[time-nuts] Datum/Exactime 9390-6000

Ziggy ziggy9 at pumpkinbrook.com
Sun Jun 24 15:50:13 UTC 2012


Just thought I'd give a progress report.

I've successfully placed a micro (Arduino) between the GPS and the system board, can spit out the packets in hex, and inject/modify data on the fly. So I can add 1024 to the GPS week and the unit does the date conversion correctly. But it gets more interesting.

What I've found is that the GPS module has a nonstandard firmware. It still sends TSIP, but the GPS time packet 0x41 has a different format:

Header (0x10)
GPS time of week (4 byte single precision float)
GPS week number (integer)
GPS/UTC offset or leap second count (integer)
??? 4 bytes
Trailer (0x10 0x03)

For reference, a standard TSIP GPS time packet 0x41 is:
Header (0x10)
GPS time of week (4 byte single precision float)
GPS week number (integer)
GPS/UTC offset (4 byte single precision float)
Trailer (0x10 0x03)

Instead of changing the week number, I can change the offset from 15 to 17 and the unit correctly flips to the next epoch, showing the right day and year. I'm off by 2 seconds then, but it proves that field is used as the leap second count.

The next four bytes in question are not just the GPS time shifted. If I can get this decoded, I may be able to use a more modern receiver with better PPS performance.

Some snippets of data are below in case it looks familiar to anyone. There's something nagging at me, but it's eluding me for now. If anyone would like to help, I would much appreciate it and we could continue off list.


10 41 45 5B E7 6D 06 9E 00 0F 30 C4 7E 56 10 03 (bytes in red decoded as binary)
10 41 45 5C 38 D9 06 9E 00 0F 30 C4 7D 39 10 03 0011 0000 1100 0100 0111 1101 0011 1001
10 41 45 5C 8C B4 06 9E 00 0F 30 C4 7C 13 10 03 0011 0000 1100 0100 0111 1100 0001 0011
10 41 45 5C EE 04 06 9E 00 0F 30 75 F5 7E 10 03 0011 0000 0111 0101 1111 0101 0111 1110
10 41 45 5D 3E C9 06 9E 00 0F 30 75 F3 49 10 03 0011 0000 0111 0101 1111 0011 0100 1001
10 41 45 5D 91 91 06 9E 00 0F 30 75 F1 06 10 03 0011 0000 0111 0101 1111 0001 0000 0110
10 41 45 5D F2 93 06 9E 00 0F 30 75 EE 5F 10 03 

I thought the GPS week increment might show something interesting, but it wasn't really:

10 41 49 13 A7 80 06 9D 00 0F 31 24 FF A0 10 03 
10 41 49 13 A7 C4 06 9D 00 0F 31 24 FF D3 10 03 
10 41 3F A7 0A 3D 06 9E 00 0F 31 25 00 10 10 10 03 (extra 0x10 is an escaped inline value)
10 41 40 BE 97 8D 06 9E 00 0F 31 25 00 47 10 03 
10 41 41 2C 87 2B 06 9E 00 0F 31 25 00 81 10 03 

Paul
----

On Jun 12, 2012, at 12:26 PM, Ziggy wrote:

The GPS unit is a module, which seems to be a a Trimble SVeeSix-CM3, or variant of some kind. It's marked 26889-81 on a sticker and 25040-  -D silk screened on the board.

Just as an interesting observation: This GPS module has two MCX connectors, one on the top of the board for the antenna and one on the bottom which mates with the system board. Directly adjacent to this is a 16.368 (or thereabouts) TCXO can. It works with a standard antenna, so it looks like this is setup to work with both downconverting and standard antennas.

I had to go to the Internet Archive to find the old Trimble page (http://www.trimble.com/y2kwnro/index2.htm) on Y2k/WNRO status. While this specific module wasn't listed, it didn't look encouraging. I do have a Trimble ACE III receiver which I am going to try using tonight. It is supposed to report the absolute GPS week since 06 January 1980 as an integer value and not truncated to 10 bits. Trimble says: "The ACE III GPS module has been designed to handle WNRO, and there are no problems
with either dates or first fix after WNRO through the year 2015." Not sure what happens after 2015...

Note that due to the 16.368MHz TCXO can being right next to the GPS module, I had to modify the ACE III by changing the connector type to match the existing connector and move it to the other side of the board. I'll report back with the results. Maybe I'll get lucky for once.

Paul



On 06/12/2012 03:52 AM, Azelio Boriani wrote:
> Is the GPS unit built-in (TBolt-like) or is it a separate module? Maybe
> there is a firmware update for the GPS unit itself to fix the rollover
> problem.
> 
> On Tue, Jun 12, 2012 at 5:26 AM, Ziggy<ziggy9 at pumpkinbrook.com>  wrote:
> 
>> I've recently acquired one of these, primarily to use as an alternative to
>> (and for comparison with) my Thunderbolt. This was a DC model that was
>> inoperative when I received it and at $225 (price from the OEM) to replace
>> the failed DC-DC converter, I installed an AC supply. So far, so good. The
>> unit powers up, tracks, locks the OCXO and things look generally good. I'm
>> giving it some burn in time before I try to characterize it. However, I
>> have a couple of issues. I did search the archives, and have contacted
>> Symmetricom (who politely told me this unit was obsoleted 2 generations of
>> product ago :)
>> 
>> The menus and serial command responses do not match any version of the
>> manual that I can find, although the Datum 9390 manual on Didier's site is
>> pretty close on the commands. Does anyone have a manual that covers this
>> model/firmware?
>> 
>> This unit suffers from a GPS epoch rollover problem. This makes the date
>> produced incorrect, and there is no way to override it, or offset it in my
>> firmware (DT813MP) other than running it in 'freewheel' or undisciplined
>> mode. It's my understanding that there may have been a firmware update to
>> address this, so if you have a firmware image, or a unit with later
>> firmware it would really be great to get a PROM copy/image.
>> 
>> And perhaps a silly question: Is the LCD supposed to be illuminated?
>> Because mine isn't. I haven't taken the front panel apart yet to check out
>> the display - I've been more concerned with getting the unit operational.
>> 
>> In the end, it's pretty usable the way it is, but the date being off irks
>> me. I'd like to fix that if I can, and docs that match the hard/firmware
>> would be nice. Any help would be greatly appreciated.
>> 
>> Best regards,
>> Paul - K9MR
>> _______________________________________________
>> time-nuts mailing list -- time-nuts at febo.com
>> To unsubscribe, go to
>> https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
>> and follow the instructions there.
>> 
> _______________________________________________
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.


_______________________________________________
time-nuts mailing list -- time-nuts at febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.




More information about the Time-nuts_lists.febo.com mailing list