[time-nuts] GPS clock error.

[Hal Murray]

> This morning at 10 AM CDT my GPS clock read 8 PM July 5th.  My wife reported
>  that the time had been 2 hours off at 6 AM local time.  She didn't notice
> if it was AM or PM.  The parabolic dish icon was missing from the display.
> I manually set the time and date but when compared to my two WWVB clocks it
>  was clear it was in holdover mode.  I waited about 3 hours then removed the
>  batteries and reinstalled them.  ...

Could you say more about this clock?

How long do the batteries last? ...

I'm familiar with battery operated "atomic" clocks that listen to WWVB.  I 
didn't know about GPS versions.

I'd expect a WWVB receiver to use much less power but maybe modern GPS 
receivers are good enough so they would have reasonable battery life.

My best guess is that your receiver got tricked by noise that looked good 
enough.  I've seen GPS receivers report that their info was valid when it was 
miles from the reported location.  Usually, that's right after recovering 
from not-enough-satellites.


albertson.chris at gmail.com said:
> There are also endless ways that logic inside th GPS can fail in a "soft"
> way.  Memory can become pattern sensitive or a tiny sense amp in a RAM chip
> can get noisy and cause one in a billion type soft errors.   I don't bother
> to fix things until I can make it repeat on demand 

Memory doesn't usually "become" pattern sensitive.  It might be designed that 
way.

Cosmic rays or alpha particles are the usual ways that DRAM gets soft errors. 
 You can also have noise/crosstalk at the board level (or on chip) or power 
supply problems.

If you want to build a reliable system, you have to pay attention to rare 
bugs.  If nothing else, you want to collect data on them so you know if you 
have a problem and/or how bad it is.



-- 
These are my opinions, not necessarily my employer's.  I hate spam.