated clearly.

To thank the Author because you find the post helpful or well done.

Comparison of Filaments and Negative feedback / DC bais 
07.Oct.11 21:18
18331 from 164820

Michael Watterson (IRL)
Articles: 1006
Count of Thanks: 48

Here the filaments of the commonly available Rod Pentodes are compared. The 1jHH17 is probably just early Roman marking of 1Ж17Б (1j17b) as the shift from Roman to Cyrillic marking in Russia was in the 1950s.

The 1j29b and 1p24b are both tubes with two filaments with common at the bottom of tube (f+) to allow 2.6V Series operation or 1.3V parallel operation.

Russian Rod Pentode Filaments

Note: The parallel filament currents of 1p24b-v are scaled by 0.5x to fit on graph, thus are about 120mA @ 0.6V and 250mA @1.75V.  

In general the filaments seem robust to manage x2.5 over voltage on the 1j17b and remain unchanged afterwards. Life is about 1s with 30V! Indeed I have accidently "burnt" an 1j18b and a 1j24b on the IC breadboard by accidental short of G2 to G1 which was connected to f- via a coil and f+ was connected to HT "0v"

The  graph has three lines, the 0.9V and 1.4V recommended limits with a suggested nominal operating point of 1.2V (normally quoted NiCd) battery.

Originally Silver Oxide (primary and expensive) cells used. They do have advantage of stable voltage and flat discharge. Silver Zinc cells were used by US Military so likely not just Silver Oxide but perhaps more likely was Silver Zinc Rechargable cells. Most sets using the Rod pentodes used NiCd for the LT and to supply a two transistor (Germanium) equivalent of a vibrator pack for HT. Some may have used more complex arrangements from a 12V or 24V truck or aircraft electric system. As was written at the start of this series, there are no known domestic radio sets using these Rod Pentodes from 1950s to early 1990s when production is presumed to have ceased. (Date codes of 1991 do exist!).

It should be noted that Alkaline batteries are higher voltage than NiCd or Silver Oxide for a good proportion of life (Alkaline are cheaper than Zinc Carbon or Zinc Chloride in terms of  time per cent). Also NiMH are about 5% or more higher voltage than NiCd, with the nominal voltage closer to 1.3V than 1.2V. This suggests that it may be advantageous when using Alkaline or NiMH to have a small resistor in series with f- dropping about 0.1V to 0.2V. This allows more negative grid, longer life of tube and battery and lower HT current as well as a slight stabilisation of DC operating point. 

If using a series chain care must be taken to not have a large value decoupling capacitor(s) as the charging of these at power on will pulse extra current in the first filament nearest LT+

Unlike DK96, DF96, DAF96 and centre tapped DL96  to give either a 125mA @ 1.5V nominal parallel or 25mA @ 7.5V Serial, there is no simple relation between filament currents on the Russian Rod pentodes. Each filament seems to have been optimised for the application rather than to meet a particular scheme of current.

The 1j29b and 1p24b are better used in parallel mode for "hot" Class A. Both are envisaged to be used also in Class B  Push Pull (Class AB really), In this case the filaments can be run in series as the quiescent current is low and the AC current can be decoupled (as advised in data sheet) by a capacitor at the centre tap so that for the AC signal the filaments are in parallel, otherwise at full power the RMS rating of Max Cathode current would be exceeded.

To thank the Author because you find the post helpful or well done.

Limits of Power due to Filament 
08.Oct.11 13:19
18384 from 164820

Michael Watterson (IRL)
Articles: 1006
Count of Thanks: 43

With indirectly heated valves (tubes) the power is usually limited by heating effects. But on the filamentary valves the limit obviously must be rise in filament temperature due to the increased cathode current.

Examining the datasheets (and measurements at Vf=1.35) gives

Comparison of Max Cathode current with filament heating current
Model Cyrillic f mA IkMax F/Ikmax
1j17b  1Ж17Б  49 5 9.8
1j18b  1Ж18Б  24 5 4.8
1j24b  1Ж24Б  13 1.6 8.1
1j29b  1Ж29Б  54 8 6.8
1j29b-v  1Ж29Б-B  54 8 6.8
1j29b-r  1Ж29Б-P  54 8 6.8
1j37b  1Ж37Б  59 4.5 13.1
1p24b  1П24Б  198 25 7.9
1p24b-v  1П24Б-B  198 25 7.9

(Only commonly available Rod Pentodes are listed)

The last calculated column suggests that cathode current can be about 1/8th. On the centre tapped 1j29b and 1p24b the datasheets do warn about decoupling the centre tap if the filaments are used in series.

A maxium cathode current is quoted as g2 has some current and with high value load or Anode open circuit the g2 current can be as much as the Anode would be with a short. At low values of load the g2 current drops to a very small value. (Current mirror between Anode and G2 at Anode voltages below "knee" but with g2 at a recommended HT)

Due to the construction of the tube we can assume the cathode current, if filament resistance was neglected would be similar along the wire, though slightly more at the top of the filament (very little more when the tube is in the flat part of pentode curve). This means the current on average is flowing on half the filament since it arrives at "both ends". This means that with say 10mA cathode current (mostly anode and a little g2) on a 1p24b in parallel mode, the f- end ought to increase by about 5mA and the f+ decrease by 5mA?

At any rate the Anode current on the filamentary tubes and thus power is related to filament design current and likely more limited by filament life than emission saturation. The fact that the 1p24b has a 800W pulse rating at very high cathode current also re-enforces this.

Other Rod Pentodes known to have existed

Model Cyrillic f mA IkMax F/Ikmax Application
1k12b 1К12Б 60     Orignal?
1j36b 1Ж36Б 75 n/a n/a Shell fuse1
1j30b 1Ж30Б  15 1.5 10.0 Mixer 12V HT
1j42a 1Ж42A 15 1.3 11.5 Mixer 6V HT
1p5b 1П5Б 120 10 12.0 P.A.
1p22b 1П22Б  115 17 6.8 P.A.2
1p22b-v 1П22Б-B  115 17 6.8 P.A.
1p32b 1П32Б  210 n/a n/a Shell fuse1
2p5b 2П5Б 198 25 7.9 P.A.

1 Since the "Shell fuse" (Russian "Crash Proof") pentodes are rated 2 hours life, the Cathode current is labelled n/a = Not /Applicable for this comparison
2 The 17mA current is for filaments in Parallel. In Series the IkMax is 10mA. This re-enforces the claim that the limit is not emission saturation but extra heating of the filament due to "Cathode" current.

To thank the Author because you find the post helpful or well done.

Rod Tubes - Principle and Design - Sukhanov, Kireev - 1960 
12.Oct.11 05:14
18689 from 164820

Joe Sousa (USA)
Articles: 664
Count of Thanks: 58
Joe Sousa

Fellow Radiophiles:

RM member Dmitri Faguet and myself have finished the English translation of the first Russian Rod Tube article that appeared in the popular electronics hobby press in the Soviet Union in  "Radio" magazine 1960 issue no. 7 pp34-38.


The original issue in Russian can be found in djvu format at Radio no7 1960 Russian.

The benefits of beam focus to keep current out of the screen grid rods are explained. Partition noise is thus reduced by over 2x, and so is wasted current on the screen grid to less than 10% of the anode current. The most efficient of the rod tubes, the 1j24b, only wastes 2% of the Anode current on the screen grid rods.

Best Regards,


To thank the Author because you find the post helpful or well done.

Rod Tubes - Application notes - Sukhanov, Kireev - 1960 
31.Oct.11 19:34
20749 from 164820

Joe Sousa (USA)
Articles: 664
Count of Thanks: 67
Joe Sousa

Fellow Radiophiles:

Once again, RM member Dmitri Faguet translated another very interesting period article about Russian Rod tubes. I very gladly contributed with proof reading. This article gives a series of recommendations and illustrations about the application of Rod tubes in receiving and transmitting applications.


The original publication in Russian was in the October 1960 issue pp49-52 of the Russian Radio Magazine “РАДИО" (Radio). See the original at Radio no10 1960 Russian

There is a general focus on the application of the low noise and efficiency advantages of rod tubes, but other circuits also explore other Rod tube characteristics. The applications include RF front end design, IF design, FM limiters, Mixers, low drift VHF oscillators, high efficiency transmitters and super-regenerative detection.

Dmitri also contributed valuable editorial remarks in his translation, that situate the article in the wider Rod tube development history. Thank you Dmitri!

Thank you also to our fellow Russian Radiophiles who have scanned and posted "РАДИО" on the web.

Best Regards,


To thank the Author because you find the post helpful or well done.

1Ж42А photos - curves 
22.Nov.15 08:41
82264 from 164820

Joe Sousa (USA)
Articles: 664
Count of Thanks: 54
Joe Sousa

Fellow Radiophiles:

Photos, and curve sweeps are now available for the 1j42a.



To thank the Author because you find the post helpful or well done.