Do you get a different phase every time you change the carrier?

[illigdw]

Hello,

I'm working on an application which uses a very wide bandwidth (50 MHz - 1 GHz) and am encountering some behaviors that seem to suggest that my phase is changing every time I change the carrier. Since the USRP's instantaneous bandwidth is much smaller than what I need, I iterate through different carrier frequencies to cover my full bandwidth.

I was running some simple tests where I just have the TX looped directly into the TX and found that my phase data didn't seem to make much sense, and basically looked like it was random instead of fitting the pattern I expected.

As a sanity check, I also ran some repeated trials using a single carrier, where I would transmit my packets and then start up my VI again. Within one trial, everything seems to make sense within some constant offset. But when I move to the next trial, my phase changes by what seems to be a random offset compared to the previous trial.

I did see on a previous thread (http://forums.ni.com/t5/USRP-Software-Radio/usrp-reverse-signal/m-p/2061118/highlight/true#M291) a comment indicating that the LO clocks lock at different phase shifts each time an application runs... so does changing the carrier frequency count as "running" a new application?

Thanks for any insight into this issue

-David

[ErikL]

Every time you press run, the carrier phase changes. What is the application. There are usually ways to compensate for this.

[illigdw]

I am trying to generate a stepped frequency modulation for a radar application. So if the phase changes by an unknown value when I switch between frequencies, it creates some big problems for me.

Based on the other thread I saw, I expected the carrier phase to change every time I told the program to run, so I tried to compensate for this by creating a loop structure with as many USRP configuration blocks as possible placed outside the loop. That way I only "press run" one time outside the loop, and then go off and do my frequency sweeps within my loop. However, it still seems like something gets reconfigured between loop iterations because my phase is radically different between sweeps.

Is there some trick I'm missing here?

-David

[cochenob]

Hi Erik-

Do you have any follow-up suggestions here? 

It seems like the only way around this is to measure the phase difference "locally", say with a switch right after the Tx, and then measure the phase of the DUT.  Then compare the two.

I guess this presumes that there are separate Tx and Rx oscillators...and that while they're phase locked...the phase difference between them will change every time your reconfigure the radio. 

Related question...why have two oscillators in the first place?  Why not just share the same one between the two?  What extra flexibility are you getting here that you couldn't do at baseband?  Is there any way to share references on the board level on our own without significantly "voiding the warranty"?

---

Brandon

[USBarnes]

Hello all,

Could you please provide the USRP model you are using? To my knowledge I am not aware of known changes on the board level that can be made to correct for this.