Re: [Openbts-discuss] OpenBTS 5.0 + USRP B200

[Gerhard <ope...@gk...>]

Hi Tom, Hi all,

sorry for not answering earlier. I've just got some spare time and was
able to test your suggestions.

It worked! By using the OsmoTRX instead of the OpenBTS transceiver my
phone was able to register on the network! Thanks for this suggestion"
Now there is a follow-up question in my mind: Why did it work? Is
OsmoTRX more sensible when it comes to detecting and handling RACH bursts?

Furthermore, I would really like to use the "original" transceiver,
because my goal was to do some GPRS. Citing from the OsmoTRX page [1]:
"OsmoTRX is fully compatible with OpenBTS for voice and SMS services.
Due to differences in handing of GPRS, OsmoTRX does not support GPRS
when used with OpenBTS, however, GPRS with the Osmocom stack is supported."

Any hints on what I can do to get my setup to work with the OpenBTS
transceiver? Or is the only solution the hard way (enable DEBUG logging)? ;)


As always: any help is welcome.

Cheers
  Gerhard


[1] http://openbsc.osmocom.org/trac/wiki/OsmoTRX#OsmoTRXwithOpenBTS



On Sep 11, 2015 at 00:33 AM Tom Tsou wrote:
> On Thu, Sep 10, 2015 at 8:02 PM, Gerhard <ope...@gk...> wrote:
>> I'm struggling to get OpenBTS 5.0 with an USRP B200 to work. After I
>> have build, installed and run OpenBTS, all I get is the network showing
>> up on my Android device. When I try to connect to it, it fails (even
>> with open registration enabled). There are no hints in the logfiles. The
>> "tmsis"-table stays empty.
> 
> Likely case is RACH bursts being missed.
> 
> Can you give OsmoTRX a try as a replacement for OpenBTS transceiver?
> Note the section on compatibility with OpenBTS.
> 
> http://openbsc.osmocom.org/trac/wiki/OsmoTRX
> 
>> Another mail said that the internal clock of the B200 is way to bad for
>> using it with OpenBTS.[2] Is that so? How can I find out, if the clock
>> is causing the issues?
> 
> Typically, but not always, If the handset is able to detect the
> network, then clocking is sufficient for test calls other basic
> services.
> 
>> The OpenBTS book and several other sources state to set "rxgain" to 0-10
>> for USRP devices. However, Matt Ettus said (regarding the gain of a
>> B200) via the USRP-users mailing list: "0dB of gain is unusably low. You
>> should be using at least 40dB unless there is a very strong signal."
>> What value should I use?
> 
> The point of confusion is that gain ranges vary across different USRP
> devices. For example, gain range on the RFIC based B200 is roughly
> 0-75 dB while the range on X300 with UBX daughterboard is 0-30 dB. The
> OpenBTS book assumes a range closer to the latter, which ends up in
> the low to middle part of the range.
> 
> 40 dB on RFIC is close to mid-level, which is a reasonable starting
> point. With GSM handsets we are dealing high transmit power levels, so
> lowering the Rx gain at close range makes sense. Setting gain down to
> 0 dB is excessive and typically reserved for directly coupled use
> cases (i.e. not OTA).
> 
>> My setup is
>> * Ubuntu 12.04.05 LTS Server (32 bit)
>> * OpenBTS 5.0 ($ ./switchto.sh 5.0)
>> * USRP B200 (no external clock)
>> * UHD 003.009.000
>> * No Antennas attached (Phone is next to the USRP)
> 
> Nothing abnormal. Attaching antenna(s) with lowered gain settings is
> recommended over raw SMA connectors for consistency. Also check
> logging for anything unusual; INFO logging level is fine.
> 
>   -TT
>