Re: [Discuss-gnuradio] USRP1 Clock drift

[Matt Robert]

Hi Mike/Nick,

Thanks for your help, I tried offsetting the transmitter by a couple of kHz and it all came to life (isn't software radio great like that!) The error of 13kHz was only at 940.2MHz, which is definitely within spec. As Mike explained this correlates to an error of about 6kHz @ 435MHz.

Seems that the ppm unit of measurement had caught me out - once I figured out that it is a ratio and not a specific value everything suddenly became clear.

Cheers,
Matt


On 7/01/11 10:12 :13, Michael Ossmann wrote:

On Fri, Jan 07, 2011 at 10:01:18AM +1100, Matt Robert wrote:

I ran Kalibrate against my pair of USRP1's using GSM900 towers
(around 940MHz) and found my first unit to me 5kHz off and the
second was 13kHz off.

Is this within limits for the built in oscillator?

Yes, those are both within 20 parts per million.

Perhaps clocktamer or a secondhand GPSDO is needed for my Tx
applications.

Have you tried adjusting the tuning of your USRP to compensate?
e.g. you could tune to 400.013 MHz instead of 400 MHz.  In theory
you might also have to correct the symbol rate, but it is likely
that the receiver will be much more tolerant of symbol clock error
than carrier frequency error.

mossmann