John -
It is my understanding that the current FPGA image used for OpenBTS will fail on the transmit side if the FPGA clock is slower than the USB clock. The USB clock is 48 MHz.
Also, the GSM symbol clock is derived from 13 MHz, so if you use a multiple-of-13 clock, you can simplify a lot of the decimation.
These two facts together led us to choose 52 MHz for the current OpenBTS USRP clock.
-- David
On Jan 16, 2010, at 4:33 PM, John Orlando wrote: Hi all, I've seen posts about folks providing an external clock of 44 MHz to the USRP, and with a few software tweaks, getting it to work. I'm wondering if there is a floor as to how low this clock be reduced without causing problems. I know some A/D converters have minimum sample rates, but the AD9862 datasheet doesn't seem to indicate that it does. Is there any reason I couldn't run my USRP with an external reference at 26 MHz? How 'bout 10 MHz? The only datapoint I could find was a reference to the fact that the current openBTS FPGA image doesn't run (?) at 26 MHz due to a "firmware fix" that is needed, though I'm guessing that is referencing the FPGA code: http://www.mail-archive.com/address@hidden/msg21130.html
I'll probably make the USRP hardware modification on Monday to try these out with an external sig gen, but I figured I'd throw the question to the list prior to that to see if I could get a definitive answer. Thanks much...
-- Regards, John Orlando CEO/System Architect Epiq Solutions www.epiq-solutions.com
_______________________________________________ Discuss-gnuradio mailing list
David A. Burgess Kestrel Signal Processing, Inc.
|