Hi All - this has turned into a good PWM discussion ! Lots of good comments.
I'll relate some of my experience with high power tube PWM transmittes. First, you can put the modulator anywhere - in the next room if you want. However, as Frank has pointed out, there are a few critical leads which must be kept short.
1) The damper diode is connected directly to the plate of the modulator tube at one end. The other end must go directly to a capacitor to the local ground of the modulator. You need to also provide a good, low impedance local capacitor between the high voltage and the ground of the modulator, at the modulator. This local capacitor and the local ground (chassis) of the modulator are the high current points involved in the modulator and damper diode. You will generate serious ringing without these.
2) You *must* make a provision for analog compensation, or you will have distortion at high negative peaks. This means you need to provide analog control of your modulator tube. Either you need to vary the amplitude of the input pulse train with modulation (the drive follows the modulation - higher drive for higher modulator output), and/or vary the screen voltage of the modulator tube. I have found varying the drive is very important and effective. The only alternative available is to provide an active pull down (up, in this case) to actively discharge the filter network. Doing so will compromise the modulator efficiency, but the method does work. The broadcasters all used analog compensation, and so did I in my rig. The other problem you will solve with analog compensation is the fact that a tube is an imperfect switch and will exhibit an increasing voltage drop as more current is drawn through the tube. The analog compensation will linearize the tube at high currents, and will also allow you to achieve 100% negative modulation with much lower distortion. There is extensive documentation about this on my web site in the "Vacuum Tube PWM Article".
3) Important: Beware of coupling around the filter - through the ground. I have found it important to isolate either the RF amplifier ground (impractical in this case) or the power supply / PWM filter ground. You might think everything is "grounded", and your large rack and all that aluminum will surely never allow any ground currents, but you very likely will have some, and this will reduce the effectiveness of your filter. In my PWM rig, I actually DC isolated the modulator and PWM filter ground plane (and it was a plane).
One suggestion is this: The power supply ground connectes to the modulator (isolated) and PWM filter (isolated) ground point. This is where everything related to the modulator joins: The bypass for the damper, the modulator local bypass capacitor and the first PWM filter capacitor. Then you go through the filter. Since the filter is large, there should be a ground plane provided for the filter (isolated). The last capacitor in the filter is located physically inside of the RF amplifier cabinet, and is usually the RF bypass as well. A ground strap is run from the RF amplifier chassis (which is presumably the actual cabinet/rack ground as well), to the filter ground plane. The high voltage, of course, runs from the filter output as well. In doing this, you will reduce considerably the chances of coupling around the filter.
When running high power, an 80dB down spur from your pulse width modulator will be very apparent.
A series resonator at the output of the filter (preferably located very near to the RF amplifier and grounded to it) can be of help, although your 6 element filter should be sufficient. I have not needed to provide any other circuitry when using a 6 element filter with the corner at 12.5kHz and a 100kHz switching frequency. The PWM spurs are unmeasurable by other operators who are a short distance from my residence.
Anyway, just a few random thoughts !!!
Regards,
Steve
