On Monday, 11 June 2012 03:39:00 UTC+10, flipper wrote:
Okay, so I like to make up cutesy names 
I wanted to do 'something' with a 6ME8 so I tried using it for an AM
transmitter and the first prototype works rather well. Bandwidth is
'too much', less than 1 dB down at 18 KHz, but we'll worry about that
later.
This is a 'dollar days' special, 2 whole bucks worth of tubes, and L2
is a UHF converter coil scramble rewound, so that was 'free'. Power
supply is from the same converter and if the remaining issues get
worked out it'll probably end up in that cabinet as well (but where to
put the air variable?).
The schematic is rough, but should be rather self explanatory, and the
ganged tuning hasn't been fleshed out yet. I'm still using separate
caps while jiggling things around.
Orphaned web page, not yet ready for prime time, has a recording of it
playing through a table radio.
http://flipperhome.dyndns.org/Beamus.htm
I couldn't download Fliper's page above on his transmitter last week but can now, and the circuit looks well. Possibly some improvement could be to used cathode followers to drive each beam delctor plate.
If he has 35mW and say 50Vrms anode signal at antenna output, then
0.035 = 50 x 50 / RL so RL = 71kohms. This is rather high but the tuned tank circuit probably has such a high Z at Fo.
The comments made about coils are of interest, and it should make no difference what type of core is used, because 270uH is easy to achieve with many recipes. But in the best old radios the 455kHz and MW RF coils are pie wound windings using litz wire and large Al cans to keep the Q high, ie, RF resistance low. In an AM receiver I built from scratch in 1999, I made copper cans for two RF input coils and I wound solid wire on 50mm long x 10mm dia pieces of ferrite rod normally used for ferrite rod antennas. The Q became higher than using air cored windings on say 20mm PVC pipes. In my receiver, the Q was so high the BW was less than 15kHz, giving slight cutting of sidebands and thus reducing possible audio BW, so I used TWO such coils, and a 3 gang cap, with 2 gands devoted to the the two RF coild and the other to the oscillator. To overcome the reduction of BW, I used antenna input to a tap on one coil, then a resistor from top of coil to top of next coil and then I tuned the coils so they were "stagger tuned", ie, slightly apart at the low end of the band where the LC Q is higher than at above say 1.2MHz. All that worked well, but compact fluorescent lamps and other junk creates a rectified version of incoming RF and I've had to change to a long rod ferrite antenna with a shielded coil and very short wires to keep out electrostatic portion of RF wave.
In my RF generator which I could use as transmitter, I get the same very nice AM wave on the CRO as Flipper's site shows. I have a 6BX6 for the RF output tube, and the coil is a hand wound thing on ferrite with 3 layers of not very tightly wound windings of solid wire, and with PVC tape used between layers. The self capacitance of the coil must be kept low if you wish to be able to tune the tank to as high as possible, say 1,750kHz.
Many old radios went from 550kHz to maybe 1,550kHz, a 3 fold F increase. Say you have 20pF for coil self C, then if the cap gang = 365pF max, then total max C = 385pF. Say the coil turns or ferrite position adjusted for Fo = 530kHz, then L must be 234uH. Say the C gang min C = 25pF, and you have 20pF self C
then Cmin total = 45pF which gives Fo = 1,551kHz, and not as high as anyone should accept. If C minimum could be reduced to 22.5pF, Fo = 2,194kHz, but you will never see this, and besides, one has to be able to use a trimmer cap to set the top of the F range. So the less stray C or coil self C, the better.
Keep the coil cans as big as possible, and never use iron cans.
Patrick Turner.