1Ж42А. As a mixer/Frequency changer????

[Murphyv310]

Hi.
Well this is pulling out the remaining hair I have. I tried the arrangement of the diodes in a doubler arrangement, it wasn't good, I then tried the diode with no DC path again and things improved but nothing like it was before the disaster. What I noticed was two issues, with two diodes on reconnecting the PP3 it was very loud then gradually fades, I tried the RF amp and the result was the same. With the single diode there is nothing until around 20 seconds after connecting the PP3, this might be space charges but this didn't occur with the original valve. If this isn't something you can replicate with these valves then it's pointless. So I will try two 1J24B instead. Incidentally I've messaged a few sellers in Russia and the Ukraine about the 1J42A and apparently they are unobtainium.

[Mike Watterson]

Something else is wrong!
1) A DC path works better.
2) A single Diode ACROSS the RF fed with a capacitor needs a series resistor before RF decoupling cap.
3) You need RF decoupling, or else the next valve is more RF detector than audio amp
4) The only reason the dual diode peak detector isn't always used for signal level or AM demodulation is the cost of a 2nd diode! It's always better.

The 1j42a was
a) A very late design
b) Hardest to make
c) Very specialist application.

So unless you really need 9V HT or the special difficult to use mixer action (which needs two valves for mixer and oscillator) the 1j24b is a better choice.
While a one valve mixer/oscillator does work (using same principle as VHF valve mixer/oscillator), using two valves is dramatically better, still less than 25mA (22mA typical) and it's close to Hexode-triode performance, same circuit works. See the DK96 -> 2 x 1j24b thread.

The 1j29b is superior for frequency response, transmitter PA, audio PA and RF front end with big overload margin, but 2 x 1j24b in Class AB about 1/2 filament current of one 1j29b and similar or more audio or RF power than a DL94. An entire superhet radio with 1j24b and 1N60 is possible. I know the 1j29b is about as good as EF80 for VHF, but I don't know yet if the 1j24b works.

For headphones or lower speaker volume using 3 x PP3 (18V end of life, approx 28V fresh) works and depending on the part of circuit then you use the full HT or the 2x PP3 point as HT for the g2, to solve the issue that a single series grid resistor isn't great, you need a potential divider (wastes HT) or an HT tap.
I've also considered 2 x 12V alkaline minicells for the g2 supply.
For serious volume you want 45V to 90V HT. The 45V with a push-pull AB dual 1j24b should be good enough. Also PP has the advantage any centre tapped primary mains transformer works as the standing currrents cancel. Use a low pass filter from the oscillator grid to get the -Ve bias for the output stage.

Edit:
And people were looking for the 1j42a for years before I found one supplier! It's rare. Though there are other rod tubes (maybe mostly from late 50s early 60s) that have not been seen.

[Murphyv310]

One last point is the valves have been stored near a powerful speaker with a big magnet. I'm wondering if that is the issue, will try degaussing later on.
I've been through all the Diode combinations again and yes I've tried series resistor before decoupling with no success. All very odd as I've made loads of regen radios in my life, first one 54 years ago.

[Murphyv310]

Hi.
Progress again and very odd to boot.
I have a tape head demagnetiser.......... Somewhere. So I found a wound coil with a suitable hole in it for a 6mm Allen key. I connected it to 45vac and it gave a really strong pull. I then degaussed the audio valve and after ward the output nearly doubled. So the 1J42A appears to be effected by magnetism. Take one even a foot away from a mains transformer and it buzzes through the phones far more than any other valve I know, I'm not saying that it's the reason for the residual magnetism as they are stored near a speaker.
Performance isn't as good as before but at least it's much better.

[Mike Watterson]

So the 1J42A appears to be effected by magnetism.

Yes. It's mentioned by Joe Sousa on the Radiomuseum articles and in the Russian literature. It's a pair of precision low power electron beams, in wedge shape. That's why the Russian stuff usually has tinned steel rather than copper or aluminium screens.
I use cut up Lidl coffee tin.

[Murphyv310]

Thanks Michael.
I'm busy with the B16T so when I've time I'll make some sreeens with thin ferrous metal.
I've now got four 1J37B's to play with at a later date.

[Mike Watterson]

If you had a suitable ferrite or laminations to make a C shape the thickness of the active rod height and have flux / pole tips at right angles to the anodes, then have a coil on that?
I've wondered would it be a way of using 1j24 as a mixer, put f1 on grids and f2 on the coil.

[Murphyv310]

I'm sure it would Michael. At what level would you require though? I'd surmise it would need to be fairly high.
In fact the valve that was blown up was the one that was in my one valve pantry transmitter and must have been run for over 1000 hours in total, perhaps they improve with use.

[Murphyv310]

Hi.
Well I've now tried the 1J18B and a 1J24B in the RF amp position. Running at 36 volts HT I find that selectivity is pretty poor even with reaction at the point of oscillation, my local is heard at low level right across the band and it floods the two adjacent stations. Drive is good though and using a 1J18B in the audio its powerful enough to drive a speaker at listenable level.
For definite the 1J42A is the best valve for the RF stage.
My plan now is to rebuild this on another board as this one has been hacked about with and is getting rough. I'll do this on a section of double sided copper laminate board for screening with a better layout. Valves will be a 1J42A & 1J18B or 1J24B. May be a couple of weeks as I'm concentrating on the Pye B16T. After the restrictions are over we will be returning to the work we do for vulnerable people in the community that has been stopped by the pandemic. So I'll have a lot less time to devote to projects.

[Mike Watterson]

The 1j18 is poorer than 1j24b or 1j29b. It's an older type.
The 1j42a is only better than 1j24b or 1j29b at low HT. In a proper circuit, the 1j24b is far better RF than 1j42a and better audio than 1j24b. The 1j24b replaced the 1j18b. The 1j29b is higher gain, higher power etc, but about x4 filament current of 1j42a or 1j24b (which are the same). The 1j18b is about twice the 1j42a or 1j24b filament current.
The 1j24b and 1j29b gain/performance depends on an optimal "stiff" g2 voltage. You can only tie it to HT, if the HT is optimal for g2, else you need a potential divider, not a single resistor. Look at g2 current to see why. Or else a tap on the HT battery pack at about 54V or 63V for 1j24b's g2, and higher for 1j29b's g2.
You can use 1/2 the filament of a 1j29b for same filament current as a 1j18b but better performance. The only penalty is 1/2 the gain (which is still high) and 1/2 the power which is still higher. The 1j24b is OK for 27v (poor) to 90V HT designs, the 63V or 72V is good. The 1j29b is about 54V minimum, good at 72 to 90V and excellent at 120V.

A pair of 1j24b is about the same filament current as a 1j18b and gives about x4 power in Class AB, and then can use a small centre tap mains transformer without saturating. I used a VFD indicator as phase splitter, but another 1j24b or a transformer is lower power. The only advantage at all for the 1j42a is the 9V HT or the unusual mixer action, which is kind of opposite to how a dual gate FET works.