[time-nuts] A silly question ...

[Tom Curlee]

This is assuming that your oscilloscope is set at 100 nS/DIV.  My Tek 2465A will sweep at 5 nS/DIV normally, and 500 pS/DIV when the sweep is set to X10.  If I'm figuring correctly, this will allow 1e-12 in 432 seconds.  I use a stop watch to time the zero crossings of the sine wave - something like:

Assuming the wave moved 5 divisions in 185 seconds:   (500 pS * 5 DIV)/185 seconds = 13.5e-12 ppm.    (I think this is correct.  tvb?)

Since a 10 MHz sine wave tends to look like a flat line at 500 pS/DIV, I often set the vertical V/DIV to 2 mV/DIV.

Tom




      From: Tom Van Baak <tvb at LeapSecond.com>
 To: Discussion of precise time and frequency measurement <time-nuts at lists.febo.com> 
 Sent: Thursday, September 27, 2018 11:19 AM
 Subject: Re: [time-nuts] A silly question ...
   
> "I think", that if for example, it takes 1 second to drift one cycle,
> that works out at 0.1 ppm. If it takes 2 seconds, it's 0.05 ppm, if it
> takes 5 seconds, it's 0.02 ppm etc. Is that correct?

Yes. At 10 MHz one full cycle is 100 ns. So if the cycles are drifting by 100 ns per second that's 100e-9 s / 1 s = 1e-7 = 0.1 ppm.

At these levels of frequency accuracy, using a 'scope is plenty good enough. In fact, it's more educational and somehow more enjoyable to watch analog sinewaves drift past each other than it is to see the digital display of boring frequency counter.

Where the 'scope method starts to break down is when the frequency error gets down to the ppb level. At 1e-9 it will take 100 s for the waveforms to drift by one cycle. And at 1e-12 you would have to wait an entire day (100 ns / 86400 s = 1.157e-12).

On the other hand, with frequency offsets this low you don't have to sit there the whole time. One trick would be to take a photo of the 'scope once an hour, or, say, once every 1000 s. If you played that back at 1 fps you'd have a 1000x "time magnifier".

/tvb

_______________________________________________
time-nuts mailing list -- time-nuts at lists.febo.com
To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.