Joined the club, Rod pentode superhet radio.

[Mike Watterson]

Oh and compared to ANY valve the Rod Pentodes have nearly non-existent internal shields. The cans or perforations you see on most valves are screens to replace the outer conductive paint, not the anodes. The screening is just some support rods connected to f- or s or g3. The 1j37b has the fewest screen rods. I make a springy holder from cut up coffee tin or similar as the rod tubes are also sensitive to magnetic fields.

[Murphyv310]

Hi.
The Band 1 405 line modulator oscillates perfectly at 45 mhz with a 1ZH37B even down to 18v. A third overtone 54mhz crystal also works.
I think I'll have a go with higher frequencies in a week or two.
Right now I've a 1948 Marconi VT50 taking up all the bench space, a restoration for a friend.

[AdrianPH]

If I get chance I will do some basic tests with Mixers of various types, I have two synth signal generators I can use for signal and osc sources, if I wind a 10.7 Mhz coil for the anode load I could possibly use a SDR receiver stick to measure 10.7 MHz output, or perhaps do a several stage 10.7 IF with a simple detector. All valve mixers of course, and simple types that I can follow.

Adrian

[BusyBee]

I am working on a signal source as well as a local oscillator to test my circuits. It's a little long winded in the signal source though as I am using a crystal oscillator with multiplier chain for frequency precision plus the ability of modulation which requires multiplication such that the deviation is multiplied through the chain. It should be a good long term solution though and I will be able to test it at the lower frequencies as they are multiplied with my existing test euipment.

I do have an SDR receiver. Mine does not actually work at 10.7MHz (as many others don't I think), it's range starting at around 24MHz. There are HF converters about for them but it gets expensive. I will be able to use mine for checking at the higher frequencies.

A while back I remember buying one of those diode balanced mixer modules. It would be worth testing alongside for comparison and would probably be a good piece of test equipment in itself.

Tracy

[Mike Watterson]

Certainly the 1j37b ought to work as osc, mixer or mixer-osc up to 60 or 70 MHz, but the 1j24b and  1j29b are probably superior. Unlike a Wunderlich
https://www.radiomuseum.org/tubes/tube_wunderlich.html
or a dual gate fet it's actually just a rod tube with the grid plates that are either side of the filament brought out to separate connections.

   
Top view of all single filament rod tubes. The dual filament types have a 3rd g1 plate between the two filaments as well as an extra pair of g2 rods. All the tubes have two wedge like electron beams electrostatically focused by g2 and g3. The pair of g1 plates "pinch" the beam and eventually cut it off. If AC is applied ONLY to g1, biased at zero volts then the two g1 plates act as anodes of a rectifier.
   
The 1j37b and 1j42a can thus be two rectifiers to 0V. The output at g1 is negative.
The rod tubes with g1a and g1b, i.e. the two plates are not internally shorted will thus steer the two beams with any differential voltage, so to enhance the drop in anode current the anodes are wedges instead of plates.

No doubt prototypes where made with the two pairs of g2 rods or anodes brought out separately, but it's obvious that a differential voltage on the two g1 plates will have the greatest effect as they are very close to either side of the filament.

Here is sketch of one half of the tube, one beam:
   
The dotted rods would be screen rods, connected to f-, f+, s or g3 wires depending on model.

Here is 1j42a, the 1j37b is similar
   

and to compare, any single filament tube with the g1a and g1b plates internally connected:
   

Advantages of the 1j37b over the 1j24b

• The grid capacitance is lower in some circuits because there isn't the internal short and the g1 pair are separately connected.
  
• Frequency doubling is simple. Feed AC differentially to g1a and g1b.
  
• Detector has more gain. Feed AC differentially to g1a and g1b.
  
• DC gain control can use a low positive OR negative  voltage on g1a or g1b, or differential. The 1j24b needs 15V (near cut-off) to 55V (max gain) on g2 assuming 80V HT. Varies with Anode volts/HT. The control range of the 1j37b is largely unaffected by HT/Anode/g2 volts.
  
• Mixer operation uses a low level LO. But any LO is likely x20 level needed.
  
• A full level LO can ALSO be fed to g2 as well as differential signal on g1a and g1b.
  

Disadvantages of the 1j37b over the 1j24b

• The common mode g1a and g1b capacitance may be higher due components needed.
  
• The filament current is 58ma to 60mA vs 11mA to 12mA at ideal 1.25V, FIVE times!
  
• The LO needs attenuated. The 1j24b g2 can be driven direct and biased at DC by the anode of the LO valve.
  
• The transformer balance to g1 and g2 is critical and centre tap common mode gain must match. Tricky as frequency increased.
  
• DC control of (differential or one of g1 and g2) must not affect RF match at g1 or g2 or max gain is  reduced and there can be partial frequency doubling/rectifying (though you might want this).
  
• Less screen rods.
  
• It's a larger tube.
  

Filament currents vs voltage is interesting.
   

I'm not sure on the differences on electrode capacitance or maximum frequency.

Note the 1j24b used to replace an existing valve in an aligned radio needs padding cap on at least anode as it is lower capacitance than DK91, DK92, DK96, 1L6, DK40, 1T4/DF91, DF96 on g2, g3 and anode. I'm not sure about DF97 and DC90. I'm not sure if grid needs padding in any circuit. I do plan to replace a 1L6 in USA Worldband sets (I have maybe 4 models). I have to add series and parallel resistors in filament circuit to match the 50mA nominal at 1.35V approx series operation as even two 1j24b as triodised osc and pentode mixer is less than 25mA! I've done one pair with additional capacitors to replace a DK96 without retuning LO or IFT.

The DK92 (without adaptor or DK40 with adaptor) will directly replace the 1L6, but mucks up alignment at shorter wave bands. Some UK/Europe sets used DK40 with Dx9 series valves till the DK92 was available. The DK91 and DK96 are inferior, hence even some 1957 model with SW used the DK92. 

The electrode capacitances of all rod types is very low compared to regular B7G types.

Also remote cutoff/variable mu while allowing greater AGC range causes intermodulation if there is an unwanted strong signal. It's achieved by a varied spacing of the helical grid wire. All rod tubes are more linear with a sharp cut off.  Some Western commercial or military sets didn't use variable mu valves except at IF to give better IM performance. Also tended to not to use heptodes or octodes (as was common even on USA mains sets) but a separate LO and mixer for better IM performance, no FM modulation of LO and less LO leakage to the aerial.

[Mike Watterson]

The 1j37b and 1j42a can also be used as the earliest ever kind of triode, the Gammatron.
Use one of g1a and g1b as the anode and the other as the grid. Tie actual anode, g2 and g3 to 0V or f-.
It won't work very well, but neither did the originals, hence the second idea was the open wire or mesh between the cathode and anode.

What you can't do with a rod tube is simulate the Tetrode for a tetrode negative resistance oscillator. You can with a real pentode on some tubes by connecting g2 to g3, or even ignoring g3 (suppressor grid). The rod tubes are NOT pentodes, just same number of grids and similar anode characteristics. Hence when triodising you connect g3 to 0V, f- or f+, never to anode and g2 as with real pentodes. The EF80 and DF97 are often triodised at VHF for mixer-oscillator use, in which case g2, g3 and anode are shorted.

In common grid mode you need a pentode where g3 has a separate pin, not connected to filament or cathode. It goes to 0V. A battery valve/rod tube CAN be used in common grid mode. The f+ and f- are fed by RF chokes from LT- & LT+ and the RF signal via caps to f- and f+. At lower frequency where a core is needed to get enough inductances you use like a 1:1 transformer with the f- and f+ DC cancelling (think of bias cancels in a push pull transformer).

[AdrianPH]

Not getting emails about the thread so sorry for delays in responding. I get emails from other forums, just this one seems to miss for some reason.

Tracy, Two of my SDR's have the HF converter built in so I can go down to about 500KHz with them. I also have a Hackrf One which goes from 1Mhz to 6GHz if I reload the software for it.

Thanks Mike for the information, it will still take me some time to understand it all, back to reading Scroggie.

Adrian

[Murphyv310]

Great piece of info Michael.
The gammatron has always interested me and I'd love to have a fiddle.
Just wish I had another couple of 1J42A's