[time-nuts] GPS week wrap in HP Z3801A

[Magnus Danielson]

On 12/26/2011 10:40 PM, mike cook wrote:
> Le 26/12/2011 21:31, Hal Murray a écrit :
>> I turned on a Z3801 that had been idle for a while. The status screen
>> said:
>> UTC 20:03:44 [?] 11 May 1992
>>
>> That looks like 1024 weeks (19.7 years) ago.
>>
>> When I first turned it on, before finding any satellites, it said:
>> UTC 12:05:00 [?] 01 Jan 1992
>>
>> Has anybody else noticed troubles like this?
>>
>> I've got another one that still works. It's got a bigger serial number
>> so I
>> assume it's more recent.
>>
>> Does anybody know how to read the firmware date or version? There is a
>> command to get info from the GPS unit, but I haven't found a way to
>> get any
>> info for the main firmware.
>>
>> After a while, the "[?]" goes away.
>>
>>
> IIRC the GPSs in the Z3801As are Oncore VPs. They don't suffer from the
> GPS week rollover, but If the RTC in the receiver is not set or its
> battery has failed, and the reference year stored in EEPROM is also
> wrong, the receiver may select the wrong 20-year window. This may be
> fixed by entering a date command when the receiver is not tracking
> satellites.
>
> send the receiver
>
> /@@Cj
>
> and you will get the revision data.

You do not talk directly to the Oncore VP, but to the clock CPU, which 
has a very different command-set. You can get the information out using 
the commands which is in there. It was some time that I fiddled with the 
Z3801 command-set. I've done some reverse-engineering of the 
command-set, but I think the Z3801A manual suffice to get started.

You will get the Oncore VP info indirectly from there.

I have not figured out if you can do something useful with the built-in 
FORTH interpreter. It would be handy if one could add monitoring 
processes and additional commands for it, but I don't think it is possible.

The Z3801 runs pSOS and includes the pFORTH. I have no details on those, 
but would value information to aid me in further reverse-engineering 
efforts. The Z3801 is written in C, which is evident from the pSOS libc.

Cheers,
Magnus