[time-nuts] 1 pps correction

Said Jackson saidjack at aol.com
Mon Apr 2 03:06:57 UTC 2012


Hi Azelio,

its a dual slope interpolator, much like the HP 5334A counter. In fact the 5334A service manual is very nice to go through to get lectured on how this works. The capture hardware is similar to the Linear Tech app note written by Jim Williams (mentioned in the time nuts archives). Basically its a very fast constant current source, and a high quality capacitor. Except Jim charges the cap, then uses an analog to digital converter to capture the time difference. We use a micro controller to capture the time difference on the cap, then capture how long it takes to discharge the cap with about ~1000x slower current than the charge current. Hence we get ~1000x to 1 time dilution, which means the underlying 16.66ns counter resolution becomes a ~16.7ps resolution. While I have never seen the PRS-10 Rubidium schematics (anyone have them in PDF format?) I gather from the description in the service manual that they do something similar to this. The Wavecrest DTS user manuals floating around on the internet also explain how this works. So in short, all that is required to build a unit like this is a bunch of fast analog charge hardware, and an analog comparator that can trigger a counter capture event, and some software for calibration and control...

bye,
Said  

Sent from my iPad

On Apr 1, 2012, at 4:07, Azelio Boriani <azelio.boriani at screen.it> wrote:

> Said,
> how complex is your 20pS time interval counter? Is it analog, FPGA,
> something else (if you can disclose some info, of course)?
> 
> On Sat, Mar 31, 2012 at 11:44 PM, Hal Murray <hmurray at megapathdsl.net>wrote:
> 
>> 
>> martyn at ptsyst.com said:
>>> I’ve seen that the peak to peak jitter is reduced from something like 27
>> ns
>>> to < 10 ns.
>> 
>>> Is this a reduction of just the jitter, or is the actual accuracy to UTC
>>> also improved by this amount.
>> 
>> Have you read the hanging-bridges paper?
>> Tom Clark and Rick Hambly: Timing for VLBI
>> http://gpstime.com/files/tow-time2009.pdf
>> I think that is the key to understanding this area.
>> 
>> If you could average over many sawtooth cycles, you should get an accurate
>> answer.
>> 
>> The problem is that you don't get to pick how many cycles fit into your
>> averaging time.  The sawtooth pattern is the beat between two frequencies.
>> One of them is drifting with time/temperature.  If you are unlucky, the
>> beat
>> frequency can be very very low.
>> 
>> The sawtooth correction lets you correct on a cycle-by-cycle basis.  You
>> don't need to average over many samples.
>> 
>> 
>> --
>> These are my opinions, not necessarily my employer's.  I hate spam.
>> 
>> 
>> 
>> 
>> 
>> _______________________________________________
>> time-nuts mailing list -- time-nuts at febo.com
>> To unsubscribe, go to
>> https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
>> and follow the instructions there.
>> 
> _______________________________________________
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.



More information about the time-nuts mailing list