My First Homebrew; Noob Questions.

[KD1SH]

   Currently pine-boarding my first AM transmitter, and all is going well other than the expected teething pains.
   The quick rundown:  it's an analog series mod transmitter using an 809 triode PA driven by a 2E26.  The oscillator is a series tuned Clapp in its ECO iteration, employing a 6AG7, followed by a 6SN7 dual triode, the first section of which is a cathode follower and the second section a conventional class A buffer.
   My only regret so far has been the choice of the 6AG7 in the oscillator.  It runs wonderfully, but I think it's a bit too much tube - maybe something like a 6AU6 would have been better.  With a parallel LC tank on the output, I get an output of 400V peak-to-peak with only 250V on the plate!  I backed the plate voltage down to 125V and the screen to 75V, and now the output is down to a more comfortable 130V peak-to-peak, where a simple voltage divider can tailor it as needed.
   It's up and running on the pine-board from the oscillator to the buffer, and now I'm at the point of figuring bias and drive levels for the driver and PA.
   Biasing the 809 will be pretty straightforward, using tube data from old "East Coast" handbooks, but but I've got a few questions about biasing the driver.  Up until now my low power stages have all been linear - outputs identical to inputs on the scope.  Now, I figure that biasing the 2E26 driver for class A linearity would be pointless, since the 809 will be biased so far beyond cutoff that most of the input waveform will do nothing whatsoever.  The plate dissipation of the 2E26 probably wouldn't allow for class A operation and still generate enough output to feed a hungry triode, anyway.
   So, assuming my 2E26 will be biased for class C, just how deeply into class C should I push it?  My understanding is that running the final PA well into class C, resulting in a very short conduction angle, is best for plate modulated AM, since the tube will more faithfully follow square-law.  The driver, though - my 2E26 - doesn't need to give a fig about square law, giving me a good deal of fudge room when choosing the bias.  The handbook gives a bias of -45V for the 2E26, assuming it's being used as a CW only final PA.  My question is, will this be appropriate for a driver?
   

[DMOD]

I would bias the 2E26 driver in Class C and let the parallel resonant circuit create the sine wave.

The driver circuit below should provide about 3.5X the power needed to drive a single 809, hence the variable 25k potentiometer in the screen grid circuit.

I have no idea what the Xcoast circuit shows but here is a suggested schematic for starters:


Phil - AC0OB

Addendum: Driver Grid Metering as you will need it. The metering uses inexpensive Analog Meters with 2.5 V scales (Commercial Electric M1015B). Metering Jacks are Antique Electronic Supply, S-H260X).

[KD1SH]

   Very nice - thanks, that's a big help.  My original thought was to use a fixed, regulated bias supply, but your grid leak bias is simpler. 
   Somewhere around 40 to 60 volts p-p was pretty much what I was figuring for grid drive on the 2E26.  For the 809, I'm thinking it'll take at least 100 volts p-p, so having a surplus of driving power available - and adjustable via the driver screen - will work well.
   I was also contemplating using pi-section coupling between the 2E26 and the 809, to better match the impedance and ensure best efficiency of power transfer, but with a surplus of driver output, would I be gaining anything?  Maybe better harmonic attenuation?
   Metering, definitely, especially grid current.

   Haven't even started on the series mod part yet, but I'm figuring a pair of 807's, sourcing on the B+ side.  Of course, that means floating the entire modulator, or at least the 807's and their DC biasing circuitry.  A common DC ground won't be possible, but I can establish a common signal ground with appropriately rated capacitors.

   The plan is to have this thing up and running for the next AM Rally, but that may prove a bit ambitious.  I'm not going to start cutting holes in the chassis until it all works on the pine-board.

I would bias the 2E26 driver in Class C and let the parallel resonant circuit create the sine wave.

The driver circuit below should provide about 3.5X the power needed to drive a single 809, hence the variable 25k potentiometer in the screen grid circuit.

I have no idea what the Xcoast circuit shows but here is a suggested schematic for starters:


Phil - AC0OB

Addendum: Driver Grid Metering as you will need it. The metering uses inexpensive Analog Meters with 2.5 V scales (Commercial Electric M1015B). Metering Jacks are Antique Electronic Supply, S-H260X).

[DMOD]

 
...Somewhere around 40 to 60 volts p-p was pretty much what I was figuring for grid drive on the 2E26.

For the 809, I'm thinking it'll take at least 100 volts p-p, so having a surplus of driving power available - and adjustable via the driver screen - will work well...

I was also contemplating using pi-section coupling between the 2E26 and the 809, to better match the impedance and ensure best efficiency of power transfer, but with a surplus of driver output, would I be gaining anything?  Maybe better harmonic attenuation?
   Metering, definitely, especially grid current.

  

I try to go by the tube curves.

It appears one needs +10 volts on the control grid of the 2E26 to push it into Class C. So with a nominal grid current of 2.5 mA, the DC bias should be about -55 Volts. So 65 volts Peak or 130 V peak-to-peak is needed on the grid of the 2E26 for class C.

Likewise, it appears one needs to get the 809's control grid  to +20V for the stated plate current at 750 volts. So with a nominal grid current of 32 mA, the DC bias should be about -70 Volts. So at least 90 volts Peak or 180 V peak-to-peak is needed on the grid of the 809 to drive it into class C.

As far as designing a Pi-network input for the 809, the output impedance of the 2E26 has to be determined and the input impedance of the 809 has to be determined.

Since the output power will be about 55 Watts, you only need about 30 Watts of Modulation power.


Phil - AC0OB

[KD1SH]

  Yeah, I get my peaks and peak-to-peaks confused sometimes.  The negative half of "peak-to-peak" doesn't get you "over the hump".
   I'm still getting used to working with the data sheets; all too often I resort to experimenting and finding out for myself.  It's fun, and it works, but plotting it out with the curves gets you in the ball-park before you heat up the soldering iron.