BladeRF Calibration

Off topic threads and discussions

Moderator: robert.ghilduta

Post Reply
xarion
Posts: 1
Joined: Sat Dec 23, 2017 1:08 pm

BladeRF Calibration

Post by xarion » Sat Dec 23, 2017 1:10 pm

Hi Guys

I'm trying to calibrate my x40 as per the wiki on github

~/kalibrate-bladeRF# kal -s GSM900
Actual filter bandwidth = 1500 kHz
rxvga1 = 20 dB
rxvga2 = 30 dB
kal: Scanning for GSM-900 base stations.
GSM-900:
chan: 92 (953.4MHz + 11.503kHz) power: 27093.53
chan: 93 (953.6MHz - 672Hz) power: 25353.03
chan: 94 (953.8MHz - 680Hz) power: 20875.63
chan: 95 (954.0MHz - 667Hz) power: 8527.52
chan: 100 (955.0MHz - 672Hz) power: 9893.46
chan: 101 (955.2MHz - 667Hz) power: 12172.33
chan: 102 (955.4MHz - 667Hz) power: 19865.76
chan: 103 (955.6MHz - 667Hz) power: 25914.98
chan: 104 (955.8MHz - 676Hz) power: 28278.21
chan: 105 (956.0MHz - 667Hz) power: 30534.90
chan: 119 (958.8MHz - 667Hz) power: 36344.74
chan: 120 (959.0MHz - 667Hz) power: 32870.62
chan: 121 (959.2MHz - 667Hz) power: 31070.63
chan: 122 (959.4MHz - 919Hz) power: 28242.02
chan: 123 (959.6MHz - 667Hz) power: 26128.25
chan: 124 (959.8MHz - 672Hz) power: 21426.78

....


~/kalibrate-bladeRF# kal -c 93
Actual filter bandwidth = 1500 kHz
rxvga1 = 20 dB
rxvga2 = 30 dB
kal: Calculating clock frequency offset.
Using GSM-900 channel 93 (953.6MHz)


Found lowest offset of 436736.125000Hz at 953.600000MHz (457.986708 ppm) using DAC trim 0xffc0


This seems a little excessive compared to the what the wiki suggests, is this a workable value to set as a trim in the cli ?

(by the way when i do kal -c 103 it hangs and doesnt seem to calibrate - no dots for progress, anything above ARFCN 95 doesnt seem to progress in the kal -c)

Many thanks!

X

bglod
Posts: 202
Joined: Thu Jun 18, 2015 6:10 pm

Re: BladeRF Calibration

Post by bglod » Wed Jan 03, 2018 9:39 am

I wouldn't use kalibrate, it's really bad and doesn't work as advertised.

Ideally, you'd grab a super accurate frequency counter and tune the VCTCXO against that (this is what we do in the factory). Another option that might be easier/cheaper is to use a 10 MHz reference if you have one, or purchase a GPS disciplined oscillator (e.g. Leo Bodnar) and use our VCTCXO tamer function in bladeRF-cli / libbladeRF. Take a look at this blog post. The tune algorithm is pretty rudimentary, but it does a pretty great job. With the setup explained in the blog post, a QPSK signal transmitted by the bladeRF will get within 5 Hz of 1 GHz (i.e. the VCTCXO will be off by ~0.2 Hz).
Electrical Engineer
Nuand, LLC.

Post Reply