[time-nuts] Tbolt TC ( was 'time-nuts Digest, Vol 63, Issue 52')

WarrenS warrensjmail-one at yahoo.com
Mon Oct 12 19:38:56 UTC 2009


>>The only downside of the longer time constant seems 
>> to be that the PPS seems to move around more, ...

Because LH's Plots and Adevs are based on the difference of the OXCO and the Tbolts own 
internal filtered GPS signal and not on an external quiet Reference as is the more common practice,
Interpreting the results of  these unique type of Plots and Adevs correctly can be somewhat counter intuitive.
 
The Phase data is really the noise difference between the GPS signal and the OXCO.
At low Tau times of less than the Loop TC (which will not be what it is set to if the Dac gain is off),
the PPS phase noise is mostly due to the noise of the GPS signal, and NOT the OXCO output.
As the TC is reduced the Phase noise reduces because the OCXO noise is increasing to match the noisy GPS signal.
This Shows up on the Dac as an increased noise which is mostly CAUSED by the GPS  when the TC is set TOO short.
The OSC noise on the other hand is the sum of the DAC noise and the GPS noise, 
so it goes up if the TC is set too fast and also goes up if the TC is set too slow. 
One problem is that high freq noise on the OSC plot mostly hides what you really need to see to make that data more useful.

with Increased TC, If the PPS Phase is moving around more at Higher Freq this is a GOOD thing, 
It means the OXCO is doing the right thing and NOT following the noisy GPS signal. 
On the other hand if the PPS Phase is moving around at what is a slow low freq,  DC like offset, 
while not necessary a bad thing, but it does mean that that the OCXO 
is trying to changing freq due to aging &/or external effects like temperature.

And the answer of how to minimize the 'pain inflicted to the unwary innocent':

If you're not a nut & only care about freq accuracy in the e-9 to e-10 range
none of this matters, it's better than that most of the time if you do nothing.
OR
If the Tbolt is not exposed to the outdoor environment 
and it is not being tilted and moved around 
and has had a few days to run in and settle 
and has a OCXO which is not broken or just a non oven Osc, 
and the +12 PS is stable
THEN:
First set the Tbolt's Dac gain to the correct value, 
and then set its TC to the 300 to 500 range, 
This can give significantly lower Osc and PPS Phase noise errors than 
the default 100 setting at short time periods of 100 sec and below.

ws 

*********************
>I must admit that I've generally tended to leave these as "they come out of 
> the box", perhaps it's time to be a bit more adventurous.
> 
>None of the five Thunderbolts that I've had on test at one time  or another 
>have shown what I would consider very significant short term  frequency 
>excursions once settled, just a few parts in 10^-11 using a 1 second  gate time, 
>if there was any noticeable improvement on that I'd be  hitting the 
>unprocessed limits of my test gear anyway.
> 
>Lady H certainly shows a more dramatic picture though, perhaps  another 
>ploy for extending her influence over the unwary and inflicting yet  more pain 
>on the innocent:-)
> 
>regards
>Nigel
>GM8PZR
*****************
>>In a message dated 12/10/2009 10:02:14 GMT Daylight Time,  
>>david.partridge at dsl.pipex.com writes:
>>
>>If you change the scale factor on the oscillator graph to about 100ppt, the
>>pps to 10nS and the DAC to 50uv it isn't so scary :-)
>>
>>My oscillator trace is varying about =/- one fifth of a division in the
>>short term.   My time constant is set at 400 seconds which seems to have
>>considerably reduced the size of the frequency excursions compared to the
>>default, and also reduced the large excursions of the DAC voltage.
>>
>>The only downside of the longer time constant seems to be that the PPS seems
>>to move around more, but as my primary concern is to use this as a frequency
>>standard, that's not an issue for me.
>>
>>Dave
>>
>


More information about the time-nuts mailing list