On Tue, Jul 3, 2012 at 3:27 PM, Alexander Huemer alexander.huemer@xx.vu wrote:

On Tue, Jul 03, 2012 at 03:08:02PM +0200, Dario Lombardo wrote:

...

On Tue, Jul 3, 2012 at 1:44 PM, Sylvain Munaut 246tnt@gmail.com wrote:

...

...

In some past post of the ML, I've read that someone was trying to get the GSM 10MHz clock reference from the motorola phones. This aimed at feeding usrp with a stable reference, if I'm not wrong. Anything new this side? This would be extremely useful if GPS signal can't be got but GSM cell can.

1. It's not 10 MHz but 26 MHz

Mmm... but the front panel of usrp is fed by 10 MHz/1pps... isn't it?

There are different revisions of the USRP. Sylvain is talking about the USRP1, you of a later rev. The USRP1 has an (unpopulated) direct clock input. The other USRP revisions have a reference clock input, which is something different.

You are definitly right. I was taling about N210, which can be fed by reference clock input.

...

...

2. Yes it works, but you can't feed the USRP directly, you need to

multiply it by 2 (using a PLL chip) to get 52 MHz for the USRP.

Can you provide some link to read something about it?

See [1,2,3]

...

Don't you think that this way could be much more preferrable than the GPS way, if USRP is under GSM coverage? I mean: the clock from GSM is next to us... why just don't use it? Probably there's something I'm missing, since this approach is much less used than GPS clocking...

There are many ways to provide a stable reference clock. A OCXO (e.g. [4]), a cheap rubidium standard, the clock tamer[5], ... Every option has up and downsides.

Just the "pirelli" way uses clock from existing GSM network. As said before I think that "reusing" the clock from network could be very useful, but I don't know if it is more difficult than generate the signal itself. Is that possible using motorola phones? Did anyone investigate that possibility?