openbts-discuss

[Openbts-discuss] OpenBTS Clock

[Samuel L. <sam...@ho...>]

Dear List, I wrote some time ago about some issues I was having with the GPSDO unit with my N210. I had similar problems to those raised by others with phones not camping and not reconnecting once they went out of range etc - these were pointing to the clock. I had gotten linux; GNU C++ version 4.6.1; Boost_104601; UHD_003.004.000-122b947

-- Opening a USRP2/N-Series device...
-- Current recv frame size: 1472 bytes
-- Current send frame size: 1472 bytes
-- Found a Jackson Labs GPS
-- Setting references to the internal GPSDO
-- Initializing time to the internal GPSDO

UHD Warning:
    The hardware does not support the requested RX sample rate:
    Target sample rate: 0.270833 MSps
    Actual sample rate: 0.271739 MSps
kal: Scanning for GSM-900 base stations.
        chan: 40 (943.0MHz + 3.422kHz)  power: 139281.38                      
        chan: 59 (946.8MHz + 3.410kHz)  power: 16115.09                       
        chan: 122 (959.4MHz + 3.528kHz) power: 12133.84 

These issues were not resolved so I have the following questions Is it necessary to have the GPS antenna connected and locked to a GPS signal for the reference clock to work with OpenBTS?Does the ClockTamer clock work well with the N210 and if so which one is better?  Regards 		 	   		  

Re: [Openbts-discuss] OpenBTS Clock

[Thomas T. <tt...@vt...>]

On Mon, Mar 26, 2012 at 2:25 PM, Samuel Liverpool <sam...@ho...> wrote:
> UHD Warning:
> The hardware does not support the requested RX sample rate:
> Target sample rate: 0.270833 MSps
> Actual sample rate: 0.271739 MSps
> kal: Scanning for GSM-900 base stations.
> chan: 40 (943.0MHz + 3.422kHz) power: 139281.38
> chan: 59 (946.8MHz + 3.410kHz) power: 16115.09
> chan: 122 (959.4MHz + 3.528kHz) power: 12133.84

These offsets indicate that something is not working with the GPSDO
reference. These numbers are roughly comparable in performance to the
better side of an unreferenced USRP.

I don't recall from the previous emails, but do you have the recent
Kal commit that updated to the UHD 003.004.000 interface?

> Is it necessary to have the GPS antenna connected and locked to a GPS signal
> for the reference clock to work with OpenBTS?
> Does the ClockTamer clock work well with the N210 and if so which one is
> better?

I don't know enough about the GPSDO unit to say how well it should
perform in the absence of a GPS signal. But, the GPSDO unit includes a
stable OXCO, which I imagine should provide a relatively good
reference on it's own given suitable warmup. Perhaps you could ask
that question on the usrp-users list?

The ClockTamer uses a TCXO, which will be less accurate than a GPSDO
locked unit, but it can be software tuned with single Hz resolution.
Generally, the ClockTamer is used on the USRP1 at 52 MHz, but could
also work at 10 MHz. I will leave it to Alexander for further comment.

  Thomas

Re: [Openbts-discuss] OpenBTS Clock

[Samuel L. <sam...@ho...>]

Hi everyone,
 
Thanks for your responses. To answer the questions asked 
Ths GPSDO is from Ettus
I did use the last available version of Kal.
 
I believe that I have followed the instructions to set up the GPSDO  and kal does say that it found it. I will leave the unit plugged in for a 24hr period to see it it locks and produces better numbers.
 
Regards
 

 

> Date: Mon, 26 Mar 2012 17:03:55 -0400
> Subject: Re: [Openbts-discuss] OpenBTS Clock
> From: tt...@vt...
> To: sam...@ho...
> CC: ope...@li...
> 
> On Mon, Mar 26, 2012 at 2:25 PM, Samuel Liverpool <sam...@ho...> wrote:
> > UHD Warning:
> > The hardware does not support the requested RX sample rate:
> > Target sample rate: 0.270833 MSps
> > Actual sample rate: 0.271739 MSps
> > kal: Scanning for GSM-900 base stations.
> > chan: 40 (943.0MHz + 3.422kHz) power: 139281.38
> > chan: 59 (946.8MHz + 3.410kHz) power: 16115.09
> > chan: 122 (959.4MHz + 3.528kHz) power: 12133.84
> 
> These offsets indicate that something is not working with the GPSDO
> reference. These numbers are roughly comparable in performance to the
> better side of an unreferenced USRP.
> 
> I don't recall from the previous emails, but do you have the recent
> Kal commit that updated to the UHD 003.004.000 interface?
> 
> > Is it necessary to have the GPS antenna connected and locked to a GPS signal
> > for the reference clock to work with OpenBTS?
> > Does the ClockTamer clock work well with the N210 and if so which one is
> > better?
> 
> I don't know enough about the GPSDO unit to say how well it should
> perform in the absence of a GPS signal. But, the GPSDO unit includes a
> stable OXCO, which I imagine should provide a relatively good
> reference on it's own given suitable warmup. Perhaps you could ask
> that question on the usrp-users list?
> 
> The ClockTamer uses a TCXO, which will be less accurate than a GPSDO
> locked unit, but it can be software tuned with single Hz resolution.
> Generally, the ClockTamer is used on the USRP1 at 52 MHz, but could
> also work at 10 MHz. I will leave it to Alexander for further comment.
> 
> Thomas
 		 	   		  

Re: [Openbts-discuss] OpenBTS Clock

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

Samuel,

26.03.2012 22:26 пользователь "Samuel Liverpool" <sam...@ho...>
написал:
>
> Dear List,
>
> I wrote some time ago about some issues I was having with the GPSDO unit
with my N210. I had similar problems to those raised by others with phones
not camping and not reconnecting once they went out of range etc - these
were pointing to the clock. I had gotten linux; GNU C++ version 4.6.1;
Boost_104601; UHD_003.004.000-122b947
>
> -- Opening a USRP2/N-Series device...
> -- Current recv frame size: 1472 bytes
> -- Current send frame size: 1472 bytes
> -- Found a Jackson Labs GPS
> -- Setting references to the internal GPSDO
> -- Initializing time to the internal GPSDO
>
> UHD Warning:
> The hardware does not support the requested RX sample rate:
> Target sample rate: 0.270833 MSps
> Actual sample rate: 0.271739 MSps
> kal: Scanning for GSM-900 base stations.
> chan: 40 (943.0MHz + 3.422kHz) power: 139281.38
> chan: 59 (946.8MHz + 3.410kHz) power: 16115.09
> chan: 122 (959.4MHz + 3.528kHz) power: 12133.84
>
> These issues were not resolved so I have the following questions
>
> Is it necessary to have the GPS antenna connected and locked to a GPS
signal for the reference clock to work with OpenBTS?

I know nothing about your GPSDO, but it's logical to assume that GPS lock
I'd required for it to assume correctly. As Thomas mentioned, OCXO usually
have  better stability then TCXO. But the initial offset is often huge. So
you usually have to lock it before the use. Then you may be able to
disconnect antenna, but you have to check this with your GPSDO datasheet.

> Does the ClockTamer clock work well with the N210 and if so which one is
better?

I don't know anything about your GPSDO, so it's hard to guess. What I could
say for sure it's that ClockTamer is good enough to recalibrate it about
one a month if you use it indoors. And you could recalibrate without GPS
using kal only.

Sent from my Android device.

--
Regards,
Alexander Chemeris
CEO, Fairwaves LLC
http://fairwaves.ru

Re: [Openbts-discuss] OpenBTS Clock

[Nick F. <ni...@et...>]

On Tue, Mar 27, 2012 at 4:01 AM, Alexander Chemeris <
ale...@gm...> wrote:

> Samuel,
>
> 26.03.2012 22:26 пользователь "Samuel Liverpool" <sam...@ho...>
> написал:
> >
> > Dear List,
> >
> > I wrote some time ago about some issues I was having with the GPSDO unit
> with my N210. I had similar problems to those raised by others with phones
> not camping and not reconnecting once they went out of range etc - these
> were pointing to the clock. I had gotten linux; GNU C++ version 4.6.1;
> Boost_104601; UHD_003.004.000-122b947
> >
> > -- Opening a USRP2/N-Series device...
> > -- Current recv frame size: 1472 bytes
> > -- Current send frame size: 1472 bytes
> > -- Found a Jackson Labs GPS
> > -- Setting references to the internal GPSDO
> > -- Initializing time to the internal GPSDO
> >
> > UHD Warning:
> > The hardware does not support the requested RX sample rate:
> > Target sample rate: 0.270833 MSps
> > Actual sample rate: 0.271739 MSps
> > kal: Scanning for GSM-900 base stations.
> > chan: 40 (943.0MHz + 3.422kHz) power: 139281.38
> > chan: 59 (946.8MHz + 3.410kHz) power: 16115.09
> > chan: 122 (959.4MHz + 3.528kHz) power: 12133.84
> >
> > These issues were not resolved so I have the following questions
> >
> > Is it necessary to have the GPS antenna connected and locked to a GPS
> signal for the reference clock to work with OpenBTS?
>
> I know nothing about your GPSDO, but it's logical to assume that GPS lock
> I'd required for it to assume correctly. As Thomas mentioned, OCXO usually
> have  better stability then TCXO. But the initial offset is often huge. So
> you usually have to lock it before the use. Then you may be able to
> disconnect antenna, but you have to check this with your GPSDO datasheet.
>
> > Does the ClockTamer clock work well with the N210 and if so which one is
> better?
>
> I don't know anything about your GPSDO, so it's hard to guess. What I
> could say for sure it's that ClockTamer is good enough to recalibrate it
> about one a month if you use it indoors. And you could recalibrate without
> GPS using kal only.
>
The offset you are seeing is around 4ppm, which is not unheard of for an
uncalibrated OCXO. Have you previously allowed the GPSDO to lock and
stabilize for >24h? The GPSDO self-calibrates its OCXO and attempts to
compensate for initial offset, aging, etc., but must be allowed to at least
lock every once in a while to obtain a time reference to compensate with.

When locked to GPS or operating on holdover, the GPSDO should achieve
better than 1ppb frequency accuracy, so your drift as read by kal should be
around 1Hz.

Best,
Nick



>  Sent from my Android device.
>
> --
> Regards,
> Alexander Chemeris
> CEO, Fairwaves LLC
> http://fairwaves.ru
>
>
> ------------------------------------------------------------------------------
> This SF email is sponsosred by:
> Try Windows Azure free for 90 days Click Here
> http://p.sf.net/sfu/sfd2d-msazure
> _______________________________________________
> Openbts-discuss mailing list
> Ope...@li...
> https://lists.sourceforge.net/lists/listinfo/openbts-discuss
>
>