[time-nuts] Thunderbolt Rollover
Tom Van Baak
tvb at LeapSecond.com
Wed Sep 17 10:44:25 EDT 2014
I don't think you can use a GPS almanac from 2012.
Let me know what format you're using (Yuma, SEM, SP3, RINEX nav). The two fields, TOA (Time of Applicability) and GPS week, need to be consistent with the fake futuristic UTC date that you plug into the simulator. Otherwise the receiver discards this almanac/ephemeris data as being out of date. It is odd the simulator doesn't take care of this automatically. Contact me off-list with details.
Ask Mark to send you a beta version of LH and you can test it for him.
Also, with or without the fix to LH, it would be cool if you could verify that the Thunderbolt itself is fine, both July 30, 2017 (Trimble TBolt rollover) and March 31, 2019 (next GPS rollover). By fine, I mean the 1PPS and the 10 MHz outputs don't skip a beat. I suspect, like most GPSDO, it will continue to work perfectly. But it's nice to get confirmation.
----- Original Message -----
From: "Matthias Jelen" <Matthias.Jelen at gmx.de>
To: "Tom Van Baak" <tvb at leapsecond.com>; "Discussion of precise time and frequency measurement" <time-nuts at febo.com>
Sent: Tuesday, September 16, 2014 1:52 PM
Subject: Re: [time-nuts] Thunderbolt Rollover
I didn´t save the output, but the generator is located on my
workbench, so it´s no problem to repeat this experiment at
But I´m afraid it´s a bit more complicated. I need to
specify a almanac file on the simulator. Stored was a file
from 2012 which was shown to be valid for a month or so.
When I put the date in that region, I got a fix almost
immediately. The manual says that those almanac files are
available on the internet, and ideed I found the current
almanac. But I have no clue yet how to generate a valid
signal for a time in the future. Putting the date far out of
that region lead to a loss of the fix, no matter if before
or after 2017.
I don´t know to much about the GPS system, but I assume that
the week number and TOW are "broadcasted" by all satellites
and can be decoded without valid almanac data. The date in
the 0x8F-AB data packet was set to the expected date in the
past which seemed to be good news for me.
So the only way to be really sure would be to see a real fix
with the date in the future. I´ll have a second and a third
look in the manual and web, and if this doesn´t help I´ll
try to get in touch with one of the experts from the
department which developed that box - they should know if
and how this is possible...
I´ve seen that LH sources are available, but I´m afraid that
my C++ knowledge is by far not sufficient to dig into this :-)
Let me know if a copy of the TSIP packages is helpfull for
you, even without a fix.
Any suggestions are welcome!
> Hi Matthias,
> Thanks much for sharing this information. You are fortunate to have access to a GNSS simulator.
> May I assume from your comments that the TBolt itself was fine -- it continued to track SV and the 1PPS and 10 MHz outputs were not affected? This is the expected behavior of GPS timing receivers across any sort of 1024 week rollovers.
> Do you have a binary capture of the TSIP packets I could take a look at?
> Note the source code to Heather is included in the distribution. Or we can just ask Mark Sims (who wrote LH, and is on the list) to add the 936 week offset back into the timestamps for dates after July 2017.
> ----- Original Message -----
> From: "Matthias Jelen" <Matthias.Jelen at gmx.de>
> To: <time-nuts at febo.com>
> Sent: Tuesday, September 16, 2014 12:06 PM
> Subject: [time-nuts] Thunderbolt Rollover
>> I got a Trimble Thunderbolt a few months ago and implemented
>> a simple monitor on a MCU to display time, no of satellites
>> etc. on a lcd display.
>> While working on this, I found the hint for the upcoming
>> rollover in 2017 in the user manual. I couldn?t find any
>> details on this in the internet - the question for me was if
>> the unit becomes useless in about three years (which would
>> minimize my motivation to put all the stuff in a shiny
>> enclosure) or if simply the calculation of the date will fail.
>> So I took the unit to work and hooked it to a signal
>> generator capable of simulating GPS, GNSS etc. systems. As
>> expected, everything works fine until July, 2017. If I try
>> dates later than this, Lady Heather refuses to display time
>> and date. The strange thing is that time and date are still
>> sent in the serial protocol. August 20th, 2018 is decoded as
>> January, 4t, 1999. As expected, time is OK.
>> So, hopefully this is only a problem in LH and the
>> Thunderbolt will be usable after July, 2017 - I?d be happy
>> to live with a wrong date.
>> Can anyone confirm this observations or does anyone have
>> more information on this topic?
>> Thanks a lot and regards,
>> Matthias, DK4YJ
More information about the time-nuts