[time-nuts] Frequency comparison

Bob Camp lists at cq.nu
Mon Feb 8 22:28:34 UTC 2010


Hi

With the 1 pps approach, you can spot jumps or shifts of >1,000,000 cycles
worth of 10MHz without major trouble. With a mixer approach spotting a slip
of jump of one cycle can be difficult. 

The difference sounds pretty minor. If you are looking for odd things
happening over a period of many days, it's pretty significant. 

At 1x10^-11 frequency offset:

The mixer goes through one cycle of 10 MHz in ~ 3 hours. 
The counters go through one cycle of 1 pps in ~ 3,000 years. 

If the Rb is at 1x10^-9 the numbers are 100X quicker. 

Since the GPSDO takes > 1 1/2 hours to give an adequately accurate reading
to calibrate 1x10^-11, there's no rushing the process. That's not warm up
time, or lock time on the GPSDO. It's the time you need to watch if for in
order to know it's output frequency to a sufficient level of precision. 

Bob



-----Original Message-----
From: time-nuts-bounces at febo.com [mailto:time-nuts-bounces at febo.com] On
Behalf Of Joop
Sent: Monday, February 08, 2010 4:27 PM
To: Joop
Cc: time-nuts at febo.com
Subject: [time-nuts] Frequency comparison

At 07-02-10, John Ackermann wrote:
>Hi Raj --
>
>You've already gotten some good answers.  If all you want to do is 
>measure the frequency offset rather than characterize the stability, a 
>simple approach is to first get as close as you can by adjust for 
>minimum march of the 10 MHz signals across the oscilloscope, then use 
>either the Racal counter or the digital o'scope to measure the delay 
>between the two signals and how it changes over time, preferably 
>measuring at 1 PPS rather than 10 MHz.
>
>In other words, measure the time difference between the leading edge of 
>the PPS signals, averaging for a while (depending on how close the two 
>already are) to improve resolution and reduce the noise.  Write down the 
>delay figure, note the wall clock time, wait a while, then come back and 
>measure the delay again.  The change in delay over the elapsed time will 
>tell you the frequency offset, e.g., if you see 1 microsecond per day, 
>that's 1.16x10e-11.
>
>Adjust and repeat.  As others have mentioned, being a time-nut requires 
>patience. :-)
>
>It's best to do this at a lower frequency than 10 MHz, and ideally at 1 
>PPS, as there's only 100 nanoseconds between cycle slips at 10 MHz, and 
>that limits how long you'll be able to measure before you've drifted a 
>complete cycle.
>
>73,
>John
>
Ok, the past few days I have been working on exactly the same thing. That
is, adjust an LPRO to my homemade GPSDO.
Good to know I followed a proven procedure. Initially I wanted to build
two PPSDIV/TADD circuits but did not have the right PICs. Instead I used
two PIC12F devices and put them on the same veroboard. The software for
only a single 1Hz output was not too complex.

There was also a Racal 1992 I could use to measure and log the phase
shift. If that would not have been the case I probably would have tried a
XOR port to both (synched) outputs and use an R + C to measure the DC
voltage. That would make a nice little gadget for people not having a
high-res counter. Also the TADD 74AC04 drivers would not be needed this
way.

Anyway, I finally dared to adjust the LPRO.
Can somebody tell if the following seems normal?
* The LPRO measured 1.77 E-10 high (before any adjustment)
* Lamp voltage is about 5.7V
* The GPSDO 1Hz pulse seems to move (noise like) + or - 15ns around its
linear regression line
* The LPRO takes more than 8 hours (perhaps even 24) to reach the 1E-11
level?
The last point is guessing since my GPS signal is super stable. I thought
I managed to adjust it to that level after the LPRO had been powered on for
more than 24 hours. But powering it on two days later shows a higher drift
after 5 hours than where left before.

Also I would like to know if I have to repeat this procedure once it is
built into its final enclosure. Right now it is open on the bench and
clamped to a heatsink. Temperature might be different inside a box.

Cheers,
Joop



_______________________________________________
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