[time-nuts] Fwd: Symmetricom TymServe 2100-GPS currently fails with GPS offset

Esa Heikkinen tn1ajb at nic.fi
Thu Jan 29 17:59:48 EST 2015


Tim Lister kirjoitti:

> Hi Esa, I am forwarding this message as I'm not subscribed
> to time-nuts list, just a lurker... It's good to know we're
> not the only ones having problems, even if there is
> no obvious solution. Do you think there is any hope of
> getting Symmetricom/Microsemi to release an updated
> firmware to fix this or are they most likely to turn
> around and say "buy something made this decade..."

It's possible that they cannot fix this even if they wanted to do so
(which I doubt) because it's kinda old product and it may be that
there's no engineers anymore in the house who know this product.

I afraid that only way to fix the firmware is to do itself. I have a
guess that the problem is related to decoding of wrong portions of the
GPS frames. So it would be possible to find where the actual command to
decode UTC offset is used and change that part of the firmware.

This is not easy way at all, but it's possible to do. It consumes HUGE
amount of time. First thing to do is to find out what protocol is used
between mainboard and GPS. Because it uses Trimble GPS module this could
be TSIP (rough guess). To find it out, it's required to install a
"sniffer" between the GPS and mainboard. TSIP protocol specification
seems to be available on the Internet.

Firmware modification is not easy task to do and it's also dangerous,
whole unit may be permamently useless after loading the mofified
firmware, with no way to load correct firmware anymore. There's also
checksum mechanism to prevent "corrrupt" firmware and because of that
the data content must match the checksum or checksum must be updated.

Another (and recommended) way could be to add some simple
microcontroller (Microchip PIC for example) between the GPS unit and
motherboard and write own software which modifies the GPS traffic so
that UTC offset will be correct again. This way also requires some study
first but after the details are known the needed software for data
modification should be very easy to do.

I have motivation to fix this but having no time for this, at least in
near future. Quick fix was to feed it with 1PPS from Thunderbolt.
However it unconfortable to do this every time when leap second is
coming. Also, if the power is lost it will start from January first 2015
00:00:00 and the time must be set manually each time, it does not
recover automatically with PPS. That's why it would be so nice to fix
this, but without any knowledge of the unit details, firmware source
code or GPS protocol it will require many days to do.

Best regards,

-- 
73s!
Esa
OH4KJU



More information about the time-nuts mailing list