[BC] BE 4M Modulation (from other list)

[Hershberger, Dave]

> >> Harold
> >> Hallikainen writes:
> 
> In the tube case, it seems that if the grid is driven with a 
> square pulse
> train (not the normal sine wave of a class C amplifier biased beyond
> cutoff), it seems that the tube would still be pulsed into saturation
> maintaining efficiency.

The problem is that the plate voltage waveform in a tuned class C amplifier
is basically a sinewave. If you want the plate voltage waveform to "pulse"
then some bad things happen. First, you have to discharge the plate
capacitance (and the capacitance of the plate circuit), so you dissipate
0.5*C*V^2 joules every cycle of RF. This will substantially reduce
efficiency. Second, if the plate voltage is pulsed and the plate current is
also pulsed, then you are generating a lot of power at harmonic frequencies
and that power has to go somewhere (to the load). With a pure sinewave on
the plate of a normal class C amplifier, there isn't significant power being
generated at harmonic frequencies. Yes, there is harmonic current, but since
there is no harmonic voltage, there is no significant power at the
harmonics.


> On generating IQ, since the result of IQ modulation is a 
> signal of varying
> amplitude and phase, it seems much simpler to just vary the 
> phase of the
> drive and vary the overlap for the amplitude. No more parts!
>

It's simple in the way you describe, but the envelope and phase modulation
signals must have a bandwidth of up to 10 times the modulation bandwidth
(depending on signal type). So all of those high order sidebands have to
cancel in a EER PA, and that means precise matching of time delay, group
delay, amplitude response, etc. And that's where things change from simple
to complicated.


> I assume the drive is now being generated using DSP 
> techniques instead of
> a sine wave generator and a comparator, right?

BE has some powerful DSP in the box and they are probably doing at least
some of the arcsine generation in DSP. But there is a temporal quantizing
effect that limits what you can do with purely digital processing. Assume
for a moment that the DSP clock is 100 MHz and we're generating a 1 MHz
carrier. Right away you can see that you can only generate drive signals in
3.6 degree increments (every 10 nanoseconds). This would create a serious RF
PWM noise floor (quantizing) problem. 90 degrees divided by 3.6 degrees is
25, and that is only 4.6 bits of resolution. One way around this problem is
to generate your drive signal in DSP, convert it to analog, do a lowpass
(reconstruction) filter in the analog domain, and then apply it to a
comparator. That will provide essentially infinite resolution of the phase
modulated drive signal, because the reconstructed analog waveform will have
much better than sub-clock accuracy. But it requires some analog (waveform
reconstruction) circuitry.

> 
> On each side of the bridge driving a reactive load, it seems 
> that as long
> as the transistors are in saturation, we should not care what 
> sort of a
> load they are driving. Just pull that side of the bridge high 
> or low and
> let the transformer draw what it will.

At high power this can create problems. In a conventional class D amplifier,
the current waveform is a sinewave and the voltage waveform is a squarewave.
They're in phase. So the amplifier switches when the current is near zero.
In a simple bridge RF PWM system, the voltage has to switch when current is
flowing, because each side of the amplifier sees a highly reactive load.
This puts much more stress on the transistors.

> Anyway, neat idea! I always wanted to play with this on ham radio, but
> never got a chance. Since it'd be possible to IQ modulation, 
> it'd be an
> interesting way to generate SSB.
> 
> Harold

I remember seeing a very early QST article on SSB, that used two combined
high level balanced modulators in the output. It used the phasing method to
generate SSB - at high level. The two balanced modulators were push-pull, so
there were 4 tubes contributing to the output. Neat stuff.

Dave Hershberger 
Principal Engineer 
AXCERA
Tel: 530-272-HDTV (4388)
Fax: 530-272-4505 
dhershberger at axcera.com <mailto:dhershberger at axcera.com> 
AXCERA. Axcessing the new era of digital communications. www.axcera.com 
960 McCourtney Road, Suite C, Grass Valley, CA 95949-7423