Russian rod pentode circuits

[Murphyv310]

Totally agree
A very useful resource.

[Alistair D]

So useful that the link will be in the next version of the SDS spreadsheet.

Al

[BusyBee]

Hi everyone. Just a quick update as I have re-visited a module I made a while ago (before my TRF receiver) when I was inspired by Trevor's super-regen.  Originally I got a little confused with it and did not get good results. Today I was modifying a crystal radio  type coil / capacitor setup and decided to try it with my module. I just used a single turn using my hookup links for regenerative feedback.   I still get a little confused as to what is happening, but on later checking I find that the tuning was over LW and MW as I received RTE Radio 1 and Caroline (plus one or two BBC stations) with excellent sound quality (with a bit of fiddling). I was using my long wire aerial (length of loft) and earth to hot water tank across the landing.  I found that the setup gave good results so may well put together as a permanent unit at some point. The module uses a 1j29b in a fairly standard regen setup using a level pot on the screen grid. Perhaps the only difference to traditional circuits is my use of a transistor (driver?) transformer as output transformer to an external amplifier (I used my Germanium transistor based bench amp). The low tuning range is defined by my currently paralleling 2 x 500 pf capacitors. I may well change that, but it is perhaps unlikely that the tuned circuit I used will be permanently with the regenerative circuit.

Tracy

[Kalee20]

They are though sharp cut off (apart from odd circuits for 1jxxa dual grid types), serious AGC needs an amp and drive to g2.

I'm looking at AGC with rod pentodes!

I'd discounted varying Vg2 for AVC purposes. True, it does control gain. But, reducing Vg2 reduces gain, and also reduces cut-off potential on g1. So signal handling is reduced as a result of the reduced grid base.

For AVC purposes, the output is relatively constant. It's the input which varies. When the input is large-amplitude, reducing Vg2 would reduce the input-handling capability, just when we really need it. With a strong signal from a modulated signal generator, I'm seeing the AM disappearing as I reduce gain with Vg2 - I have made a super limiter...

I have experimented with varying Vg3, finding (with 24V on g2) that gain reduces to zero as Vg3 is taken to -25V. But the down side is that anode resistance ra seems to drop too, meaning that damping on a tuned circuit load increases as gain is reduced - selectivity reduces.

Trials will continue over next few weeks, but I thought I would tap into any already-existing knowledge on here in the meantime!

[Mike Watterson]

The Russian radios using rod pentodes used the 1j37b for AGC/AVC, because inherently you've only got some gain control via Vg2 and it's not great for very big signals. They did also experiment with Vg3.

The Vg2 does work reasonably, nearly as good as a DK92 mixer, for LO injection to make a mixer. One Vidor model in 1950s uses a DF97 this way for MW & LW mixer, which is a 1st IF on FM. On FM a DF97 is a mixer osc as on German sets or the EF80 on FM. On AM, that valve is the LO only.

Both grids together for signal on 1j37b. Differential voltage on the pair of g1 varies the gain very like variable mu. See Radio museum article. Naturally that also works as a mixer if one signal is the LO and one is RF, then AGC can also be added.

The 1j42a should allow something similar, but it has lower Ra and a very low max Va. Also rarer.

Another alternative is 4 x matched 1N34 or similar as they pre-date rod pentodes as ring diode mixer. Copper Oxide was used in the 1930s for SSB mixers on telephone trunks.

[Kalee20]

Thanks Mike!

I do have a few 1j37b's. In the manner you mention, I take it the two grid rods are going up and down together, so cathode current is varying cyclically - but with the two rods at different DC potentials, the electron beam is steered off-course, so that only part of the beam hits the anode (so the anode current, and variations therein, are reduced - giving gain control - and the electrons which go off-course are captured by g2).

Incidentally, the Vidor you mention must be the Vanguard. I do have one, as yet untried!

[Mike Watterson]

Yes, probably the Vanguard. It works well on VHF-FM with the loop aerial disguised as beading hiding the cloth seam. The main restoration issue was the faulty wavechange switch. Probably corrosion from the batteries.
The LW & MW is good too.
I cut one earth on the first DF97 socket and it worked with a DC90, no noticeable tuning shift.
The entire design might work with 1j24b, but with no AGC. A few 1950s Ever Ready sets only had AGC to the DK96 grid, just running the DF96 at full gain. Using even 1j24b + IN60 diode instead of DAF96 and a 1j24b (or pair in push-pull) for audio out.

I think your description of the 1j37b is correct, except it seems that the performance and implications of that are curious. You can use a LO at HALF the desired frequency, suggesting it could be used to implement a stereo decoder. The anode would have L-R audio.

Also if there is a tuned RF pre-amp you'd feed both grids with differential LO and common mode RF, but if there is no preamp, you use a RF transfomer to feed RF differentially to g1a and g1b, but feed the LO to a centre tap on the secondary, so the LO radiated can be nulled out. LO then is the common signal.


I meant 1N60 in above post, not 1N34, as it's more available and also pre-1950 originally. Though the current 1N60 is simply the same part number.

[Murphyv310]

Nice to see Michael around and the thread revitalised.
Big changes since September 2018 for me!

[Jez]

Well this is all very interesting! Where/what is the best source for some of these valves? 1ж29б seems most interesting.

[Mike Watterson]

Ukraine and Bulgaria. The Ukraine suppliers were offline or unable to ship for a while.
The 1j24b (not 1j24p, it's a monster) is the most versatile to replace most 1.4V valves. Need a small ballast resistor on f- and may need a pull down resistor on g2. The 1j29b only needed when more than 400mW needed.
Any mixer can be implemented as plug in replacement using 2x 1j24b, one as a triode oscillator and that can direct drive g2 of the RF in/Mixer. Less LO leakage than DK91, DK96, DK92 and 1L6. You need to add extra capacitance to have tuning free plug-in replacements as the 1j24b (and rod "pentodes" in general) have very much lower capacitance. The 1j24b will replace a DF97 as triodised mixer/oscillator or as Pentode Mixer (Vidor VHF/MW/LW set) or 10.7 MHz IF. Will also replace DAF9x (add 1N60), DL91, DL92 & DL96. Really a DL94 at full wack needs a 1J29b, but a 1j24b can be used with some bias adjustment in some sets. A pair of 1j24b in push pull class B can certainly do HiFi at 1W.
https://www.radiomuseum.org/forum/russia...es.html#11

The 1j24b can typically be at about 11mA at 1.2V (small series resistor for Zinc, Alkaline or 1.4V supplies!). The 13mA is a worst case!

Maybe link to my graphs
https://www.radiomuseum.org/forum/russia...es.html#18

Dmitri Faguet also has done a lot.