Accuracy of NIST tones over SW (Cross-posted)

[daniokeeper]

Hello,

(My first post here, so please take it easy on the noob )

Can anyone here tell me just how accurate the tones broadcast by NIST over SW are when received by a SW radio?

Would they be at least at +/- 0.01 cents?

Would they be more accurate than listening to them over the phone?

I'm a piano tuner and I would like the most accurate reference pitch possible to calibrate some equipment. I started a discussion on this over at the PianoWorld Forum and there are two differing opinions. One is that the pitch transmitted is the one received. The other is that due to the signal reflecting back and forth off a varying ionosphere that the signal may be subject to the doppler effect, decreasing its accuracy.

NIST over shortwave as a reference - Piano World Piano & Digital Piano Forums

I've done a little Googling on this, but I haven't been able to find exactly this topic discussed.

Thanks,
-Joe G.

 

[C2]

The difference between 1000.000 Hz and 1000.005 Hz is about 0.01 cents.

The difference between middle C at 278.4375 Hz and 278.4391 Hz is about 0.01 cents.

Those are some pretty small differences at low frequencies.

The accuracy of the NIST broadcast should be received at about 0.01 Hz accuracy, and at 500 and 600 Hz, that is 0.035 and 0.029 cents, respectively.

Propagation and your equipment could have a profound affect on that accuracy.

The best accuracy is received from the GPS system, 0.001 Hz.

That is for absolute time. You can use that to synchronize a high frequency oscillator for extremely accurate timing.

 

[daniokeeper]

Thank you for your time and for sharing your knowledge, C2.

I assume it.s OK for me to post a link here from the thread at PW.

Thanks,
-Joe

 

[Wa8kwm]

...The accuracy of the NIST broadcast should be received at about 0.01 Hz accuracy, and at 500 and 600 Hz, that is 0.035 and 0.029 cents, respectively.

I don't know where you came up with the figure of 0.01 Hz. Given that the tones are present for 45 seconds, and assuming that with a good signal the phase of the audio tone can be detected to within 10 degrees of a cycle, that is 22500 cycles at 500 Hz, plus or minus 1/36 of a cycle. That gives 500 Hz plus or minus 0.00062 Hz. Of greater concern is the effect of propagation delay variations, which according to one source is 0.2 msec between Colorado and Delaware. This translates to 0.1 cycles at 500 Hz. When added to the 1/36 = 0.0278 cycle mentioned above gives a total phase uncertainty of 0.1278 of a cycle. At 500 Hz over 45 seconds that comes to 0.0028 Hz, which is quite a bit better than the 0.01 Hz that you quoted.

 

[C2]

Where's Beetle when you need him?

Well, I shouldn’t have said the accuracy of the received signal. The WWV/WWVB transmitted frequency uncertainty is 1x10^-13 over 1 day. That is the standard and is nearly identical to the global reference standard of 2x10^-13/day.

The received frequency can be rather unstable, depending on several factors, stability of propagation delay being significant. How is one going to generalize these affects? Moreover, how is one going to use the received standard to perform a calibration?

Audio frequency calibrations typically have uncertainties of parts in 10^-4 (0.05 Hz @ 500 Hz, or 0.173 cents)...,

G. Nelson, M. Lombardi, D. Okayama, NIST Time and Frequency Radio Stations: WWV, WWVH, and WWVB, NIST Special Publication 250-67, Time and Frequency Division Physics Laboratory, January 2005, pp 124 & 137.
(available at: http://www.nist.gov/calibrations/upload/sp250-67.pdf)


...but the oscilloscope pattern drift method has been shown to be capable of measuring the audio frequency tones (such as 500 Hz) with an uncertainty of 1 × 10^-6 (0.0005 Hz @ 500 Hz, or 0.00173 cents) under ideal conditions.

N. Hironaka and C. Trembath, The Use of National Bureau of Standards High Frequency Broadcasts
for Time and Frequency Calibrations, Nat. Bur. Stand. (U.S.) Tech. Note 668, May 1975, page 15.
(available at: http://tf.nist.gov/general/pdf/453.pdf)


I assume that the OP is using something like TuneLab, a computer based frequency comparison software, or another similar method, so his measurement uncertainty might fall somewhere between the two. Where depends on his measurement technique. I really do not believe that he is using a calibrated laboratory oscilloscope to tune his piano.

I also think the question is related to the SW broadcast vs. the phone signal. I think the differences will be inconsequential from a practical standpoint.

 

[Wa8kwm]

"audio frequency calibrations typically have uncertainties of parts in 10^-4 (0.05 Hz @ 500 Hz, or 0.173 cents)...,

G. Nelson, M. Lombardi, D. Okayama, NIST Time and Frequency Radio Stations: WWV, WWVH, and WWVB, NIST Special Publication 250-67, Time and Frequency Division Physics Laboratory, January 2005, pp 124 & 137.
(available at: http://www.nist.gov/calibrations/upload/sp250-67.pdf)"

...but the oscilloscope pattern drift method has been shown to be capable of measuring the audio frequency tones (such as 500 Hz) with an uncertainty of 1 × 10^-6 (0.0005 Hz @ 500 Hz, or 0.00173 cents) under ideal conditions.

N. Hironaka and C. Trembath, The Use of National Bureau of Standards High Frequency Broadcasts
for Time and Frequency Calibrations, Nat. Bur. Stand. (U.S.) Tech. Note 668, May 1975, page 15.
(available at: http://tf.nist.gov/general/pdf/453.pdf)

OK, I see where you got your figure. And it does depend on how the calibration is done. Theoretically you could do even better by projecting over the discontinuities when the 500 Hz tone is interrupted and form an averaging period much longer than 45 seconds. That assumes that the 500 Hz is maintained phase synchronous at WWV during the interruption, and that the lock you achieve in the first 45 seconds is good enough so that less than one cycle of the 500 Hz is lost during a typical 1:15 interruption. Of course no one does that, so we might as well confine ourselves to optimal measurements that can be taken in a single 45 second session.

I assume that the OP is using something like TuneLab, a computer based frequency comparison software, or another similar method, so his measurement uncertainty might fall somewhere between the two. Where depends on his measurement technique. I really do not believe that he is using a calibrated laboratory oscilloscope to tune his piano.

I also think the question is related to the SW broadcast vs. the phone signal. I think the differences will be inconsequential from a practical standpoint.

I agree. Piano pitches are not stable enough to even define to better than 0.1 cents in most cases. But being the author of TuneLab, I am well aware of its potential calibration accuracy. It is quite similar to the oscilloscope drift pattern mentioned above in its function. If someone is patient enough to trim the offset in TuneLab so the the pattern is stable for one full 45 second session, they will have acquired a 0.01 cent lock on the standard. Of course it may take a number of these 45 second sessions to find the right offset trim.

 

[daniokeeper]

Thank you both so very much for your help with this. So, I need to keep it steady for 45 seconds. Yes, I'm obsessive enough to do this.

Thanks
-Joe