[amsat-bb] DopplerPSK

Phil Karn karn at ka9q.net
Wed Jun 19 20:20:18 UTC 2019

On 6/12/19 19:58, aflowers--- via AMSAT-BB wrote:
> The same algorithm in DopplerPSK could be used on a complex signal in the SDR architecture, and it could work equally on TX and RX in the processing chain.   To the accuracy of the orbital elements everyone uses, if everyone corrects their uplink and downlink properly there is no perceived frequency drift by anyone using the satellite.  Just what you need for a GHz-range LEO with a narrow-band transponder, eh? 

That's exactly what I do in the SDR I've been working on. It uses a
Funcube dongle and Raspberry Pi (or any general purpose Linux or OSX
system) and applies an optional frequency rate to the digital local
oscillator. The Funcube samples at 192 kHz, so you have a lot of room.
Right now it can do a quadratic phase function, i.e., it can sweep at a
constant rate in frequency, which is a constant acceleration with
continuous phase and frequency. (I'll have to think about which order
PLL this corresponds to.)

It would be easy to extend this to higher orders, but because you can
update everything very rapidly (it's all in software) I've seen no real
need to do so.

Originally the SDR read the velocity data from a pipe, but more recently
I've changed over IP multicasting for everything. A completely separate
program, which can be on a different computer, multicasts the tracking
data to the LAN and the SDRs (there can be more than one) pick it up and
make the frequency computations based on the current frequency.

It all works very well. I tested it on one of the older Japanese
cubesats that transmits a CW beacon on 70 cm. At first I thought I had a
bug in my program because the CW shifted every time it started. Then I
realized it was just a soft power system in an old satellite; the beacon
remained rock steady in a 200 Hz CW filter from horizon to horizon.
Naturally this requires accurate elements and clock time. It was a
little weird hearing miscellaneous spurs sweep through the passband as
the signal stayed in the middle.

All said, PSK31 still isn't the best choice for a LEO satellite with a
lot of Doppler...


More information about the AMSAT-BB mailing list