Rod Pentode TRF

[BusyBee]

I must admit some of the information floating about is very confusing, as good as it may be. Personally I prefer to understand the original Russian use before modifying as they appear to follow things more conventionally. I would expect them to modify things only if there was a clear advantage although I would admit that it would be sensible for them to keep things as conventional as possible to ease the burden of training with many different Warsaw pact forces. I was reading the pair of translated ex Russian 'Radio Magazine' articles yesterday (available on the Radiomuseum site). They answer quite a few of the points we have raised here. The issue I find is that western sites / articles / people have put their own slant on these valves and how they should be used. A good example which really throws a spanner in the works to me is the Gammatron circuit put together in a rod tubes article on the Radiomuseum site. Although I can see the point it really goes off at a tangent and may inspire, but really confuses their use to me. I also tend to be a little wary of some of the data given. As an example - the translated Russian documents mention that there can be massive variation of the characteristics of individual valves of each type. They point out that this is the reason for using a potential divider for g2. This may well explain the variation of results obtained with circuitry.

Tracy

[Kalee20]

the translated Russian documents mention that there can be massive variation of the characteristics of individual valves of each type. They point out that this is the reason for using a potential divider for g2. This may well explain the variation of results obtained with circuitry.

Not only a potential divider, but an adjustable one at that!

I've got an AF circuit using two 1j24b's, from same filament supply and same HT. Each has a 270k anode load. Each has grid-leak returned to 0V so zero bias (and measuring with a DVM it really is zero.

But to get the same 45V quiescent anode voltage, one needs 23.6V on g2 and the other  36.4V.

I'm sure I could run them from the same potential divider but then I'd need to provision adjustable bias for g1.

I haven't tried any other specimens, unfortunately it's a soldered-together circuit!

I'm just accepting that I'll have to have several trimpots.

[Murphyv310]

Hi.
Interesting posts both from Tracy and Peter. I have also noted variations in specimens of the same type. The 1j42a can be quite wide. I also agree that a potential divider is the best route for the G2. In this little project I deliberately didn't use them as I've left the HT unswitched.
I often wonder if the performance of the original 2 valve TRF was dictated by a 1j42a that was bang on for the project and when it failed due to my mistake a replacement didn't cut the mustard.
I suppose that the differences between valves just make the whole aspect of these valves more interesting or time consuming which way you want to look at it.
No matter what, experimentation seems to be really the only way.

The other thing I've come across is like a "running in" effect. I noticed this early on in 2016 when I got my first batch of 1j18b valves, after a few hours of running characteristics changed and for the better. Also in this set the HT current was a total of 4.8ma (with the higher voltage valves in the audio) now its settled at 5.5ma with an increase in audio output before clipping. Also seen this on the 1j42a with an increase in reaction level. Values may need to be adjusted to reflect this and of course filament volts as this effects bias. I believe bias is derived in the filaments too of course and that's why gain can change with filament volts, not directly proportional to heating effect IMO.

[Murphyv310]

If we look at the "D" series valves and how they behave in a battery radio, we see that a fresh LT battery giving often more than 1.5v gives the set a boost, as it reduces to 1.1v performance takes a noticable drop.
The Rod Pentodes though often go the other way, with a reduction increasing gain. At the moment the project filament volts are 0.93v and its performance is the best so far. One day I was fiddling with a 1j42a and unknowing to me the battery had dramatically dropped to 0.6v and with only 3v on the anode it was quite amazing on gain, not as high as the sweet spot but working well, a small increase or drop in filament volts and gain went well down, perhaps there is two sweet spots.

[Kalee20]

It's interesting, but my curiosity doesn't extend to operating them while detonating a small yield nuclear weapon in the potting shed

I've been thinking of this...

A nuclear explosion liberates energy, as heat, light, and higher-frequency gamma radiation. Plus there are copious quantities of sub-atomic particles. What there isn't, AFAIK, is a large-energy magnetic pulse of the type you'd get if you discharged a huge electrolytic cap into a big coil.

That being so, I'd expect the pulse of gamma rays to disrupt electron flow, but as the transit-time in these little valves (nanoseconds) is going to be much longer than the period of gamma oscillation (picoseconds) there will be an averaging effect and not much variation of current will occur. I'm assuming that eddy-currents induced in the electrodes is insufficient to heat them and damage them.

The sub-atomic particles will easily penetrate the glass envelope and play havoc, but only for a short time. They may disrupt the crystal structure of the matter of the valve, but unlike transistors and IC's with their micron-sized features, it won't matter.

If the searing heat is shielded from the valves, they won't overheat.

I'd reckon they'll survive and work afterwards.

Somewhere in Russia, there must be data on this. It'd be a shame if Amie had to sacrifice her potting shed in the interests of research...

[AdrianPH]

MI5 or the men in white coats will be along soon! 

Adrian

[BusyBee]

I don't think I can do much about the electromagnetic pulse as I haven't even got lots of Farads and a big coil let alone a good stock of Uranium although the distances do make a difference so a tiny discharge right alongside would be just as effective as a large one out in the garden.

With regard to the variation in parameters within each type though I may be able to get somewhere. Some time ago I constructed a test unit for my valve test gear which uses an IC zero insertion force socket purely for subminiature wire ended valves. Using that it is easy for me to exchange valves without any soldering in a test circuit. It may be useful to put together a basic amplifier and subsitute valves, tabling readings of anode and screen voltages / current, gain etc. The only bugbear is obviously the running in time. I have come across this with all non recently used valves but maybe it has greater effect with rod pentodes? For DC work settling times of hours are often quoted which will be impractical and not really needed for our purposes unless someone is attempting an electrometer or analogue computer. Perhaps to try a couple first and see how readings change. My heater supplies are always regulated DC and if I remember correctly I have made a special regulator for these valves for my valve tester (which is home made and more attuned to switched point curve tracing than go / no go). Testing would really be a bit limited as all of a type would be from one batch (as bought) rather than randomly selected from different makers etc but would likely give some idea.

If anyone has any suggestion before I start please let me know (as well as letting me know if this is a poor idea). I may not test all but I do have 1j17b, 1j24b, 1p24b, 1j29b, and 1j37b types.

Tracy

[Kalee20]

But I don't think a nuclear blast generates the same sort of EM pulse as a coil with a burst of current through - so failing this latter test (which Amie has shown it would) doesn't necessarily mean the valves will die in a nuclear blast. I tend to think (very basic reasoning, as outlined previously) that they'd survive.

Your filament supply being regulated - so is mine! I have nevertheless seen a slight drift over some seconds, nothing which alarms me though.

[BusyBee]

the translated Russian documents mention that there can be massive variation of the characteristics of individual valves of each type. They point out that this is the reason for using a potential divider for g2. This may well explain the variation of results obtained with circuitry.

Not only a potential divider, but an adjustable one at that!

I've got an AF circuit using two 1j24b's, from same filament supply and same HT. Each has a 270k anode load. Each has grid-leak returned to 0V so zero bias (and measuring with a DVM it really is zero.

But to get the same 45V quiescent anode voltage, one needs 23.6V on g2 and the other  36.4V.

I'm sure I could run them from the same potential divider but then I'd need to provision adjustable bias for g1.

I haven't tried any other specimens, unfortunately it's a soldered-together circuit!

I'm just accepting that I'll have to have several trimpots.

I have conducted a few test over the past couple of days and my results seem to compare similarly with yours.

Please see the attached information.

Initially I got a single 1j24b and tested it at three different voltages. Please see the diagram. The nominal 100V HT is in pink, 60V in black and 20V in green. Only the anode and g2 resistors were changed, for optimum gain. Please note that these resistors may not be optimum as my switched range boxes have limited values but general coverage is Ok. The other components were fixed in value (generally part of my test gear). Input signal was set to 50mV p to p 1kHz sine wave. The output signal was measured on the output, both being taken from the oscilloscope screen so of limited (but acceptable in this context) precision. The input probe was x1 and output x10.

Below the diagram the three lines give the results of initial tests. Note – The distortion was that of the output waveform noticed by increasing the input signal and was very subjective, just really and indication that the amplifier was behaving sensibly and, as such, is not particularly useful for these tests.

From these initial tests I went on to test four further examples of the 1j24b at nominal 60V with the resistors determined for maximum gain from  the previous tests (100k Ohm for anode and g2 resistors). As a further test I tested each again with the anode load increased to 270 kOhm.

The results indicated a wild variance with the 270 kOhm anode load (no change of g2 resistors) so I then went on to do a set of further tests with a 47 kOhm anode load. This did give sensible results although lower gain as would be expected. For the last numbered sample (#4) I did afterward try a different g2 resistor and found increased gain.

As regards valve run in and warm-up, I found some variance in the first few minutes of running but after that, little change. That said I have not run things for hours. These valves have not been used since I have had them, and perhaps not for years but they are all NOS. I practice they seem similar to the western NOS I have as regards initial use.

Conclusions:

I have found great differences as previously reported and the results do indicate that individual setting up may well be beneficial. My samples were 1j24b's. I do wonder whether 1j24b-v's would have closer characteristics but I do not have any to compare.

If anyone has suggestions I have put the test valves to one side. If I think of any other tests that would be useful I will try but, at the moment, I'll let things sink in. If anyone has any question please ask as I may obviously have overlooked something.

Tracy

[Mike Watterson]

Another reason for a potential divider on the G2 is that over a significant range the g2 is a mirror of Anode current. Not a problem where the g2 connects direct to a battery or HT rail, but otherwise strange things can happen, especially on an IF amp. The current variation drawn on G2 means the voltage can change dramatically with load current. On the 1j24b the gain is very low when g2 is below 20V and is at a maximum at about 45 to 55V depending on other voltages. As there is no variable mu / remote cut-off on normal rod pentodes (achieved on regular valves by increasing spacing of the spiral of g1 turns along the length) and amplified AGC to G2 was often used. It's also why the rod pentodes are reasonable mixers when g2 is fed direct from the anode of an oscillator that's biased to about 40V DC.

I'm not sure that the variability on later models is much different to regular valves, which have quite a variation in Ia vs vg for new ones, hence push pull using adjustable bias or large enough cathode resistors (negative feedback). Naturally direct filaments can't use a cathode resistor, though from about 1937 most battery sets used a series resistor from -HT to 0V to bias the output valve. Some 1950s German sets used the fact the LO grid goes negative and filtered this to get a more negative bias for a pair of DL96 in near class B push pull, but the quiescent current on battery push pull sets isn't well matched.

But the Rod Pentodes SEEM less well matched than regular battery pentodes due to the sensitivity of gain to g2 voltage.