[time-nuts] how good an oscillator do you need for a GPS simulator

[Jim Lux]

On 12/16/11 7:46 AM, Azelio Boriani wrote:
> A used Spirent is only 26K to 37K. Interesting: playing back bits from
> RAM... can it be that simple? Obvious: a DAC is required.
>

No.. you don't even need a DAC.  The underlying waveform is a binary 
code that is BPSK modulated.

there are a variety of commercial GPS record/playback units that 
basically are single bit digitizers with no mixer (other than the 
sampler with a well chosen sample rate to put the GPS signal at a good 
place in the passband)

The GPS receiver we're flying for the CoNNeCT project samples the three 
bands with a single bit at about 38.6 MHz, which aliases the L1 of 1575 
down to about 9 MHz: close to fs/4, which is a convenient place.  The 
other bands alias to convenient places as well.

There is some art in picking a good sample rate... you want something 
that aliases to somewhere convenient, and you want to not have the 
Doppler push you past a folding point.   Doppler from GPS satellite 
motion to a stationary platform is on order of 5kHz max.  For a receiver 
LEO satellite (probably a worst case) you'd need to add a platform 
motion of 7 km/s, which is about 10 kHz.  Real worst case would be 
something that's reached escape velocity, which I think is 40-50 kHz 
Doppler.

So don't go picking sample rates that are exact fractions of L1.  You'll 
wind up in alias unwrapping hell.

For generating, though, sample rates that are a multiple of the chiprate 
(1.023 MHz) might be the best strategy. Say you clock the bits out at 
10.23 MHz, and you could use the 154th harmonic of the same oscillator 
as your L1 carrier and all that.