[time-nuts] Measuring speed of light or reproducing a metre
Hal Murray
hmurray at megapathdsl.net
Mon Jun 24 02:48:17 UTC 2013
jim77742 at gmail.com said:
> My actual application is as a quick cool demo showing what I can do with
> this gear in my garage when people go "why"?
What's the geekiness level of your target audience?
I'd probably start with a scope. I think that would be easier to explain to
not-quite-geeks. Look at the pulse coming out of your GPSDO and also after
going through 100 feet of coax.
For geeks, you could measure the delay through 100 feet of coax, then compute
the velocity (assuming the length is correct) and see if that matched the
specs for the velocity of the coax.
Assuming your scope cooperates, you could also look at the next PPS and use
that to calibrate the osc/clock in your scope. (Last time I checked, mine
was off by 6 ppm.)
A 5370B is (much) more accurate at measuring the delay than a scope. But it
doesn't give you the pretty picture and may be way off if your signal has a
slow rise time and the trigger levels are not right.
It might be fun to analyze the possible errors in that sort of setup and
compare them to state-of-the-art measuring setups.
----------
You could setup a 5370B to measure the frequency of a typical low cost
oscillator package. Step one is to show that it doesn't quite match the
number printed on the can. Step two is to put your thumb on the can to
change the temperature a bit and watch the frequency change.
It would be fun to put a knob on the power supply voltage so you can see that
change too. That probably needs a voltmeter.
-------
Maybe you should keep a clipboard on the wall so you can add a note
describing why you used a piece of gear each time you turn it on. Better
would be when you turn it off. Then you know what you really used it for
rather than what you thought you were going to use it for and/or if it did
what you wanted it to do.
----------
I started working at Xerox late in 1976. Shortly after I got there, Ed Taft
released a new version of the Alto OS that significantly improved timekeeping.
The machine was designed with a 170ns cycle time. Crystals come in MHz
rather than ns so they were built with 5.88 MHz oscillator packages. That's
170.068 ns, off by 400 ppm. The initial software used 170 ns. The fix was a
minor tweak to that constant.
--
These are my opinions. I hate spam.
More information about the Time-nuts_lists.febo.com
mailing list