The transmitter is mechanically easier, but a bigger challenge electronically. Various inductances have to be calculated for 6 bands; the crystal circuit and the grid resonance circuit.
At first I checked the dimensions of all components and see if they will fit. Because none of the parts are original, it takes some puzzling to fit everything in the available space. The transmitter subchassis is about 100mm wide and that leaves about 140mm for the meter with the switches left and right of the meter. Sounds easy, but the meterbody (the part that goes through the frontpanel) is 57mm. The ceramic 3-position switch also measures about 57mm across its contacts, leaving 26mm for the 6-position meter switch. And that is not enough.
In the original, the switches are mounted partly behind the meter, what means the axis is very near the meter body, see original:
Front view of the transmitter. See how close the switches are to the meter.
And that is my first problem. The knobs are only 25mm long and asymmetrical: the mounting hole is at the end of the knob and not in the middle. I found some nice vintage bakelite knobs for the receiver and the position of the mounting hole was not critical there. But here it is, and I have t ofind a solution to that. I bought 5 vintage knobs at a fleemarket but they are way to large for this purpose:
Nice bakelite knob, large enough to switch on a nuclear plant.
42mm is way to much. Since 25mm knobs will be very hard to obtain, I have to find another solution... First let's position all components on the front panel. Drawing:
Draft drawing of the front panel layout.
Not a very clear drawing, but it gives an idea. The ventilation slots at the upper right corner are still missing on the drawing. In the mean time, I found a grid tuning capacitor that looks very much like the original:
Grid tuning variable capacitor
Looks like the BFO-capacitor, but this one has two mounting holes in stead of a mounting nut around the axis. The thread looked like M3 but I should have known: the diameter of the spindle was not 6mm but 1/4" and of course the mounting hole thread was not metric. Again, my special bolts-and-nuts-stock provided two bolts, ex DB-9 connector, that fitted.
Now I have all the parts for the front panel, I can start working on the subchassis. In the original, the subchassis is mounted directly to the front panel. Probably both were less than 1mm thickness, but with 1.5mm printed circuit board and 1.5mm front panel, that will be too thick. So I used the same method als in the receiver: the subchassis is mounted with 5mm spacers, which leaves room for the mounting nuts and the key socket between chassis and front panel. Here also is a isolated support for the grid tuning capacitor, which is at the 500V level.
Subchassis front. The cut-away on the lower left is for the power supply cable.
After placing the sidewalls and painting the chassis, it looks like this:
Front view of the subchassis
Rear view of the subchassis
Notice that the left valve socket is mounted from above, not from underneath, just like the original. And the socket of the PA valve is mounted on a lowered platform. Why that was necessary is unknown to me. The metal 6L6 has a flange and that now levels with the chassis:
Rear view with 6L6 in place
The two supports on the right will be used for mounting the anode choke. You can see the lowered PA valve socket from underneath:
Bottom view, with some paint left...
I tried to figure out how the grid circuit of the PA tube was configured. Let's have a look at that part of the schematic:
Detail of the PA driving circuit.
That looks rather complicated, but that is because the resonance circuit consists of 6 switched parts. Simplified, it looks as follows:
Simplified driving circuit
That would be pretty easy, except for components L9, C10E and C30. The reason is clear to me: C10E lifts the resonance circuit from ground and C30 routes a bit of the anode-energy back to the input. That is called neutralization and compensates for the grid-anode capacitance, which can be considerable with these kind of valves, causing instability at higher frequencies. See the great explanation at this website. My problem is L9. According to my calculations it should either be less than about 500nH, or greater than 12,5uH. Otherwise it will resonate with the 220pF capacitor somewhere between 3MHz-15,5MHz where the transmitter has to operate.
Next: calculating the coils L7 and L8. I measured the capacitance of the circuit and used that to calculate the required inductances. I used some old grips from Ikea cabinets als a coil former; the diameter was 9.7mm. I cut off some straight parts of 35mm length and used the Mini Ring Core Calculator to determine the number of turns. And that matched pretty well!
Coils for the grid tuning.
Coils in place. See the anode choke mounted outside the chassis at the far left
I figured out a way to mount the two switches left and right of the meter. I used a thick piece of PCB (2.2mm) which is even more rigid than aluminium. That is used to support the switches, and is mounted to the meter terminals:
The meter as support for the switches
At the right side there is 1.5mm left to the side and that is enough. I could have mounted the left switch directly to the frontpanel, but then I'd have had the mounting nut on the frontpanel and that looks ugly.. The transmitter looks as follows at this point:
The first parts in place.
After arrival of the Anode Tuning and Aerial Matching capacitors (bought from Ebay - the only missing holes in the picture above) I finished the oscillator/driving circuit. I put a crystal in the socket and switched on the power supply. And it worked! At least, when the switch is in Fundamental position. I calculated L5 and L6 for the fundamental frequency of the crystals, but that was not correct. Let's have a look at the schematic:
Transmitter schematic. clicik here for a blown-up version
The oscillator circuit appears to be a Tri-tet oscillator. Never heard of it before, but this article tought me everything I needed to know. Which meant I had to re-calculate the coils for a frequency somewhere between fundamental and second harmonic. After re-making the coils, I used some 3.5 and 7MHz crystals to test the frequency doubling in tri-tet mode. And now it worked in all cases. As you can see on the oscilloscope:
Anode voltage when in frequency doubling mode.
The doubling is clearly visible. The picture is taken using a 7020kHz crystal, doubling to the 20m band. There is less energy than in fundamental mode, but that was to be expected. Vertical resolution is 10V/cm so this represents 35Vpp here. When in fundamental that is about 55Vpp. This is without PA valve, but looks good so far.
Now the dimensions of all components are known, all holes can be drilled and the front panel painted:
Painted front panel
After mounting all components, it starts to look like a B2 TX:
Most components in place
Rear view, with EL32 and 6L6
Top view of the transmitter.
To connect the power supply, I made the connectors for both the RX and TX from 5mm pertinax, PCB and banana-plugs:
Power supply connector
And finally the tank coils. I asked a local pottery to make the coil formers 78mm length and 35mm diameter. I even specified the colour and I'm really pleased with the result:
Coil former, bottom view
Using a sample of acoustic shielding as used at highways, I made the base for the tank coils, again using banana-plugs. The result is great:
The tank coils
I tested the transmitter and 15-20W is achievable in fundamental. 20m using a 7MHz crystal provided 10-15W - still OK. Only the finger plates have to be made now...