Page 1 of 1

bladeRF micro phase coherency

Posted: Wed Oct 03, 2018 7:03 am
by kazam
Hi,
Is it possible to set up two bladeRF micros such that they are phase coherent? According to AD it should be possible:Analog Devices

I'm interested in 4xRX MI applications primarily so that's why I ask.

Thanks

Re: bladeRF micro phase coherency

Posted: Tue Oct 30, 2018 7:03 pm
by rtucker
G'day,

Maybe! We did plan for this during design, but as far as I know, it has not yet been attempted.

From the looks of things, the FMCOMMS5 uses an ADF5355 to generate LO signals to be fed into RX_EXT_LO / TX_EXT_LO pins on the '9361s. We expose these traces on U74 (the expansion connector) on pins 3 (RX_EXT_LO) and 4 (TX_EXT_LO), so a similar sort of arrangement could be done. We also connect the SYNC_IN to the FPGA.

The FMCOMMS5 wiki section also includes a page on multi-chip synchronization, which details use of the SYNC_IN to synchronize the sampling clocks, as well as a method of using the RX*_C inputs to measure the phase error vs the other chip's TX. (We have unpopulated UFL pads for RX1_C and RX2_C, which are independent of the rest of the RF section). With a common 38.4 MHz clock connected to the CLKINs on both boards and some phase correction logic in the FPGA, it has the potential to work.

I hope this is helpful! Thanks for your question. :)

Re: bladeRF micro phase coherency

Posted: Mon Feb 25, 2019 3:44 pm
by maddog
Hi Rey,

I am trying to implement the same using two bladeRF xA9's in 4x4MIMO.

As far as my understanding goes the multichip sync for AD9361 has two methods one is to measure the phase difference in internal LO's and correct it in FPGA and the other is to use an external LO. I will try to use the extenal LO option and could you elaborate on what changes might need to be done in the FPGA?

For the SYNC_IN can I use the FPGA trigger functionality to control the SYNC_IN pin on two different devices at same time?

Thanks.

Re: bladeRF micro phase coherency

Posted: Wed Nov 11, 2020 4:14 pm
by Keaton
Hello,

kazam & maddog have either of you had any success synchronizing the two boards. Thinking about trying this out myself, did either of you utilize the "Trigger" functionality (https://nuand.com/libbladeRF-doc/v2.2.1 ... r_i_g.html) on your bladeRFs? I've been reading the reference manual for the AD9361, and under the section "Multichip Synchronization" it talks about feeding a reference oscillator signal to the XTLAN pins on both boards, but does not mention using the RX_EXT_LO / TX_EXT_LO pins mentioned in rtucker's post. Any insights would be appreciated!

Re: bladeRF micro phase coherency

Posted: Wed Nov 11, 2020 4:34 pm
by iperper
To follow up on this, I'm currently trying to phase synchronize the TX and RX. The multi-chip sync seems to be more for phase sync TX of one chip with TX of another (and same with RX)?

My issue is that I'm constantly re-tuning, which leads to a new random phase offset whenever I re-tune, and for my application I need to compare the phase of the response before and after.

I'm working with backscatter communcations, so the TX signal goes through the environment channel and is received by RX. Currently, my solution is to use the MIMO abilities to feedback the TX2 signal directly (attenuated) to the RX2 signal. Then I divide by over-the-air response by this signal, which cancels the phase difference because the direct path signal received should be the same except for phase shift. Obviously this is a non-ideal solution, so I was wondering if there are other options.

Additionally, eventually it would be nice to have multi-chip sync as well, but I'm not well-versed with FPGAs. Are there implementations of the phase correction described in what you linked?