I asked about this not long ago on the time-nuts@... mailing
list.
Most receivers have a fixed frequency oscillator driving the
numerically control oscillators so the 1pps output has significant
sawtooth jitter because of limited timing resolution. The better
timing receivers that work this way report the difference between the
1pps output and the calculated 1pps value so external correction can
be applied.
The Thunderbolt design is different in that the 10 MHz reference
frequency itself that drives the numerically controlled oscillators is
phase locked to GPS time so the PPS output only has random jitter
error.
On Mon, 15 Oct 2012 17:38:38 -0400, "Tom Miller"
<tmiller11147@...> wrote:
Usually the 1 pps comes from the 10 MHz (or some higher reference) but they are both common.
Tom
----- Original Message -----
From: Geoff Blake
To: TekScopes@...
Sent: Monday, October 15, 2012 5:19 PM
Subject: Re: [TekScopes] Re: How to calibrate a frequency counter (on the cheap)
If I understand correctly, the 20nS spec. of the 1ppS signal is
relative to time, not frequency, in other words, it means that the 1
ppS pulse is within 20nS of GPS (I think) time once the delays are
taken out. It is rather like saying your clock is one minute slow, but
looses only one second a week. The accuracy of the 1 ppS period is the
same(ish) as the 10MHz, i.e. ~ parts in 10-^12 from which it is
derived, and is quite suitable to look at Rb sources.
At least, I think so.
Geoff
On 15 October 2012 22:02, Patrick Wong <patwong3@...> wrote:
>
>
> After seeing this post, I looked into the specs of various GPS receivers.
>
> I note that the Thunderbolt is rated at 1 part in 10^12 accuracy with respect to the 10 MHz signal that it produces - can't beat that unless you have your own cesium standard. Apparently this requires the Thunderbolt to do some averaging over a day, to reach that accuracy.
>
> However the 1 pps signal is rated at much lower accuracy, within 20 ns. I believe that this translate to an error of up to 0.2 Hz at 10 MHz. Is my analysis correct?
>
> Therefore, it seems to me that using the 1 pps signal as a real-time reference to adjust a rubidium oscillator is not helpful unless you know the Rb oscillator was off by more than 0.2 Hz to begin with.
>
> I recently bought an HP 5335A which can read out frequency up to 1.3+ GHz to 11 digits, and this has caused me to become interested in an effective way to reference my rubidium frequency standard as well as the built in 10 MHz standard within the HP 5335A. Right now the difference between the two is around 0.01 Hz which reflects ~1 part in 10^9...
>
> I guess I also will look for a Trimble Thunderbolt or equivalent.
>
> Patrick Wong AK6C
>
> --- In TekScopes@..., "denyhstk" <denyhstk@...> wrote:
>>
>> About 2 years ago, I picked up a Rubidium oscillator on ebay for around $60. Put it in a box with power and convection ventilation. Picked up a Garmin miniature hockey puck sized GPS receiver for $59 (I think it was either direct or an Amazon purchase). Only problem is the Garmin put out a 1pps signal and I was not in the mood to get fancy with this.
>>
>> I ran the 10MHz Rubidium into an HP 3336B and generated a much lower frequency at the output of the synthesizer. Both the lower frequency (can't recall, maybe 10kHz) and 1PPS were run into a scope, triggered on one and the timebase adjusted for a shorter and shorter interval as my adjustments of the Rubidium got closer.
>>
>> After a few hours, I think I had it between 10^-9 and 10^-10 (an estimate by observing the rate of drift of the two).
>>
>> At that point I called it a day and haven't checked since.
>>
>> Den
>
>
>
>
> ------------------------------------
>
> Yahoo! Groups Links
>
>
>
--
#################################################
Geoff Blake, G8GNZ JO01fq: Chelmsford, Essex, UK
<geoff@...> or <melecerties@...>
Using Linux: Ubuntu 11.04 on Intel or Debian on UltraSparc
and even on the NAS. Avoiding Micro$oft like the plague.
#################################################
David Hess