[time-nuts] Interesting jump in my Z3805

[Magnus Danielson]

Steve Rooke skrev:
> 2009/7/4 Magnus Danielson <magnus at rubidium.dyndns.org>:
>> Steve Rooke skrev:
>>> Attached plot of jump in my Z3805. Looks like there was a jump and
>>> then then a rapid track back to the "normal" frequency drift. Comments
>>> please?
>> Small picture, a bit hard to interprent. The green is?
>>
>> I think I correctly interprented the length of the plot to a week (the lower
>> plot shows a regular pattern of 7 cycles and I think I interprent the red
>> text to day 168,71 hrs). If it would have been a frequency jump it would
>> track in faster than half a day. Can you come to think of something else
>> that could have happend to the equipment that day?
> 
> Your right, it is a bit small so I've managed to figure out how to get
> the bigger picture (attached).
> 
> From the help help file of the application"
> 
> "Finally, graph mode presents a detailed graph of the EFC (electronic
> frequency control), TI (time interval value), PU (predicted
> uncertainty) and the number of satellites being tracked. The cursor
> readout indicates the time and date, and either the EFC or the TI
> corresponding to the location of the cursor.
> 
> On typical Z3801A's, measurements confirm that 1 count of EFC is
> roughly equivalent to 1 part in ten to the twelve."

That clear up all those issues! Thanks! :)

> The event seems to start at about 04:00 30/6/09 and is over at about
> 16:00 on 30/6/09. There is no indication that there was a complete
> loss of lock to the birds at that time and the number of sats looks no
> lower than at other times (OK it's not an ideal number of sats but my
> antenna is in a temp position at the moment and cannot see all of the
> sky).

Infact, if you look at the sat lock trace, you see the same pattern as 
every other day. Your antenna position do make you loose sight of a few 
birds as you can see very clearly as a type of sine-shaped pattern 
occuring, but only the 24 hour occurences is as sharp, but you do see 
the 12 hour pattern if you look a bit more closely.

The antenna position could improve, but I would not say it is the root 
cause here.

> There is no actual loss of lock, it shows that the peak
> deviation at that time was about 35ns but obviously something changed
> to cause the EFC needing to track away from the "norm" to remain in
> lock. I think I spent most of that day in bed with the flu that's
> going on here so it was very quiet. We have been having a fair bit of
> rain here but it has been over many days, not just this Tuesday. Just
> thinking now, but these times are in UTC so that would represent about
> 4pm on Tuesday to 4am on Wednesday, again most of time in bed.

Hmm. Hope you are clear from that flu now. Nasty buissness.

I don't think it is the crystal making a frequency jump, or is it?
Such a jump would be high-pass filtered by the locking PLL. Hmm. Looking 
at the TI is looks like it starts with a steep slope, so it could be 
such an event, or any other similar event. The EFC responce looks a bit 
pale thought, so I wonder if that is really what happend. Need to ponder 
some more.

Cheers,
Magnus