Providing your own LO

[cochenob]

I have some applications that require phase coherency, and have been reading about some of the MIMO applications using the USRP which have similar requirements.  In all instances, one is required to make a bunch of calibration/reference measurements since every time the radio is reconfigured (frequency changes, BW changes, etc etc), the phase relationship between the LOs chage.  In other words...even though USRPs may be phase locked, there's a constant phase offset that will always change unpredictably with each reconfigure unless you measure/calibrate it.

I've read both here and over on the Ettus board about some tricks that allow you to do this calibration (one example in the Developers Zone uses the coupling of the switch between channels to allow you to make a reference measurement without a reconfigure of the radio...cute!).  However...I can't help but to think that all of these extra calibration efforts would immediately disappear if one could just provide their own external LO at the desired frequency.  That way not only would the LO's between radios be phase locked...but their phase relationship would never change either.  In terms of experimental procedures...this is *huge*...esecially depending upon the complexity of the measurement.

I know this comes down to time/money/complexity for Ettus and NI, and the eventual customer.  But I'd totally be willing to accept that if it made life easier in implementing the measurement.  Any plans for this in the future?  Is it possible now?  Would I have to void the warranty to try it?  🙂

---

Brandon

[Joshua-B]

Hi Brandon,

I hope you are doing well today!  Unfortunately this cannot be done with the USRP.  However, A number of our PXI form factor RF modules can use an external LO.  I'm provided a link below to a white paper that talks about doing phase coherent generation and acquisition with some of these modules.  

Configuring Phase-Coherent RF Measurement Systems: From MIMO to Beamforming

http://forums.ni.com/t5/forums/replypage/board-id/500/message-id/1130

If this is something you would like to see in future models of the USRP, I recommend posting it the Idea Exchange on the NI Community at:

http://www.ni.com/community/

Thanks!

Regards,

Josh Brehm

Applications Engineer

National Instruments

[cochenob]

Thanks.  Hoping to keep the advantage of the small form factor with the USRP though.  The PXI chassis is good for benchtop experiments, but wieldy for field measurements or sensor integration.  I'll post something over on the other board.  Thanks!

[ErikL]

For RX phase coherence, it can be done as a post-processing step.  Demonstrated in this code and video:

Code: https://decibel.ni.com/content/docs/DOC-25716

Video: http://youtu.be/qBvpllCqDR0

As for TX beamforming, its more challenging.  We recommend the FlexRIO SDR platform.

http://www.ni.com/sdr

[cochenob]

Thanks Erik-

This is precisely the example we've been looking at.  It's a clever procedure, but one that seems could be elimianted if you could just provide your own LO.

I've looked at the PXI SDR in the past.  We really like the flexibility/portability/cost of the USRP though.  This is the part where you tell me, "Yes, that's the point." : ) .

[ErikL]

how many channels and what frequencies are you looking for?

Erik

[cochenob]

Ideally...1-4 channels from DC to a few hundred MHz I suppose?  Are you thinking there's possibly another hardware option?  Again...the FlexRIO stuff is great...but the PXI form factor is a bit of and end game for life 'beyond the test bench'.  The USRP is perfectly acceptable in terms of channels, frequency band, real-time BW, actual front end performance, size, cost, ease to program, etc.......but for coherent applications, this re-tuning issue becomes a bit of an inconvenience to deal with that could be eliminated by providing your own LO.