27 Oct
2020
27 Oct
'20
1:28 a.m.
It works!
Long story short: the LimeSDR-Mini was off by 3.5kHz, and not the phone.
Even after 30 years, the phone is only off by 14Hz, what a performance.
Made in West Germany :-))
After the recalibration of the LimeSDR Mini, the phone successfully updated
the TA and went to normal idle state. Calls are also possible, SMS will be
tested when a capable phone arrives. All TA combinations were tested
(1-15).
Will create an updated patch for the country specific settings, now that it
is tested.
The lesson is: trust but verify :-) Especially with cheap hardware.
Regards,
Csaba
Tomcsanyi, Domonkos <domi(a)tomcsanyi.net> ezt írta (időpont: 2020. okt. 26.,
H, 19:54):
Hello Csaba,
I am not sure if you know this, wanted to mention it: you could easily
measure your SDR’s oscillator accuracy using the kalibrate tool. It uses
nearby GSM BTSs if I am correct to give you a good estimation about how
much off your oscillator is.
I used to use it with rtl-sdr a lot, because those USB dongles are crazy
off in terms off accuracy.
Once you have an offset figured out for the current temperature I guess
you could try to offset the frequency set by osmocom by that amount. Maybe
even try to expose it as a variable if needed.
Other solution would be using some kind of a GPSDO or similar 10 MHz input
if the LineSDR supports it.
Sorry again if this is known info, but wanted to chime in with some
hopefully helpful bits.
Cheers,
Domi
2020. okt. 26. dátummal, 16:08 időpontban Sipos Csaba <
dchardware(a)gmail.com> írta:
Hi Ralph,
I attached the patch for the country specific settings for Hungary, you
can apply it against "master" . TAs are not yet verified, but the phone
goes "green" with all TAs from 1 to 15. The RF part and country code are
verified.
One thing, maybe your MCR has drifted, due to
ageing?!
Well, that was my first impression too, as if you look at my picture (
http://www.imagebam.com/image/d17e881357285965 ) you will see that the
center of the downlink and uplink transmissions are not exactly 10MHz
apart, they have a couple kHz drift. On the picture the two markers are
indicating the nominal frequency of the channel. As you can see the uplink
burst is actually closer to the nominal frequency, compared to the
downlink. It is not in the standard (or I was not able to find it, in
general the RF spec is not very detailed...), so I am not sure if the SDRs
downlink signal is off, and the phone is correct or the other way around.
Although it sure looks like a low layer fault, as the NMT network is not
even detecting is as bad or incorrect frames.
I did a loopback test with a jumper between the TX and RX of the SDR and
that seemed to work (frame level and frame quality were 90+%). On the other
hand if the local oscillator would be off, the phone would not be able to
lock on the DL signal as well. I have no high precision counter at home,
but with a relatively primitive multimeter I was able to measure 12.7999MHz
on the 12.8MHz local oscillator. And both identical phones behave the same.
The question is if I can somehow offset the uplink frequency on the SDR
side or via config/source modification to remove this couple kHz drift?
If someone has seen a lot of these signals and can take a look at the
image above, would be lovely to hear if any of the signals are off.
I will also get another more "modern" phone hopefully soon, so I can try
with that.
Regards,
Csaba
Ralph A. Schmid, dk5ras <ralph(a)schmid.xxx> ezt írta (időpont: 2020. okt.
26., H, 15:33):
Hi Chaba,
I have a Nokia handheld phone from Hungary, I can try with your settings
and let you know.
Andreas seems to be difficult to reach these days, I found the same :/
With best regards
Ralph.
*From:* OpenBSC [mailto:openbsc-bounces@lists.osmocom.org] *On Behalf Of
*Sipos Csaba
*Sent:* Monday, October 26, 2020 3:16 PM
*To:* openbsc-request(a)lists.osmocom.org <openbsc(a)lists.osmocom.org>
*Subject:* Fwd: NMT-450 - partial success
Hi,
I know it is a bit off topic, but as osmocom-analog has no dedicated mail
list and my every attempt to contact Andreas lead to silence, I thought
this is the closest one to discuss it.
I try to create an NMT-450 network with Motorola MCR4800XL phones, and a
LimeSDR-mini. As the phones are locked to "Hungary" using a specific raster
and a large gap in the middle, first I needed to dig out the details, find
out the country code and create a patch so at least the phone is willing to
lock onto the DS signal. I managed to do all that, so now the phone is
actually able to decode the network and lock onto the signal.
My issue is with the uplink: when the phone tries Traffic Area update
(the phone's uplink transmission burts is clearly seen with a spectrum
analyzer), the network side is not able to detect the uplink burst at all.
Not even as bad, or incorrectly formatted frame. Andreas has a site which
describes how to set up the uplink side and do some tests:
http://osmocom-analog.eversberg.eu/docs/sdr.html
I followed that guide and when the uplink burst from the phone arrives,
the RX IQ constellation monitor indicates a correct burst with proper power
(the burst is nicely round and in the green area). If I try to set up a
call to the phone using the correct country code and phone number, the
phone clearly responds to the paging request, as the 3 paging attempt
generates 3 uplink bursts. Again, with no reception/decoding on the network
side. Tried with two phones of the same type, the effect is the same.
I have two questions:
1. Where to send patches for the osmocom-analog project?
2. Does anyone have an idea what can be wrong with my setup?
One more thing I noticed: compared to the channel frequency used to set
the NMT network up, the uplink is a couple kHz shifted:
http://www.imagebam.com/image/d17e881357285965
As it can be seen, the uplink burst appears 3-4kHz left relative to the
downlink signal.
The network is started with the following command:
nmt -k 239 -k 235 -Y HU,1 --limesdr-mini --sdr-rx-gain 20
Any and all help is appreciated.
Regards,
Csaba
<0001-Add-country-specific-settings-for-Hungary.patch>