[time-nuts] Rb lamp lifetime...
bruce.griffiths at xtra.co.nz
Tue Nov 18 19:50:42 UTC 2008
> Yuri Ostry wrote:
>> Sunday, November 2, 2008, 6:31:32, WB6BNQ wrote:
>> W> Hi Yuri,
>> W> Unless you need the longer term stability of an undisciplined Rb source, I think
>> W> you would be better served to get one of the “Timenuts group” Trimble Thunderbolt
>> W> GPS units to use as a home standard. It is available via this page:
>> W> http://www.tapr.org/kits_thunder.html
>> W> and info about the unit is available at this URL:
>> W> http://www.leapsecond.com/tbolt-faq.htm
>> W> Add an external very high quality crystal oscillator locked to the Trimble and
>> W> you will have the same stability of a good Rb source. Most would use a high
>> W> quality Rb source, undisciplined, as a tool for comparing or generating
>> W> specifications of crystal oscillators.
>> Unfortunately, I'm currently living in a place that have problems with
>> good location for GPS antenna. The best place available provides less than
>> 40% of open sky view and a lot of buildings around, lots of reflected
>> signals. No access to roof, security is tightened heavily after 9/11
>> and terror acts in Moscow, so it is big problem to get access to
>> attics or cellars of residental buildings.
>> I curently have uBlox LEA-5S (SuperSense nav module) connected to this
>> antenna.... Not very promising... Fix is available about 70% of time
>> but drifting heavily (due to multipath?), often I see coordinates
>> about 400-500 m away from real location.
>> I have Motorola Oncore M12+T module as well (purchased it with
>> intention to build GPSDO controller), but (IMHO) it will be nearly
>> useless in such conditions with original W5OJM controller design.
>> I thinked about some FLL implementation where microcontroller
>> constantly monitor output of GPS for T-RAIM solution status and
>> discard measurement cycles that have T-RAIM alarms, restarting cycle
>> once T-RAIM reports good timing solution again... Will be fun project,
>> of course, but will take some time. And it will be a good reason to
>> familiarize myself with new 16-bit PIC24 series... Their cascadable up
>> to 32 bits timers/counters with input capture looks quite promising...
>> Best regards,
>> Yuri, UA3ATQ/KI7XJ mailto:yuri at ostry.ru
> Hi Yuri,
> Someone correct me if I am wrong, but I think the M12+ is like the UT+ and uses one
> bird, at a time in the time mode, to do the timing process after you got a position
> fix. If that is true then maybe it will yeild a better result ? I think it would be
> worth a try. So, put the Shera controller together with your M12+ and see what happens.
> I am wondering if limiting the antenna's view would help with the multipath ? For
> instance, what happens if you build a cone shaped skirt that is pretty tall such that it
> prevents a majority of the side view energy from reaching the antenna ?
> Worst case, you could use the M12+/Shera arrangement with a Rb unit and go to a park or
> other open space for a day, or if camping is available, maybe a weekend camping trip
> would be in order. Then compare the Rb oscillator, setting it if need be, and then take
> it home. Not an ideal solution, but with the much better holdover of the Rb source you
> would, maybe, at least have perhaps a few parts in 10E-10th for a month or so before
> having to do it again.
The M12+T uses all SVs above the user defined altitude limit to compute
the timing solution.
However the individual time offsets for each SV from the global solution
are available, so in principle one can remove one or more SV's from the
This requires either hardware or software timing error correction to do
Thus, in principle, one could implement a much more complex SV selection
algorithm than a simple altitude limit.
Since switching to holdover mode will be relatively frequent a more
versatile controller than Shera controller will be required.
If you use an appropriate micro most of the required hardware is built
in to the chip.
Neither external counters nor auxiliary oscillators are necessary as
everything can be clocked at 10MHz.
More information about the time-nuts