I wanted to duplicate the original QST version as closely as possible. The 500 uufd Cardwell variables, 14 gauge rubber covered wire and one 45 came from my junque box. Advertising brought me the needed brown beehive insulators, tube sockets and Sangamo micas. A friend of mine contributed the second 45. Close study of photos in the original QST article (and lacking a real breadboard) led me to lay mine out on a 8"x 20" piece of shellacked ash. The 1/4" copper tubing in the tank circuit is not just for looks. A lot of current flows in the tank circuit and it also needs to be as mechanically stable as possible.
My transmitter is shown in the photos. I've used it (tubed with sturdier VT25's) on 40 and 80 meter CW running about 10 watts. It's surprisingly stable and sounds good on the air. It's a blast to operate. It does have a certain amount of personality though. It's microphonic, sensative to hand capacity and hard to get on frequency. One sweep of the dial covers multiple MHz. 40 meters is crammed into a few degrees at the bottom end of the dial and 80 isn't much better. I use a separate (well calibrated) receiver to make sure that I'm in the band. Microphonics were cured by keeping it on a table separate from my operating position. Hand capacity is lived with, but I do have a 6" insulated shaft extension that helps a lot. I power it using a regulated 200-315 volt supply. Once on frequency it is surprisingly stable.
C1 - 500uufd C2 - 500uufd C3 - 250uufd mica R1 - 50K (for type 45 tubes) or 10K (for type 10 or VT25 tubes) R2 - 75 ohm center tapped L1 - 40 mtrs: 8 turns 1/4" copper tubing, 1 5/8" ID, 4 1/2" long 80 mtrs: 12 turns 1/4" copper tubing, 2 3/8" ID, 4 1/2" long L2 - 7 turns: 1/4" copper tubing, 1 5/8" ID, 3" long L3 - 40 mtrs: 64 turns, 22 guage, 7/8" dia, center tapped - 80 mtrs: 72 turns, 32 guage, 1" dia, center tapped
Layout the tubes so that the wiring can be as symetric as possible. Both plate terminals should be on the C1 side and both grid terminals should be towards L3. Keep the underside wiring towards the center of the breadboard and away from L1.
Do the wiring between L1 and C1 using copper tubing. Use stiff, heavy gauge wire or bus bar between C1 and the tube plates.
Type 10s or VT25s may also be used with adjustment to R1 and the plate and filament voltages. Filament voltage for type 10 or VT25 is 7.5VAC. The filament voltage for type 45 is 2.5VAC. With type 10s the plate voltage may be raised as high as 600VDC. I suspect other triodes such as '01s or 2A3s would work with voltage and R1 adjustments but I haven't tried them.
The RF choke is under the "breadboard" and connected to L1 via a brass bolt through the breadboard and an alligator clip.
The six binding posts are ordered (front to back) key, key/B- , filament, filament, key/B-, B+.
Use 1/4" outside diameter (not inside diameter) copper tubing for the coils.
Both L2s must be wound in the same direction.
Six beehive standoffs were used. Two support L1, one is used for each L2 and two are used for the antenna connections.
First of all, remember the voltages involved are lethal.
The grid coil needs to be adjusted so that with no load (ie. no antenna connected) the plate current dips at the low end of the band. This should be with the plate capacitor near maximum. Typical plate current with no load is 25 to 45mA. Connect the antenna and adjust the loading for maximum output and a clean sounding signal. Once tuned up, back off the loading coils to 75% of maximum output. This minimizes the antenna's effect on the stability of your signal.
See http://www.prismnet.com/~nielw/1929Stn/1929Station.htm for video of this transmitter on the air.
Addendum
At the 1989 AWA Convention this transmitter was awarded the Matlack Transmitter Award given for excellence in constructing or restoring transmitting equipment.