Discussion in 'Amplitude Modulation' started by KA9Q, Oct 10, 2019.
"Audio is something that messes up a nice, clean carrier." ~ John Battison
I've been interested in poor-man's frequency references, so I was looking AM and TV broadcast stations to see how stable their carriers are in practice. (The FCC is no help since the legal tolerances are rather loose.)
So I exchanged some email with the chief engineer of a local AM radio station who confirmed my on-air observations. (They've been about 0.9 Hz low since the transmitter was installed in the early 1990s. Very stable). Then I said that quite frankly, I found his carrier to be far more useful than his sidebands. He knew exactly what I meant, and agreed.
So from AC0OB's excellent post, it seems like the solution may be as simple as neutralizing the final amplifier tube. I don't have the schematic to the transmitter (only a block diagram) so I don't know what tube it uses. If I had to guess it would be something like a 4CX15,000 tetrode. Maybe just better bypassing on the screen grid would help, since its function is to reduce grid/plate capacitance.
Another mechanism occurred to me: electron transit time. As the plate voltage increases with modulation, the electrons move faster. This is consistent with what I see in the I/Q plots produced by Kiwisdr, assuming they're not "upside down": as the carrier power increases, the carrier phase advances on all three high level AM transmitters (5, 10, 15 MHz). In a physically large tube like a 4CX15,000, I could see the electron transit times at HF as being significant. Nominal plate voltage is 5.5 kV.
I don't think transit time becomes a factor until you get up into UHF.
Electron transit times are involved, but mostly as transit time loading of the grid circuit.
If the transmitter is properly designed and neutralised, this effect should be negligible as compared to the grid current loading in a Class-C final.
This effect looks very much as dynamic detuning of the tuned circuits in the amplifier.
Even for a tube as physically large as a 4CX15,000 or so?
The "innards" of the 4CX5000A are shown here;
and the higher power 4cx15000A is about 50% larger.
The electron path from filament to anode is then about 50 mm, and the average speed of the accelerated electrons in the order of 0.1 c. It would then take 0.05/30000 = 1.6 ns to travel between the filament and the anode.
Transit time loading occurs when the momentary voltage gets out of phase with the electron stream, which starts to be influential when the phase shift approaches 45 degrees or so.
The frequency where 1.6 ns is 45 degrees phase shift would be about 80 MHz, which is close to the upper frequency where full power operation of the 4CX15000A is still permitted.
Magnitudes of phase shifts would be influenced by the anode and screen accelerating potentials which indeed are modulation-dependent.
On 10 MHz the influence on phase-shift by transit-time effects would be lower, but maybe large enough to be noticed in an I-Q representation.
Thanks, you just saved me the exact calculation I was going to make!
I'm not sure what transit time loading is, but I agree transit time can't be the only factor at work here. The distortion is worst at 10 MHz, not 15, and 5/10/15 all use the same type of transmitter.
My thinking was that the variation in anode voltage with amplitude modulation would vary the speed of the electrons, at least in the space between the screen grid and anode, and this would vary the RF phase shift between the input and output of the amplifier. It's been a very long time since I looked at this stuff, and I don't remember how to model the electric fields and electron flows and accelerations within the tube as a function of relative electrode potentials.