[time-nuts] Line Frequency

M. Simon msimon6808 at yahoo.com
Mon Feb 10 03:29:45 EST 2014


Chris, 

What you are suggesting is different from what I'm doing. Sampling is different from time stamping. I'm looking for events. You are looking for "data". You want a PC. I want a small micro (specifically the LPC1114). You want to overpower the problem. I want to finesse it. Nothing wrong with either. It just depends on what you are trying to accomplish. 

My desired accuracy  is based on what I have seen displayed on line frequency monitoring sites on the 'net. Numbers like 60.053 etc. That says (roughly) 1E-5 accuracy. Now it may be that they show more precision than is warranted. Or they are averaging over a number of cycles. 


Having thought about this for quite a while I think synchronizing a free running clock to the line may be the best that can be done. And the accuracy per cycle is probably no better that 1E-4 - maybe worse. Line noise and varying line voltage make this a wicked problem. But then again any time you try for a lot of accuracy the problems get wicked. 


My backup plan is the H bridge suggested earlier in the discussion. The size of the capacitor can be adjusted for the amount of noise rejection desired. And I must say that it is an elegant solution and the parts count is small. With a 10 mV offset op amp (comparator) and an input voltage of  20 V peak (roughly 14VAC) and a 10 us RC, the timing should be quite consistent. 

I will report back once I have some results. Which may take a while. 

Simon

Engineering is the art of making what you want from what you can get at a profit.



On Monday, February 10, 2014 4:02 AM, Chris Albertson <albertson.chris at gmail.com> wrote:
 

>
>
>
>
>On Sun, Feb 9, 2014 at 4:27 PM, M. Simon <msimon6808 at yahoo.com> wrote:
>
>Tom,
>>
>>I was hoping for  1E-5 precision or better. My time stamping counter will have a 30 MHz clock (for convenience). 
>
>
>I think noise will not allow that level of accuracy with only one time stamp per cycle.   I think you'd be better off making many samples per cycle.  Taking 48,000 samples per second give 800 samples per cycle.  48K is the "standard" sample rate for audio used with video.    Likely your computer already has the means to sample two audio signals at 48K samples per second.   It is not taxing on any modern computer or even for a smart phone.
>
>
>Doing this on my iMac is so trivial I don't bother to save the setup.  I have an audio interface to samples 24 bits at 96K SPS.  I placed a 1 volt peak to peak signal on it and then brought up a spectra display.  I can log the spectra to a file.  This kind of software is available for free.
>Chris Albertson
>Redondo Beach, California 
>
>


More information about the time-nuts mailing list