[time-nuts] Datum Starloc II GPSDO issues

Mark Sims holrum at hotmail.com
Wed Jun 29 18:02:57 EDT 2016

It appears that they are brand new... no signs of ever being handled or installed.   The anti-static bag is sealed with a yellow warning sticker.   The date codes in the firmware/software ID message indicates year 2000.   The manufacture date message returns 0's in all the fields. 

They have a DC-DC converter module soldered to the PCB.  The second board in the stack is the Motorola timing receiver.  These are OEM boards like the gold boxed Thunderbolts,  not the metal boxed "retail" units like the Trimble "red box" Thunderbolts.  The connectors are a match to the Trimble units,  except the RS-232 connector is a male and requires a null modem cable/adapter.

And for your viewing pleasure, a few more warty warts...  a couple of times an hour the UTC offset field reports 0 even though the receiver has a proper  UTC offset value.   

 In the TSIP binary protocol any 0x10 bytes must be sent as 0x10 0x10.   The end-of-message flag is 0x10 0x03.   If a message contains more than one 0x10 byte,  there is a very good chance that the Starloc will send 0x10 0x10  as the end-of-message flag instead of 0x10 0x03!  This causes the next message to be merged with the previous one, and it will not be decoded.

The satellite health message (0x59) should have a info type byte followed by 32 bytes of info flags (one for each possible satellite).   Our fiends,  the intrepid Starloc coders,  send an info type byte of 00 (invalid, should be 3 or 6) followed by 31 bytes of 0's.

Ahh, the ephemeris status message (0x5B) should have 16 bytes of data in it.   Starloc gives you 15 bytes of zeroes.  

The tracked satellite list message (0x6D) has a byte that says how many bytes of satellite ID numbers follow it...  good ole' Starloc sends however many bytes if ID's it wants to.  The count byte is pretty much useless and if the two don't match, well,  the message is bogus.

The satellite solutions message (8F:A7) should have lots of useful info in it like per-satellite clock bias.  Well Starloc sends a message with just 13 bytes of who know what...  I don't... it ain't nuthin' like what Trimble documents...


More information about the time-nuts mailing list