Message: 6
Date: Mon, 31 Jan 2011 12:24:08 -0700
From: Eugene Blanchard <eugene.blanchard@sait.ca>
Subject: Re: [Openbts-discuss] 52 MHz clock oscillator - clock output
       spec
To: OpenBTS <openbts-discuss@lists.sourceforge.net>
Message-ID:
       <F0980110198A0D47AA3ED9D26017D3B40DDD7C822D@EX3.ACDM.DS.SAIT.CA>
Content-Type: text/plain; charset=us-ascii

Excellent points. The TCXO that I had built is actually a VCTCXO so it would work by software control. Looking over the USRP docs at

http://gnuradio.org/redmine/wiki/gnuradio/UsrpFAQ

I see that there are eight auxiliary 8 bit D/A output ports available on the USRP. Four of the eight are assigned to RXA/TXA and the other four to RXB/TXB. Are all eight being used? Would it be possible to use one for control of the VCTCXO?

I know little to nothing about the actual programming of the USRP and how to do this but I can make a piggyback board with a voltage control input available from the USRP.  Moving in this direction is way beyond what I was planning as I was looking for a simple TCXO for testing.

The Bitshark USRP RX (BURX) daughterboard has a 26 MHz TCVCXO on board, and uses one of the DACs on the USRP1 to provide the needed warp voltage.  These DACs can be controlled from software land, and there are interfaces in both Python as well as C++ for controlling them.  See:

http://www.epiq-solutions.com/twiki/bin/view/Main/BurxWithOnboardTcxoAndUsrp1

for more details about using these DACs. 

Thus, if a user has their USRP1 modified to take an external reference clock, you can wire this 26 MHz oscillator to drive the entire USRP1, and then control the warping from software-land.  We use this scheme for processing UMTS signals, where the frequency error in the reference clock of a typical USRP1 presents a significant issue if not corrected.

--
John Orlando
CEO/System Architect
Epiq Solutions
www.epiq-solutions.com