k9qi
Member
Offline
Posts: 12
|
|
« Reply #28 on: November 30, 2014, 01:40:29 PM » |
|
Hell all,
I am currently out of town with family for the holiday but will link some schematics, etc when I get back. The Johnson counter approach is only applicable to the Sainton screen or suppressor modulated Doherty; what I call Type 2. A classic Doherty requires the use of a phase shifter (90 degrees) between the carrier and peak tube as the low power linear envelope must remain intact. And, the peak tube must have exactly 2X the AC (signal) grid voltage as the carrier tube so that the conduction points, with bias, are nearly identical at 100% positive modulation. The networks are not difficult; especially if you use a tetrode and run the tube Class AB1 so no grid current is drawn from either tube at any time. 4CX250's would be great for this application but anything from a 4-65A through a 4-1000A will work fine as will smaller versions using 6CL6's, 12BY7's, 2E26's or 6146's. One could even get cute with the design and use an 829B dual tetrode; they won't work for a Type 2 since the screens are tied together, but they work fine for a classic Doherty. Type 1's are less efficient but Type 2's are more work as you must also build the audio section; really a class A cathode follower modulator complete with some form of 1:1 mod transformer albeit at low power compared to plate modulation.
On the PDM side, I built a small PDM using series 6CL6's; it made about 5 watts. I then built a bigger one using tetrodes (4-65A's) basically copying the Harris MW-5 but without feedback. The performance was mediocre; maybe 5 - 6% THD. The problem running it open loop was no modulation tracking or power output reference for stability. Basically, it was OK except that low frequency power supply regulation (bounce, if you will) was not tracked out so a 4:1 IM measurement (60Hz/5kHz) was on the order of 10- 12% at 95% modulation. CCIF 1:1 IM (4/5kHz) was much better at around 5 - 6% if I recall. That rig has been sitting for a number of years; I acquired a Continental 314R1/828D PDM rig that I converted to 160M. I now plan t convert it to 75M; John Lyles, K5PRO has the same rig and is contemplating the same thing with his. The 314 has a -8.5kV supply, the anode of the PA is grounded and the series PDM runs the filament at -8.5kV. I loved that rig in broadcast work but disliked that high voltage in the shack. So, I replaced the power supply transformer with one that produces about 3.7kV and run the rig at 250W. I also run only 1 3-500Z in the PA and use the designed single 3-500Z in the switcher PDM. I gave up a bit of positive mod capability; it would have been better to have a 4.5kV supply but it works fine. The big trick with all the PDM boxes, homebrew or otherwise, is matching the lowpass filter of the series PDM to the PA impedance. The design of that filter is a bit of work - a 5th order transitional that is essentially linear group delay to about 20kHz and then has zeros at 70kHz to eliminate the PDM frequency. The 3dB point is about 40kHz. It is far easier to do a PDM using switchmode FETs and use either an H-bridge or Class E PA. Voltages required are much lower by at least 10x over a tube, the filter impedances are also much, much lower and more forgiving of mismatch as a result. I love using tubes but if I ever did another PDM, I'd do it with FETs. Plus, the Class E rigs do sound extremely good on the air.
Recently, I've been thinking it might be good to experiment with some of the Freescale linear FETs in a solid state "Type 1" Doherty. But that will wait as there are too many projects already started and I do want to finish the Sainton 4-1000A rig - if for no other reason, as a tribute to Joe Sainton who mentored me some 30+ years ago.
Again, I'll post links to much of this when back home. I'll also make images of the pertinent sections of various manuals that discuss the theory and practice of these amps.
73, greg - k9qi
|