Replacing a DK96 wiith two 1j24b

[Mike Watterson]

Here is the high performance 2x 1j24b to replace a DK96
   

   

Circuit
   

Note, this isn't suitable for a 1j42a (too high HT) and not suitable for 1j37b.

It should work for any battery Heptode/Octode. However the pin connections vary and series valves need a ballast resistor between the B7G filament pins for 24mA or 48 mA approx, thus for series it's not good for the DK96, but may work for the 1L6, DK40, DK91 and DK92 etc.

[BusyBee]

That looks a useful idea!

Tracy

[Murphyv310]

Thanks for sharing Michael.
One question, where can you get B7G plugs these days?

[Mike Watterson]

No idea.
I cut up zinc plated garden tie wire as pins.
I used a B7G base protector as a template to punch holes in plastic.
Set pins through with solder blobs at top
Added glue gun.
Soldered on the Rod Tube(s)
Removed from base protector.

I've been looking for B7G plugs intermittently for years. Only Octal and the related higher pin count plugs seem to be widely available. I tried a broken valve base, but you can't solder to those. The internal structure is welded on.

I'd be able to make nicer ones if I had plain paxolin or tufnol sheet.

[Murphyv310]

Thanks Michael.
I have some tufnol sheet and could use a hole cutter and retrieve the waste part to make the base. PCB pins might work, I think they are thinner gauge to valve pins though.
I've a couple of battery sets that I'd not mind modifying.

[Mike Watterson]

No modification at all needed to the actual radio to replace all the battery valves. Should be no issue on any parallel filament set.

Also the 1j29b with filaments in series can replace 2V valves and run off the 1.8V to 2V lead acid. I have a pair inside the envelope of a KLL32 that arrived cracked at the base. The two homemade screens give it a similar internal appearance. I had to add two resistors as a divider from external g2 to internal g2 and f-  for stability and also so the g1 was at the same bias point. But this was inside the package. The g2 and g3 of the two valves tied.

McMichael 484, a 1948 2V lead Acid set. Basically old pre-war "British base" valves simply given octal bases in 1945/1946. A fifth of homes had no electricity in the UK in 1948 and a Lead-Acid Table model was much cheaper to run than an "All Dry" set.

Since the 1j24b will work down to 0.9V or possibly less, the circuit above might replace a KK32, with possibility it might work with the two filaments in series. I have tried a DK92 in a KK32 socket, with a resistor on f+ to drop the 2V to 1.3V, that worked perfectly.
   
Schematic 

As received
   
(cracked at base)

   
Old insides

New insides
   

And envelope glued on!
   

Home made B7G plugs
   

[Mike Watterson]

I've discovered that a jumbo "chrome" effect paper clip makes good pins as it's got a steel core, stiffer than zinc plated garden wire and solders easily. Most importantly it's almost a perfect diameter. The garden wire I found is too fat and my tinned copper wire I used before only works on "perfect" sockets as it's a little skinny.

I've found a load of pin protectors that have eight holes, wide at one side and pin sized at the other end, so they sort of work as straighteners too. They have two bands, really fine grooves, on the side thus the bottom 1/3rd can be cut off. Most are some sort of thermoset plastic or Bakelite. A few seem to be softer stuff.

So I set a slice on an uncut one, poke in straightened paper clip and cut off about 1.5mm above the surface and then bend over outwards pressing with the flat of a screwdriver, then pliers. Solder blob goes into the wider gap at the top of the cut disc. Finally prise out the new plug when all seven pins are done.

Photos next time. I'm making up two, one triodised for the oscillator and one where the oscillator is a Pentode with the g2 going to the Heptode or Octode final screen grid (variously numbered) that takes HT.

The two g3 wires, two screen wires and two f- wires all bend gently to 180° and solder to the piece of coffee tin shield bent to hold both rod tubes. That then has a single heavier wire to the B7G f- pin. The two f+ wires are sleeved and also bent up to connect to 2 x 10Ω resistors to drop 0.2V maximum. The version for 50mA series or 25 mA (nominal) will have one of two values of resistor from pin 1 to pin 7, as the total current of both Rod tubes is about 21mA to 24mA. That might need adjusted on test; I maybe should measure all my 1j24b tubes to see what the variation of current is at the same supply voltage. The current is increased by the mixer HT current and the oscillator HT current.
Another aspect to look at is that on 67V and 45V designs the screen grid HT might be connected direct but on 90V it's via a decoupled resistor. Thus oscillator g2 would need a pull down resistor for the 50V (a 90V HT). The triodised version of the oscillator has no HT for a screen grid and the mixer g2 in all configurations is biased by the oscillator tube's anode voltage as you want about 35V with 20 to 50 swing for maximum conversion, ie about 30V AC, though even 5V drive does convert the input RF.

So though I know the circuit worked to replace a DK96, I didn't make any measurements. I'll also want an RF probe that doesn't load a shortwave LO.

[Mike Watterson]

A triodised (but g3 to 0V) 1j24b working well as the oscillator section of a DK96 circuit, though aim is to replace an 1L6.
LW has 40 V pkk RF
MW has 30 V pkk RF
The original DK96 does also give more on LW than MW.
The DC is about 40V.
So that's nearly perfect to drive g2 on a Pentode RF amp as a mixer, either the 1j24b or anything else.
Also means same circuit will replace a Triode-Pentode or Triode Hexode.
Also triodising means no pesky g2 HT, which isn't very compatible between Rod Tubes and conventional Pentodes or Tetrodes unless their g2 is tied to a battery tap; if it's just a series resistor you need to know the value and add a second resistor to ground to give about 2/3rds to 3/4 HT.

Next looking at the Pentode RF amp as a mixer by driving g2 direct from anode of oscillator.

[Mike Watterson]

I'm obviously holding it wrong, or have made a wiring error as this originally worked well. I shall have to retrieve the two sets I used then from the attic. It was a few years ago and I don't remember it being poor, in fact I thought it was better. The mixer gain is very poor.
On the plus side, the LO is at almost the same frequency, possibly very slightly less capacitance as a slight touch closing the tuning capacitor is louder.

[Mike Watterson]

The 1L6, DK92, DK96 etc are all about 7.5 pF at the anode. The 1j24b might be only 3 or 4 pF, though exactly how it's measured in each case isn't revealed.

The capacitor inside the IFT can might be 65pF

I think an experiment with a small variable cap might be in order!