On Sun, Jul 22, 2012 at 3:27 PM, Alexander Chemeris
<alexander.chemeris(a)gmail.com> wrote:
On Sun, Jul 22, 2012 at 11:11 PM, Thomas Tsou
<thomastsou(a)gmail.com> wrote:
On Sun, Jul 22, 2012 at 10:12 AM, Andrey
Sviyazov
<andrey.sviyazov(a)fairwaves.ru> wrote:
Hi Thomas.
It is very good new.
Please let us know more information how you wins!
I mean which PLL filter, ICP and other LMS paremeters do you use or its done
by just only IQ balance correction and Laurent C1 pulse and all other by
default?
Register dump and startup sequence are attached.
Uhm, this is a register dump at what time? Originally I asked for the
dump before and after every line in the umtrx_tx_init.sh.
That was for Andrey. That dump was taken during the previous 945 MHz
measurement.
These dumps are for the calibration failure, which are taken at the
following points.
0
./umtrx_lms.py --lms 1 --lms-init
1
./umtrx_lms.py --lms 1 --lms-tx-enable 1
2
./umtrx_lms.py --lms 1 --pll-ref-clock 26e6 --lpf-bandwidth-code
0x0f --lms-auto-calibration
3
LO leakage was
also recalibrated for different frequencies.
How different are calibration values? I've never checked how is it
related to the PLL frequency.
A single value within 900 MHz. At 1900 MHz it's maybe up to 3 or 4. LO
leakage is easy to calibrate though. IQ calibration is rather tedious
- calibration values at 900 MHz are useless at 1900 MHz.
For
comparison, after C1 pulse is added USRP1 is 1.3/3.5 and USRP2 is
1.2/3.0 when measured at 925 MHz.
tip: Would be great if you push this to github just as you send e-mail :)
I'm posting the measurement captures and the Matlab for the pulse
sequence generation. The transceiver code is pretty ugly right now.
There are many stuffing bits and zero padding added to minimize the
start / end error on the E4406. It may not even work with real phones
now.
PS Sylvain is waiting for a howto to run OpenBTS with
UmTRX to test his setup.
I'll hack something together later tonight. Can he receive a tone? If
yes, then everything else is straightforward.
Thomas