DSPRelated.com
Forums

Frequency measurement

Started by SVS June 30, 2006
Dear Members
I want to measure frequency of power line. I am using 16-bit ADC (sampling frequency is 12.8KHz). My question is how to measure frequency with high accuracy (+/ - 8mHz) in presence of Harmonics ( I think it can be done by Appling FFT) . Can any one through some light on this?
Thanks in Advance.
Regards,
SVS

---------------------------------
Do you Yahoo!?
Get on board. You're invited to try the new Yahoo! Mail Beta.
>Dear Members
> I want to measure frequency of power line. I am using 16-bit ADC
>(sampling frequency is 12.8KHz). My question is how to measure frequency
>with high accuracy (+/ - 8mHz) in presence of Harmonics ( I think it can
>be done by Appling FFT) . Can any one through some light on this?

The possibel approaches are
1. Use hardware zero cross detection and take the inverse of timer gap between two interrupts and calibrate.
2. Implement the same in software
3. Use a digital PLL, means keep estimating the phase angle for every sample using a PI control algorithm and use phase incremental between two samples to calculate frequency

BHASKAR REDDY . N
Mysore
Mobile: +91 9880 177365
Sita-

> I want to measure frequency of power line. I am using 16-bit ADC
> (sampling frequency is 12.8KHz). My question is how to measure frequency
> with high accuracy (+/ - 8mHz) in presence of Harmonics ( I think it can
> be done by Appling FFT) . Can any one through some light on this?

To exceed +/- 8 mHz accuracy using an FFT approach, then you would need 4
mHz magnitude bin spacing. This would require a 262,144 pt FFT size at
12.8 kHz sampling rate.

So you're going to need a substantial amount of SDRAM in your C6x design,
at least 2 Mbyte -- maybe 4 Mbyte to be safe.

-Jeff
First you should remove the harmonics and superposed noise on the signal by
applying some kind of low- or band-pass filter. After that, you can measure
the frequency by counting the time between two subsequent zero crosses of
the filtered signal. If you find out that your measurements aren't accurate
enough, try measuring the time between a larger number of zero crosses.

Besides the FFT, another useful tool for spectrum analysis is the Chirp-Z
transform (see http://www.embedded.com/showArticle.jhtml?articleID301593
).

Regards,

Guy Eschemann
Vienna, Austria.
On 7/1/06, Bhaskar Reddy N wrote:
>
> >
> >
> >Dear Members
> > I want to measure frequency of power line. I am using 16-bit ADC
> >(sampling frequency is 12.8KHz). My question is how to measure frequency
> >with high accuracy (+/ - 8mHz) in presence of Harmonics ( I think it can
> >be done by Appling FFT) . Can any one through some light on this?
>
> The possibel approaches are
> 1. Use hardware zero cross detection and take the inverse of timer gap
> between two interrupts and calibrate.
> 2. Implement the same in software
> 3. Use a digital PLL, means keep estimating the phase angle for every
> sample using a PI control algorithm and use phase incremental between two
> samples to calculate frequency
>