Re: [Openbts-discuss] OpenBTS clock finally fixed

[Alexander C. <ale...@gm...>]

Hi Joshua,

It's definitely interesting. Even that I don't plan to use FA-SY 1,
I think it should be easy to adopt it to other clock boards.

On Mon, Aug 24, 2009 at 23:03, Joshua Lackey<jl...@th...> wrote:
> I recently purchased an FA-SY 1 from Funkamateur that I'm actually quite
> pleased with.  One nice feature is that it warms up the synthesizer to
> reduce temperature drift.
>
> I used the method you mention below, that of synchronizing to a local
> BTS, to calibrate it.  The factory calibration was roughly 6kHz off of
> my local tower.  (~7ppm.)  After calibration, I am normally < 30Hz off
> the tower.  (.03ppm.)  I say normally because last night I was +/- 3Hz
> off the tower frequency!
>
> I wrote an implementation of a rather good paper ("Robust Frequency Burst
> Detection Algorithm for GSM/GPRS," G. Narendra Varma, Usha Sahu, G.
> Prabhu Charan) which uses an adaptive filter to identify the FCH burst.
> Once I have the burst, I calculate the offset.  My current
> implementation actually averages 80 offsets to get the results I give
> above.  (I take 100 offsets and throw away the highest and lowest 10.)
>
> I'd be glad to post my calibration program if there is any interest.
>
>
> Quoting Harald Welte (la...@gn...):
>> Hi David,
>>
>> On Tue, Aug 18, 2009 at 08:25:34PM -0700, David Burgess wrote:
>>
>> > First, we tried to fix this problem by putting a 26 MHz TCXO on the
>> > USRP.  The problem with that is that the USRP firmware fails for
>> > clock rates below 48 MHz.  So we looked for a 52 MHz TCXO bout could
>> > not find any available in small lots with reasonable lead times.  So
>> > then we got someone to build us a clock module PCB with a 26 MHz TCXO
>> > and a clock doubler.
>> >
>> > We got our first sample PCB a few days ago.  The doubler adds a
>> > little jitter to the clock, but not enough to break GSM.
>>
>> I'm quite surprised to read this.  What accuracy is the TCXO specified
>> for?  GSM spec says you have to be within 0.03 ppm accuracy, and I have
>> never seen any TCXO at that accuracy.  Only OCXO, which are obviously very
>> expensive.
>>
>> And even all the OCXO based GSM BTS clocks that I've seen can and need to be
>> calibrated.
>>
>> The question is also: over which temperature range and which aging do you
>> want to be that precise? Some (not even all) OCXO's are specified to have
>> this precsion over 10 years.
>>
>> I have a device with internal OCXO that can measure the difference in ppb
>> between its internal OCXO oscillator and the clock of all the BTS's in its
>> vicinity.  I have never seen any of those BTS exceed 27 ppb difference from
>> that reference - mostly actualy within 10ppb difference.
>>
>> All those measurements where done in the GSM900 band in Germany. I have not yet
>> tried in GSM1800.
>>
>> Have you ever thought of calibrating the OpenBTS clock against the carrier
>> clock derived from other BTS?  I think that would provide a very high accuracy
>> when you need it (i.e. when phones are receiving other GSM networks).  Once you
>> don't receive any other BTS, you just use the latest PLL calibration value.  It
>> is no longer too critical, since you can safely assume the phones also don't
>> receive the carrier of other BTS and thus have to search over a wider range to
>> get a carrier lock.
>>
>> Regards,
>> --
>> - Harald Welte <la...@gn...>           http://laforge.gnumonks.org/
>> ============================================================================
>> "Privacy in residential applications is a desirable marketing option."
>>                                                   (ETSI EN 300 175-7 Ch. A6)
>>
>> ------------------------------------------------------------------------------
>> Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
>> trial. Simplify your report design, integration and deployment - and focus on
>> what you do best, core application coding. Discover what's new with
>> Crystal Reports now.  http://p.sf.net/sfu/bobj-july
>> _______________________________________________
>> Openbts-discuss mailing list
>> Ope...@li...
>> https://lists.sourceforge.net/lists/listinfo/openbts-discuss
>
> ------------------------------------------------------------------------------
> Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
> trial. Simplify your report design, integration and deployment - and focus on
> what you do best, core application coding. Discover what's new with
> Crystal Reports now.  http://p.sf.net/sfu/bobj-july
> _______________________________________________
> Openbts-discuss mailing list
> Ope...@li...
> https://lists.sourceforge.net/lists/listinfo/openbts-discuss
>



-- 
Regards,
Alexander Chemeris.

SIPez LLC.
SIP VoIP, IM and Presence Consulting
http://www.SIPez.com
tel: +1 (617) 273-4000