Hi folks,

I'm in the process of designing the modem for a university-developed cube satellite. We'd like amateurs to be able to use the packetized data channel for interaction with the satellite. The plan is for amateurs to be able to do stuff like initiate file downlink (there's a small camera on board), write text files for use as a message board, query the telemetry files, and possibly route packets from ground -> satellite -> ground.

Right now I'm trying to gauge the capabilities of amateur ground stations. Obviously, the ground stations will be attempting to correct for Doppler shift, but there's only so far that can go. Does anyone have a good idea what the typical velocity error is between a NORAD TLE and a cube satellite TLE? This will go into my maximum frequency error specification.

The other issue, frequency wise, is oscillator drift and offset error. The satellite will be using a very stable Stratum-3 TCXO with 280 ppb or better stability, which gives me good accuracy on the satellite side (within 100 Hz or so on 70cm). The other side of the link would be the amateur transceivers. What's a reasonable oscillator stability spec for a typical SSB transceiver used on amateur satellite frequencies?

For modulation scheme I'm leaning towards either pi/4 DQPSK or standard QPSK with a BPSK barker code preamble on each packet. Both schemes will use convolutional coding for FEC. Which scheme I go with depends on which ends up being more robust under severe doppler while still fitting into hardware constraints - the comms system is all going to be done in SDR. Suggestions for alternate schemes are welcome though.

Thanks for your help,