Symmetricom Syncserver S350 + Furuno GT-8031 timing GPS + GPS week rollover

[OH2LAK]

I've managed to salvage a Symmetricom Syncserver S350 which is happily referencing to 1PPS and NTP but fails to refer to GPS.

The GPS is operational, it can provide a 3d solution and accurate location but the unit fails to end up in the favored result, GPS being the valid time reference.

I have now through multiple forum posts determined this has to doo with GPS week rollover and have found documents which advice that the GPS can be told manually which year we are now, after which it should be able to provide a accurate time & date.

I have a copy of the Furuno service bulletin SE18-100-034-02 here; https://1drv.ms/b/s!AvS_Uj9NsWu3h6RbahwXg81huphiBg?e=WpKgip
The document states in chapter 6;

(3) GT-80/GT-8031/GO-80
The receivers can update the date up to year 2079 by entering the date with ZDA command.

The GT-8031 I/O Protocol specification on page 8 states the format of the ZDA command structure which is rather self-explanatory.
The document is available here; https://1drv.ms/b/s!AvS_Uj9NsWu3h6RdEpuQda_pqochPw?e=9QfC3A

Now the actual question; Is the serial port of the Furuno GT-8031 GPS unit available to the core linux host running in the S350 or is it wired to the FPGA? I could not find a reasonable UART or TTY where the GPS would be attaching to. Just thinking how to issue the ZDA command to the GPS, could it be done from the linux host or do I need to wire a serial port converter directly to the GPS unit.

Thanks,
Erik

[JoeFloyd]

The 10 Pin connector between the GPS unit and the mainboard carries TTL serial Rx/Tx, power, 1PPS.  There are two vias on the side of the GT-8036 board facing upward when connected to the main board.  They are under one of the stickers.  It's pretty obvious which via is the Tx signal coming from the GT-8036.  This produces a burst of TTL data every second.  The Rx is much less active.

I decoded the serial data Rx/Tx and can confirm that when GPS Week roll-over occurred in  2019 each GT-8036 was on borrowed time.  The way UBlox describes the GPS week rollover issue is that the firmware date is checked to see if the receiver was manufactured within a period of time leading up to the roll-over.  I assume Furuno did something similar. 

The 1 PPS signal from the GT-8036 is still valid.  The time is still valid.  The date is completely invalid. 

There are plenty of TTL to USB serial adapters.  I'll take a picture of the GT-8036 and show which vias I'm referencing.  If you have a scope you should be able to find them pretty quick. 

Another possibility is to track down the GT-8736.  It is a drop in replacement for the GT-8036 and works fine in the S300.   I recently purchased several GT-8736 modules for about $50 US each.  Much cheaper than other options which are closer to $400 each.  As far as I know the GT-8736 is no longer in production so it did take some effort to track down a seller who had it in stock.  I bought the last of their stock as far as I know. 

https://linkwave.co.uk/gt-8736-gnss-timing-module

The website currently shows the modules in stock, so you might be able to order from linkwave.  Linkwave would not process a non-UK order, so I had to use a mail forwarding service, but it was still much much cheaper than any other option I could find to get a S300 back into GPS lock.

The GT-8036 is a M12 (motorola binary format) GPS receiver which is what the Symmerticom S200/S300 NTP servers require.  As far as I know, the GT-8031 does not implement the motorola binary format and the S300 you have won't be able to talk to it. 

What model of GPS receiver is currently in the S350 you have?   

[OH2LAK]

Hi Joe.

Thanks for very elaborative answer! My S350 has a Furuno GT-8031F GPS unit inside. I was exactly thinking about a drop-in replacement for it as I'm not completely sure would the ZDA command fix it for good or only temporarily, the protocol specification is a bit unclear is the date command volatile or backed up.

I'm good with measurement equipment so I'm sure I'll find the right pins but pictures are always nice in these threads as they might then help others too. This might be the first time I can really utilize my oscilloscope's serial data decoding option!

Decoding the lot number of the GT-8031 it seems to match the previous GPS epoch, I figured the test year 0 refers to year 2000. Or maybe 1990? The HW spec document is dated 2003 so maybe then 2000.

3 0 9 01 611 0820 1 3

Test year 0
Test month 9
Test date 01
Model code 661
Mounted date 0820
Inspection machine ID 1
Check Digit 3

Looking again at the GT-8031 spec, it seems not to be able to output Motorola binary format. I wonder then is the S350 able to identify somehow the module used and adopt to that.

Thanks for the link to Linkwave, I have engaged my elfs in the UK territory to help me get a christmas gift

Br,
Erik OH2LAK

[JoeFloyd]

To clear up some confusion.

The GT-8036 timing receiver has a GT-8031 on it.   

The GT-8036 also has an MSP430 which translates NEMA into Motorola messages and vice versa.   This is how other modern Motorola binary format GPS receivers work.

Attached are photo of the vias which are the RX/TX in Motorola binary format.  The GT-8031 has a separate pin header which should have the NEMA protocol exposed.

[dj831]

Hi everyone, just for your information: tested yesterday, and it doesn't work  . Explanation:

1. At power on, Furuno gives:

$PFEC,GPtps,020101000002,1,0,1,000000000000,00,00,000000000000,1147,172802 -> week 1147
$GPGGA,000003,3444.0000,N,13521.0000,E,0,00,00.00,000000.0,M,0036.7,M,, -> default coordinates
$GPZDA,000003,01,01,2002,+00,00 -> default date, 01.01.2002

2. Sent ZDA command:$GPZDA,235000,20,05,2023,+02,00. Command accepted:

$GPGGA,235200,3444.0000,N,13521.0000,E,0,00,00.00,000000.0,M,0036.7,M,,
$GPZDA,235200,20,05,2023,+02,00

3. Power cycle. Back to default date (2002)

4. Sent ZDA command, plus command $PFEC,GPclr,2: backup data, and reset module. After reset, date was 2023. Power cycle: back to default date again.

The Furuno module is able to backup data as per bulletin SE18-100-034-02, but in RAM only! (a complete nonsense for me). It would work if module was battery backed up, but does not work as-is. Next step: try HD-8736 card from Heoldesign (France). Not much data on it, but is is plug and play. As per Company, it can be set on glonass and Galieo through serial commands, but have not been tested yet - is is preprogrammed for GPS only. Also, no GPS week rollover 'till 2042, but that can be corrected through serial port.

[dj831]

Yet another update ...

As I wanted to test further, I added a battery backup, to overcome loss of week when powered off. The Pentar module already contains  a voltage supervisor (a TPS3619-33) that could have been able to handle battery backup, but that is just used for reset (both MSP430 and Furuno module). Vout of TPS is connected to VBackup of GT-8031, and 'battery voltage' on Pentar board comes from a diode on Symmetricom motherboard, via a 10k resistor on Pentar board.

I disconnected 10k resistor, I connected 'on the fly' a CR2032, and I cut TX track from MSP430 connected to RX of the Furuno module, to ensure that there is no conflict between my RS232 interface and processor (though I added a pair of 100R resistors on MAX3232 side, just in case).

1. Command sent to Furuno: $GPZDA,235000,20,05,2023,+02,00 (as previous post).

2. Power cycle (with battery as backup, of course).

3. Output gives: $GPZDA,000505,21,05,2023,+02,00 -> command accepted, with GPS holding time.

4. Module put back on S250. As before, satellites found, but still unable to sync to GPS (S250 was powered several hours).

5. Module removed from S250, and read with RS232 interface, with TX pin still in place. ZDA command is issued only once at power-up. Sadly, it gives: $GPZDA,172218,06,10,2003,+00,00. However, $GPGGA outputs correct coordinates.

If we compute today's date (22 may 2023) from rollover day (18 sep. 2022), we have 247 days. And if we compute difference from 6 oct. 2003 to output date after internal rollover (2 feb. 2003), we get exactly 247 days. Pretty obvious.

Conclusion: ZDA command is useless, even with a battery backup. GT-8031 will NEVER work any longer in Symmetricom S200/S300 series despite ZDA command 

[comsec22]

I have the same issue on Symmetricom XLi, where a Furuno-8731H act with a roll over issue...
In the mean time I order an HD-8736 (Heol Design), wishing well 

I am only wondering why, except the position, no way to see GPS time and date, even wrong because of roll over.. is it a normal
behavior on Symmetricom's receiver ?

Cheers

[dj831]

Ciao comsec22,

I confirm that card works flawlessly - you just have to set it to dynamic, to see satellites, and switch to hold - you no longer see satellites but system works fine. As for your Symmetricom, I can't tell you, since I have an S250, and I see satellites and a wrong date with GT-8031.