OsmocomBB layer1 with FreeCalypso changes
mychaela.falconia at gmail.com
Sun Nov 26 22:42:17 UTC 2017
> Do you have a FreeCalypso board to test? If not, we
> can make one remotely accessible (SSH to Linux box which has it
> connected to UART + USB based power cycling capability of the board
> power suplpy) in the sysmocom lab, if needed.
Please keep in mind that the board I sent you back in April has not
been calibrated: because you claimed it early, it was sent to you
before our FreeCalypso RF calibration software (replacement for
Openmoko's original which according to you has been lost and couldn't
be found) had been developed later in the summer. According to my
notes, that board I sent you did have a properly formatted flash file
system structure in the first 2 MiB of the second (non-bootable) flash
bank, but there are no /gsm/rf/* calibration files in that FFS.
If you have a CMU200 which is itself in good calibration standing and
an N-to-SMA cable whose insertion loss at the GSM frequencies of
interest is precisely known (very important), you should be able to
calibrate your FCDEV3B board yourself with my FreeCalypso tools:
As alternative options, you can ship the board back to me and I can
reship it to you with calibration, or I could send you one of my
second-batch boards (already calibrated), and then you send the old
one back to me once you receive the new one and confirm that it's
good, so you don't have a time window without any board.
In the interest of full disclosure, my own CMU200 currently has
uncertain calibration status, and I've been told that getting it
calibrated at R&S's USA (Maryland) office would cost somewhere around
$1600 USD. Unless some sponsor covers it sooner, I expect to be able
to make that expenditure some time in 2018 and get truly dependable
calibration, but until then there is about 0.5 dB of uncertainty in
the Tx power levels I calibrate on FCDEV3B boards. It should
definitely be within the tolerances allowed by the GSM 05.05 spec, but
I am a perfectionist...
There is a more detailed explanation in the doc/Tx-cal-theory article
in the fc-rfcal-tools source repository linked above.
More information about the baseband-devel