Simple micro transmitter

[Murphyv310]

Hi.
Here is the link to the Valves i have used in this and other projects http://www.ebay.ph/itm/USSR-subminiature...58dc11e91e
I think these are more scarce than other types but I highly recommend them.
The seller incidentally is very good to deal with.

[Murphyv310]

       
Hi.
Here is the finalised circuit for the two valve TX.
This works just about as good as the three valve TX that I will post up in a day or two.
Some interesting facts, the current consumption, HT 0.3ma LT 24ma. You'll notice the anode load on the audio preamp is 120k, the anode volts on my Fluke DVM is 3.6v and this stage gives 3 times gain, I haven't fitted a volume control as most mp3 players have a volume control anyway, Line output seems to be around the right level but a volume control could be fitted prior to the audio preamp. One thing to bear in mind is these valves don't like a low value grid resistor so if you fit a conventional pot of 4.7 meg etc you'll need to put the top end to grid rather than slider to grid, strange but true and this seems to be across the board with rod pentodes.
The pi tank circuit is conventional, I've used an old former from a MW coil off a ferrite rod, I have 65 turns of 24swg cotton covered enamelled copper wire, The two chokes on the modulator valve are not critical but the values I have used seem to work best. I am using a 1566khz crystal but any in the AM band will work fine, although the top of the band crystals will work best, the oscillator works fine up to 8mhz, above this frequency it gets a bit tetchy and doesn't always start, The tank circuit would need re-designing if a short wave micro TX is required.
The three valve TX only gives a marginal improvement, possibly a little better audio quality but it does need more drive due to the NFB loop so it could go to four valves.
If you have a MP3 player or other signal source with around 2-3v p-p drive then the one valver is fine.
I do hope others build this and give me feedback.
I'm happy for anyone to reproduce this circuit elsewhere but please mention who did the donkey work.
Have fun.

PS, the audio input cap isn't critical, frequency response is good anyway but if it's found to have too much bass then this cap could be dropped to 1nf or below, I'm using a 4.7nf here.
Oh, on the picture of my prototype you'll see the tank coil has a ferrite rod, with this left in and the "aerial" tuning cap fully meshed the rod will radiate the signal about three metres, the quality is great so good in fact I knackered the speaker in the DAC90A listening to Kraftwerk at high volume!!!

[Murphyv310]

Technical specs of the valves used are here: http://www.radiomuseum.org/forum/142_dat...glish.html

[tiv48]

Thanks for the circuit Trevor
I have never done anything with rod pentodes before and this looks like a great project to start.
Rob T

[Murphyv310]

Hi Rob.
These are wonderful little devices, I've done a number of projects with them and each is a learning curve, there is not a lot about with the exception of J Sousa, a bit from Michael and a few others.
This particular valve is very versatile, it's very low anode voltage and extremely high impedance make it a different beast to what we are used to.

[Mike Watterson]

Important to understand that the g1a and g1b are not like two gates on a FET. It's the same pair of grid plates that pinch the electron beams as on all the other rod tubes, but not tied together.
Thus if the signal is common mode, the characteristic (gain etc) is identical to a regular rod tube. Unless you have a modulator, special AGC, frequency doubler, detector or demodulator, tie g1a and g1b together.

1j42a and high power 1j37b
Frequency Doubler and Detector operation
The differential voltage  is like the voltage on electrostatic deflection plates of a CRT. Actually the tube is totally Electrostatic. G2 and G3 are an Electron lens. Thus no matter what the beam current is from the regular common mode (g1a and g1b portion that matches), the differential voltage deflects the beam onto the g2 rods and g3 rods (or plates, the 1j37b has rods for g2 and g3, two for each on each beam = 8 rods, the pair of anodes are plates. The 1j42a has a slot for g2 or g3, I forget which, and a triangular  pair of anodes). So the current on Anode drops and current on g2 & g3 rises, no matter which polarity the g1a and g1b difference is. So the main signal that is the same on g1a and g2a will be varied in gain. If there is no common mode signal, then actually the differential signal is full wave rectified. If you put a high pass filter, it's a frequency doubler, and if you put a low pass filter, it's a very good detector, sensitive and full wave!

Mixer Operation
You use a centre tapped transformer for the LO drive. Only low level is needed. the + and - phase drive the g1a and g1b. The RF in or Audio in (or IF in if it's part of a BFO detector) feeds between 0V and the centre tap, so both grids are fed equally. The other way of using a Rod pentode as a mixer is to to direct couple a triodised (g3 to f-) Rod Pentode oscillator anode to g2 of the RF amp/Mixer. No other bias needed. This might be preferable for RF in as the LO leakage is less, even though a higher level. A separate Osc and Mixer is always preferable to single valve solution (Hexode - Triode ECH series better than EK or EH etc Octodes, Heptodes, Nonodes. There is actually Triode Hexode and Triode Pentode 1.4V battery types as well as 2V types)

AGC or DC volume control
A remote cut off or variable mu Pentode is created by having a variable pitch grid helix instead of fixed pitch. But none of the grids exist in a rod pentode, it's not really a pentode at all, but a pair of back to back electron guns with electrostatic lenses. It can ONLY work with a filament, a indirect cathode is too fat! The pair of grid plates (the 1p29b and 1p24b are actually using two parallel beams from two filaments, hence an extra filament pin and three grid plates.) are very close either side of the filament to pinch the beam. Normally all the grid plates are tied together. On the 1j42a and high power 1j37b they are not tied together, those types have no more grids than any other rod pentode. The 1j37b has least number of rods!
So inherently there is no remote cut-off, a variable mu design isn't possible. No Rod Pentode has usable AGC via g1. However g2 will vary the gain (hence using g2 as mixer LO in), as indeed it does on EF80, or DF92 etc. My Hallicrafters uses sharp cut off IF amps and DC on g2 to vary gain. For the 1j24b, about 20V on g2 is almost no gain and about 50V or 55V is maximum gain.
So with 1j42 you can have two capacitors to couple signal to g1a and g1b, put one grid leak (1M or 2.2M ) to 0V and the other to the AGC line, or if only using 6V to 9V HT, use a positive AGC line too! The polarity of g1a and g1b doesn't matter, it's only the magnitude of the increasing difference that reduces the gain. Or you can vary  g2 volts from +4V to +9V to vary gain (1j42a only).
The 1j37b takes much more filament current and designed for 60V to 90V HT.

Gammatron mode.
You earth (f-) anode, g3 and g2. You can pick g1a OR g1b as the grid. Then the other g1 is the anode. Even the 1j39b will work at 6V in this mode. I've not tried the 1j42a yet as a Gammatron. The Gammatron is a very old idea, but the early ones were not made well and it wasn't understood as to what was happening! The control grid affects the space charge and thus the available current to the output "grid". The g1 in ALL rod pentodes is a pair of plates on either side of the filament at right angles to beam. The 1j29b and 1p24b has two sets of beams so the middle g1, middle g2 rods and middle g3 rods are shared. They simply has a wider pair of anode plates for the four rather than two beams. Use one filament only on 1j29b or 1p24b and you get close to half power and half gain.

It's a shame they were told to stop developing these (due to silicon replacing germanium transistors). I can imagine a 7660 style beam switching tube for SSB etc ... or logic gates Just a matter of not tying rods together internally, like the simple "snip" and extra pin that created the 1j37b and 1j42a.

[Mike Watterson]

Do I need to put pictures?

[Murphyv310]

Important to understand that the g1a and g1b are not like two gates on a FET. It's the same pair of grid plates that pinch the electron beams as on all the other rod tubes, but not tied together.
Thus if the signal is common mode, the characteristic (gain etc) is identical to a regular rod tube. Unless you have a modulator, special AGC, frequency doubler, detector or demodulator, tie g1a and g1b together.

Hi Michael.
I'm not so sure if you're comments really are helping in this project, are you implying this wont work?? Looking at the spec there is no reason to say it wont plus I have proof it does and that my friend is the ultimate end I've done.

1j42a and high power 1j37b
Frequency Doubler and Detector operation
The differential voltage  is like the voltage on electrostatic deflection plates of a CRT. Actually the tube is totally Electrostatic. G2 and G3 are an Electron lens. Thus no matter what the beam current is from the regular common mode (g1a and g1b portion that matches), the differential voltage deflects the beam onto the g2 rods and g3 rods (or plates, the 1j37b has rods for g2 and g3, two for each on each beam = 8 rods, the pair of anodes are plates. The 1j42a has a slot for g2 or g3, I forget which, and a triangular  pair of anodes). So the current on Anode drops and current on g2 & g3 rises, no matter which polarity the g1a and g1b difference is. So the main signal that is the same on g1a and g2a will be varied in gain. If there is no common mode signal, then actually the differential signal is full wave rectified. If you put a high pass filter, it's a frequency doubler, and if you put a low pass filter, it's a very good detector, sensitive and full wave!

As yet any experiments using the 1j42A as a FW detector have been fruitless, poor performance and with severe clipping

Mixer Operation

You use a centre tapped transformer for the LO drive. Only low level is needed. the + and - phase drive the g1a and g1b. The RF in or Audio in (or IF in if it's part of a BFO detector) feeds between 0V and the centre tap, so both grids are fed equally. The other way of using a Rod pentode as a mixer is to to direct couple a triodised (g3 to f-) Rod Pentode oscillator anode to g2 of the RF amp/Mixer. No other bias needed. This might be preferable for RF in as the LO leakage is less, even though a higher level. A separate Osc and Mixer is always preferable to single valve solution (Hexode - Triode ECH series better than EK or EH etc Octodes, Heptodes, Nonodes. There is actually Triode Hexode and Triode Pentode 1.4V battery types as well as 2V types)

A centre tapped TX cant work as it will be in antiphase to each grid thus cancelling out, so this project cannot work like that,
I'd suggest you build the single valve modulator and report back.
I'm somewhat disappointed at your comments after proving this very simple TX works.


May I also reiterate that these valves do not behave like what we are used to. The use of lowish value grid resistors increase distortion, all the curves in the world wont change how they behave in practical circuits.

[Murphyv310]

Hi .
I'm going to compile a YouTube video over the next few days of the two valve TX, I'll hopefully demonstrate the effects of different grid resistors. I'll also show in detail the feeds to the grids, the HT & LT currents and the waveforms on the scope.
I'm also quite happy for anyone to borrow my prototype for testing.

[Terry]

 I am using a 1566khz crystal but any in the AM band will work fine

Interesting choice of frequency, Trevor - I assume it was purely random, based on availability?

Basically the same channel as used by the Dutch pirate Radio Veronica - 1562kHz/192m - which, it is alleged, inspired Ronan O'Rahilly to open Radio Caroline 5 years later.

However, there were a few 8kHz channel shoe-horned into the very top end of the MW but this was rationalised by WARC '79 which not only made all channels 9kHz wide but also exact multiples of 9kHz. Thus 1566kHz, which is an excact multiple of 9, is the same channel as was used by Veronica in the revised allocations.

You'll be co-channel with two low power transmitters - 500W in Surrey and 630W in Somerset, so I don't think they will be bothering you or, more importantly, you bothering them!

There's nothing of significance in the adjacent channels, either, other than a 300kHz transmitter in Nice, if it still exists (I'm working from a three year old listing at the moment).