At 09:52 23/03/01 -0000, you wrote:
Other tasks got in the way so never perused this much further, but the major
task putting me off was to work out a way of continuously phase modulating a
transmitter over a full 360 degree range. This looked at first sight even
more difficult than amplitude modulating - had already made a PWM modulator,
although not big enough for the Decca transmitters, published in the lastest
LF Experimenters book.
Hello Andy & group,
What I had in mind was a kind of digital 'smooth' phase-shifting.
The 'rude' way to perform BPSK is just to send the RF (square wave) through
an XOR gate, what will cause an instant 180 degrees phase shift.
Let us (for simplicity) assume that we would modulate BPSK on 100kHz square
wave. Further assume that the phase shift occurs exactly on the rising edge
of the 100kHz signal.
If there is no phase shift we have a nice 100kHz square wave (5us high /
5us low). At the moment of a phase shift you will have 1 cycle with 10us
high and 5us low, we have an 'extra' 5us high level (or otherwise seen :
the frequency is down to 66.7kHz for 1 cycle = 15us). A 180 degrees phase
jump (at 100kHz) could be seen as a 5us 'time jump'.
If we want a more smooth phase transition we can spread this 5u 'time jump'
over several cycles, instead of a single 15us cycle we could have 5 cycles
of each 11us (otherwise seen : the frequency would be down to 90.9kHz for
55us).
If this is still too 'bumpy' we could have 50 cycles of 10.1us (frequency
down to 99.01kHz for 505us) or even 500 cycles of 10.01us (frequency down
to 99.9kHz for 5005us).
How smooth you can make the phase transition would depend on 2 things :
1. the transition time must be relatively short compared to the duration of
a bit (I believe that 5 to 10% of the time would be acceptable)
2. the smoother the transistion, the higher the reference frequency you need
To explain this last :
The 'base frequency' (in this example 100kHz) and the 'transition
frequency' (in this case 90.9kHz or 99.01kHz ... depending on how smooth
you want it to be) must be directly derived from the same source. So if you
take a rather fast transition (eg. the 5 cycles of 90.9kHz) you need a
source of 1MHz (divided by 10 = 100kHz, divided by 11 = 90.9kHz). But if
you want a slow transition (eg. 500 cycles of 99.9kHz) you will need a
source of 100MHz (divided by 1000 = 100kHz, divided by 1001 = 99.9kHz).
I believe that the 2. limitation will be more stringent.
Now back to 137.5kHz : A stable signal source (PLL or DDS) to to 40 or
50MHz is rather easy to make. This would give us a practical ratio of 256
(35.2MHz source for a 137.5kHz 'base frequency' and a 2 millisecond
transition time.
I am aware that this method will require substancial hardware, but it will
be at the 'low power' end (reducing the firework) and once you have it, it
can be used with any PA. And maybe there are more intelligent methods to
achieve this smooth phase shifting, I wonder if we cannot put most of it
into a PIC or similar ?
73, Rik ON7YD
|
