[time-nuts] Fury Realhamradio listing

SAIDJACK at aol.com SAIDJACK at aol.com
Sun Apr 29 01:05:16 EDT 2007


In a message dated 4/28/2007 19:14:40 Pacific Daylight Time,  
tvb at LeapSecond.com writes:


>In reporting 1PPS TI, did you measure what  averaging
>time the Z3801A uses? And what averaging time do  you
>use in your Z3801A emulation?

Hi Tom,
 
What I found in the Z3801 manual was that they use some sort of  averaging on 
their TI number (see page 3-20 for example). Does anyone know their  
algorithm for this? There is no specification or details how their TI value  is 
calculated. So far no one seems to have inquiried about averaging done  by the 
Z3801A on the TI output values.
 
We do not emulate the Z3801A for comparative reasons, we implemented the  HP 
commands simply to get GPSCon running. Again we are following the 58503A more  
closely than the Z3801A and are not trying to replace the Z3801A (That would  
mean the user would have to buy a clunky old 48V power supply now, wouldn't 
it  :)

>I ask because the single-shot granularity of the  Z3801A
>appears to be around 100 ps, at least the TI value of  a
>Z3801A syst:stat is always a multiple of 0.1 ns. For a
>fair  comparison, what is equivalent the Fury single-shot
>resolution?


They can get 0.1ns resolution on the printout because of  averaging. I don't 
know what their native granularity is.
 
Single shot resolution may not have such a significant  impact since the GPS 
used in the Z3801A is so much worse than the  M12+.
 
Our resolution is more coarse than 100ps, but our inherent GPS noise is  very 
significantly lower than the Z3801. Resolution of 100ps with a peak to peak  
noise of 100ns or more is not as good as peak to peak noise of 31ns with  RMS 
noise of <10ns.
 
So even if they have true 100ps hardware resolution (I somehow doubt  it 
since an expensive 54132A only has 150ps resolution) then that's more than  two 
orders of magnitude less than their sawtooth noise - does it really matter  at 
that point?

>Second question, I'm not sure "30 picoseconds"  means
>anything. I mean, this is a GPSDO -- and so when you
>talk  about its average value compared to UTC it seems
>you're just comparing  the Fury with itself, right? What
>am I missing?

 
Well, this is the number that GPSCon gives. You may need to ask Graham what  
the significance of this number is.
 
GPSCon calculates the average of all samples over the power-on time  
intervall. Same way to measure it for the Z3801A and the Fury. Just comparing  the 
Z3801A plots with the Fury plot.
 
>From what I see online, the Fury is better for short (55hours+)  
measurements, so overall less bias I would think. It's only been running a short  time, so 
we shall see if it can maintain this small average error  long-term.
 
bye,
Said
 



************************************** See what's free at http://www.aol.com.



More information about the time-nuts mailing list