On Fri, 14 Jul 2000, Jamie Morken wrote:
> Is the set of N samples the analog
> samples from the soundcard or is "fsk symbol" something else?
Yes, the N samples are the raw samples from the soundcard that you think
contain an fsk tone. For 1200baud afsk packet sampled at 11050hz, N is
about 9 samples. Every 9 samples the tone may change and you want to use
all 9 samples of the symbol to figure out what the tone was.
> I see the ratio of
> 1.8333333*1200 = 2200 but can you explain how you are checking for power at the
> 1200 Hz frequency?
Let's not worry about the fact that these are samples, just call S a
function of t (time). This signal happens to be a 1200hz tone that has an
arbitrary phase offset (it might be coherent fsk then no offset but that's
abnormal on ham bands as far as I know).
S[t] = A*sin( 1200*t*2*Pi+1.23456789 ) <--- A is the power we want
to know
Now when we do convolution(S[],Ia[]) we really mean integrate this
S[t]*sin(1200*t*2*Pi)
from zero to two Pi. That integral is
A*cos(1.23456789). The convolution(S[],Qa[]) ends up looking like
A*sin(1.23456789). So square both of those guys and add and you get:
convolution(S[],Ia[])^2+convolution(S[],Qa[])^2 = A^2
Tada, you have the power. Actually, you have the amplitude squared. Take
the square root and divide by square root 2 if you really want power.
>I was also wondering about the effect of doppler shift on the
> radio signal. I am putting a radio in a rocket that will be travelling over mach 3
> away from the groundstation so will the 1200/2200 fsk frequencies be shifted down
> in frequency enought that I should be compensating for this?
Well two points, normal packet is 1200baud afsk = audio frequency shift
keying = frequency shifting frequency modulation. The signal your decoding
will be immune to doppler, unless your going fast enough to invoke special
relativity in which case your going to have trouble with the symbol timing
pll and NASA.
Straight FSK and PSK will doppler shift, but unless your planning on
orbit your not going to get it going that fast. If your
worried the shifted frequency will be:
F=441Mhz*(1+(speedofrocket)/speedoflight)
The real problem is tranceiver stability. That's why we use FM for voice
and AFSK for data, they're very immune to cheap digital oscillators.
Multimode 440Mhz rigs are $$$ because it takes work to make a good stable
platform for PSK.
> Also will the UHF
> carrier be shifted a significant amount requiring frequency tuning?
If you use normal packet, no problem. 9600baud fsk, psk or any exotic
modes wil be a problem.
Spend more time on making a transmitter that doesn't care about shaking
and occasional power glitches.
-------===Dustin Moore===---------
