[time-nuts] [Solved] Looking for multiple PPS timestamp logging
kevin at rosenberg.net
Tue Oct 4 15:03:56 UTC 2011
On Oct 4, 2011, at 4:36 AM, Hal Murray wrote:
> In California, the diurnal temperature swings are big enough to be useful. :)
Here in New Mexico as well, useful or ruinous depending upon your application.
Hence, I believe, your smiley face above ;)
> The swing would be much bigger outside, especially with an open sky.
We thought about an outdoor light and temperature sensor when my son
considered measuring the stability of the PPS signal from a NetClock/2
WWVB receiver. But, we'll be measuring the temperature near the indoor XO.
> I was thinking of keeping track of when you expected each pulse to arrive
> next and sleeping until a little before you expected the soonest one.
> (Adjust "a little" by trial and error.)
Sure, and the solution becomes more interesting the more pulse lines that
are being polled.
> I agree that Tom's picPET will be a fine way to do it. I was just playing
> with how to do it with the parallel port.
I lot of fine ideas come from thought experiments.
> It's still an interesting question of how accurate you need to measure the
> time. That's probably a good one to work through with your son on the back
> of an envelope.
I think so, too. So, my son will be learning more about stability, Allan
deviations, modified Allan deviations (which should be more useful for
the number of averaging points over a month), and improving his exponent
Another statistic he'll be learning is the Pearson correlation coefficient
to compare temperature (probably in Kelvin) to the XO frequency. The more
precisely those are measured, the more likely the Pearson will reflect the
> How are you measuring temperature?
Good question, because we haven't completely solved that. We'll be using
a LM35CAH mounted next to the 32768 crystal and measuring the
voltage once a second.
What I haven't decided is whether to have my son use the 34401A and GPIB
polling via a Prologix adapter versus a simple MCU firmware using a 2.5V
reference voltage and a 10-12 bit ADC and outputs the ADC result either
once a second, or in response to a pulse (probably from the PPS of the XO).
The first has merits of accuracy and simplicity, but I prefer he use
more affordable devices than the 34401A in his experiment to even his
project compared with other students. The latter has the advantage of
price, but the disadvantage of me writing the firmware (though, the
firmware is almost trivial). But again, I'd like him to be as
independent as possible.
So, if you know of any simple ADC to UART firmwares available, that'd
be great so he can just reference someone else's code. The picPET
is a perfect device at the perfect time. But, we can't rely on tvb
to come up with a 'picADC' at a similarly serendipitous time.
More information about the time-nuts