[time-nuts] Z3801A Problem

[Ed Palmer]

On 11/5/2012 11:27 PM, Hal Murray wrote:
> azelio.boriani at screen.it said:
>> Very interesting... is it using the binary protocol? Maybe a serial link
>> error, the binary protocol has a checksum (yes, NMEA too). Check the serial
>> link levels with a 'scope, maybe that the Z3801 firmware waits to see some
>> consecutive errors before actually reporting that something is wrong,
>> meanwhile the satellite number is set to 0.
> That seems unlikely, but I've been surprised before.
>
> I have two Z3801As using the same antenna which is not in a good location.
> They both go into holdover at the same time about once per day.
> (Occasionally, it's only one.)  I don't have any data on the number of
> satellites.  I assume it goes to 0, but I don't know how long it stays there.
>   It looks like a spike on the scale of 1 sample per minute.
>
> I've been assuming it was something like multipath from an airplane.  I'm
> roughly 20 miles from the San Francisco airport.  I'm not on a main flight
> path, but it's not uncommon to see a large plane go over reasonably low.  I
> assume it's merging into the landing pattern.
>
> Does anybody know how to find out if there are any planes near a specific
> location, and if so how to get the flight path with good timings?
>
> --------
>
> The HP driver in the ntp reference package logs a line like this every minute
> (actually 64 seconds):
> 56237 776.033 127.127.26.1 T2201211060012573001036  64 0
> 56237 840.033 127.127.26.1 T220121106001401300102D  64 0
> 56237 904.044 127.127.26.1 T2201211060015053001032  64 0
> 56237 968.034 127.127.26.1 T2201211060016093001037  64 0
>
> The " 64 0" on the end are counters I've added to my local copy.  The 64 is
> the number of good T2 messages.  The 0 is the number of messages with errors.
>
> One of the columns in the T2 chunk is the FFOM, another is the TFOM.  (I'd
> have to look it up.)
>
> If I was to add some more info to that line, what would you want?  EFC and
> number of satellites are a couple of obvious ones.  EFC probably doesn't
> change very often or fast.  One sample per minute is probably good enough.
> For the number of satellites, we might want both the min and max, and even
> the count at that level.

EFC would be interesting to see.  When the problem occurs on my Z3801A, 
the EFC might do nothing,  or spike, or jump and stay at the new value, 
or jump and recover to the original value.  HUP would also be good to 
record to watch for a degradation in that value.

Ed