Problem with persistent tx of iq data

Having issues with the site, hardware, source code, or any other issues?

Moderator: robert.ghilduta

Post Reply
castilloe
Posts: 1
Joined: Tue Oct 31, 2017 4:05 pm

Problem with persistent tx of iq data

Post by castilloe » Tue Oct 31, 2017 4:54 pm

Hello,

New to bladeRF here. I am trying to understand the effects of calibration by running baseband loopback tests and am running into a strange issue. I set up the tx/rx streams in bladerf-cli as follows:

tx config file=fm_iq_data.csv format=csv repeat=0 delay=0
rx config file=rx_txlpf_rxlpf.csv format=csv n=1000

The iq data that I provide in fm_iq_data.csv is 100 samples representing 100kHz modulating baseband signal for FM. The 100 samples correspond to one period of a 100kHz sine wave with a samplerate of 10M. I set up the samplerate and bandwidth as follows:

set bandwidth rx 2M
set samplerate rx 10M

set bandwidth tx 2M
set samplerate tx 10M

Once I set it all up, I run

tx start

The first time I do 'rx start', received IQ data seems fine; it looks like what I expect. However, when I try to receive samples a second time, I just get nonsense, really slow moving or DC like IQ data.

It seems like the first capture somehow interrupts the transmitter and tx just stops sending new data out (even though tx config shows that it is in the "Running" state). In order to get good data again, I have to stop and re-start tx. Is this a configuration issue? Am I missing a step, maybe a buffer configuration, that will allow simultaneous tx/rx in bladeRF-cli?

Thanks in advance,

--Erick

User avatar
rtucker
Posts: 77
Joined: Sun Jan 25, 2015 10:38 am

Re: Problem with persistent tx of iq data

Post by rtucker » Wed Nov 01, 2017 12:17 pm

Hmmm... I just gave this a shot with a 1024-sample counting pattern as my TX file, and it seems to work okay here. I'm guessing you also did a "set loopback bb_txlpf_rxlpf" as well?

Could you give it a try with bladeRF's "-v verbose" option? It'll produce voluminous amounts of logging information, but might yield a clue. Also, the output of "print" and "version" might help narrow it down.

Thanks!
Last edited by rtucker on Wed Nov 01, 2017 12:18 pm, edited 1 time in total.
Reason: i hate forum formatting
Rey Tucker (she/her)
Systems Engineer, Nuand LLC
Rochester, NY, USA

#WontBeErased

Post Reply