Channel crossover fix for LimeSDR and XTRX SDR radios - Best fringes yet

[David Lonard]

Hello everyone,

I've finally come up with a solution for a significant problem with SDRs like the LimeSDR and XTRX that both use the LMS7002 RF chip. There is a channel cross over issue that results in common mode noise that can seriously degrade interferometer performance.

The XTRX has a pretty big signal delay between channels of 63 nanoseconds that I was correcting with a coax delay line to synchronize both channels. I figured that instead of syncing the two channels with cable that I would instead digitally delay the signals using the cupy.roll command on my IQ data. Sure enough, this delay in the digital domain suppresses channel cross over, allowing for better fringes. With a 50 MSPS sample rate, rolling the IQ data by 3 samples nearly aligns the signal and a little extra coax delay fixes the last bit of signal offset.

I've posted some fringes of Taurus A with my 210 cm dishes at 4.220 GHz with a 50 MHz bandwidth. The fringes now 'wrap' all the way from -pi to +pi. Also, signal blips from aircraft radar are visible on the left. On the right, satellite beacons and a transponder are visible as RFI. Ultimately, the only really clear signal with a C band LNB is in this part of spectrum.

The attached Python code shows the implementation of the cupy.roll command.

David

[zyrkzys]

As an XTRX user, your example code really helped me! Just wanted to say thanks!

[David Lonard]

Your welcome!

The XTRX project continues to be a headache for me though. I've got a XTRX Octopack that I would like to use for 16 channel interferometry, but the current XTRX API fails to enumerate more than a single XTRX. Furthermore, after upgrading to Ubuntu 22.04, the XTRX freezes after a random period of time (minutes or hours) when running the code provided above. 

[John Yu]

Hi David,

Just to express the same experience with xtrx: I brought 2 and find both of them had IQ imbalance issue (i.e. the two components are not the same scale and phase angle were about 75~80 deg, instead of 90 deg). I and Q scale differed about 100:90. I raised the IQ swap issue (I guess is the channel crossover as you name it) and they did resolve that one. It was not coming as I1Q1, I2Q2, I3 Q3.... instead it was Q1I2, Q2 I3,,,,,, If that was the only issue, I can tune them to be 100:100, and delete the first int16 to make it useable, but phase non-orthogonal is a big problem, one needs Lim microsystem's special tool to cfg the internal sin and cos ref signal phase on the mixer to get it right. I never had time to be able to investigate that tool further (LIM microsystem had a YouTube video describing it, but I was not sure how easy it was to setup the full system). After a few months email with xtrx producer, I gave up, as they seemed to be too busy on applications while did not pay attention to such fundamental issues. I was hoping to get  Octopack for 8 xtrx working in parallel as well, but I was told that the 8 cards can run one by one, not working in parallel (yes, the data rate for 8 parallel I/Q channel are going to be too big for such simple device to handle back in 2018/2019). It was waste of time! Blade RF and B210 are far better. More expensive, but truly worth it. Never had issue with them.

I stop chasing the issues by Oct 2019, but if anyone knew about xtrx improvement, and find it good enough for reasonable IQ collection, please update.

Thanks

John

[David Lonard]

I see that Myriad is re-working the gateware and API for the XTRX.

https://github.com/myriadrf

So far, I can't figure out if any of the recent work is operational and re-flashing my XTRX would probably brick it. Maybe at some point, stable operation will be possible in the far distant future.

David

[Marcello Montisci]

Dear All,

may be this question is inappropriate here. I should have made a better search on internet.

I have a USRP B210 with two rx input.

I there a place where I can download the software usable with Window, which allows me to use the USRP B210 for some interferometer experiments in radioastronomy? I currently have the B210 in operation with SDR Console which is no help for radioastronomy and interferometry.

I am a microwave engineer unfortunately very illetterate on computer languages and coding.

Thanks for any help

Marcello

[David Lonard]

Hello Marcello,

I don't have a USRP B210 myself, but I would look at this web site for some clues on using the B210 for coherent two-channel reception.

https://pysdr.org/content/usrp.html

It will take some programming, but by combining the examples they provide with the code I've included here, you should be able to get this to work on a Windows machine.

David