changing phase angle

[Haroon2k9]

Hi all,

        I have recorded a same signal several times for two minutes and trying to perform the spectral analysis. If I take any two data samples, as far as the waveform is concerned the wave is very similar to each other for both the samples as can bee seen from attached screenshot (waveform). Same is true for amplitude spectrum. But when it comes to phase spectrum the data between different samples is not same as can be seen for a frequency of 2150 in the screenshot attached.

       Why is there the difference in phase spectrum only for different data samples while waveform and amplitude spectrum is same which the case should be since the same signal is recorded at different intervals???????

Anxiously waiting for the response.

Thanks and best regards,

Haroon

[JÞB]

To me it looks like there isn't much differance at 2150.  (-180 deg = + 180 deg)

The rest of the changes seem to be related to the trigger resolution.  It appears that your triggering on a 50Hz fundamental frequency compoment.  comparing the phase relationship of the 86th harmonic (2150Hz) requires a very precise trigger scheme and a very high sample rate.

I'm assuming this is related to youe appliance current measurement experiment.  What Hardware did you decide on?

Message Edited by Jeff Bohrer on 05-26-2010 09:51 AM

[Haroon2k9]

Hi there. Thanks for the reponse. yes you are right this is related to my appliance current waveform experiment. I am using NI USB 6259 and as a transducer I am using Fluke i200s current clamp. I am keeping a sampling rate of 100 kHz. Well as far as -180 and 180 is concerned it may be okay but please refer to the changes in my attached screenshot. how can I avoid these small changes of around 30 degrees????? I am trying to use digital filter as well. But i am not sure, sorry for my ignorance, which filter to use.... chebyshev or butterworth and about the order required??? I only know I can use a band pass filter which will allow 5o Hz to 2450 Hz (50th Harmonic) but not sure how to use it and what impacts it may have on my results 😞

I am working around this phase issue for almost 2 months now but still clueless. Your help will be very helpful.

Best regards,

Haroon

[JÞB]

Don't appologize for being ignorant or unexperienced!  your learning and asking questions!  I grew up as an electronics technician and have worked in electronics every day for over 25 years (I'm getting old) and I STILL learn new stuff every day!

OK so you have a few misconceptions:

1) MYTH: the power company provides a stable regulated voltage and frequency with no distortions.  REALITY: the power company has no interest in providing a stable pure 50 Hz. (in fact, it would be nearly impossible) They do attempt to regulate frequency and voltage to 48-52Hz at 220 - 240 VAC but, surges, spikes and failures occur often.  Additionally, few -if any- line powered devices fail to work properly if the power source is within 10% of "ideal".

2)MYTH: the Fluke i200s current clamp introduces 0 phase distortion.  REALITY: The phase distortion is unspecified in the literature (Follow the link.)  while the clamp is "usable" up to 10KHz if you used controlled sources and graphed out the accuracy in amplitude, and spectral purity vs input frequency you'ld see an interesting curve with several non-linear components. 

3)MYTH: there is no "noise" on the line.  REALITY: TV, Radio, Local radar systems, cell phones and microwave relay towers all broadcast RF that can induce currents on the transmission and distrubution lines (not to mention solar events).  Additionally PLC technologies actually intentionally introduce modulation of the line frequency.

so in conclusion: I'm not sure there should be any spectral relationship between random samples of line currents from a residential or comercial power provider near 2KHz.  I would suggest you limit your experiment to the first 3-5 harmonics as there seems to be no rationale to compare the phase relations way outside of the 50hz that is of interest.

Perhaps you do need to gain information about these higher freq components.  If so you will need some type of AC-AC converter with much tighter regulation than the power company provides.

[johnsold]

Haroon,

In addition to Jeff's excellent comments think about a few other matters:

What is the amplitude of the 43 or 49th harmonic?  If those amplitudes are very small it may not be possible to get an accurate measurement of the phase.  You do not indicate how you are determining the phase.  The standard spectrum VIs all use Fourier transform techniques.  I have not investigated their phase accuracy with small signals but I would not be surprised to find that you are near the limit of the algorithm for harmonic amplitudes 40-50 dB below the fundamental.

The sampling rate of 100 kS/s gives a phase resolution at 2450 Hz of ~9 degrees.  Is the phase calculated relative to the fundamental or the desired harmonic?

What does the phase mean to your experiment?  If it is this hard to measure or calculate, do you really need it? 

Lynn 

[Haroon2k9]

Hi Jeff,

         Thanks for the encouraging words. I agree with all your points. Suprisingly didnt pay much attention to the literatire of the current clamp for which error increases with the increasing frequency. So thnx for bringing into my notice. 

Now about conclusion if i consider only first 5 harmonics, the total harmonic distortion(in current) for my load is only 173 % and if i consider harmonics upto 50 it jumps to 260 % so don't want to neglect higher harmonics 😞 one option I am considering is to take into consideration phase angle for first 5 harmonics (which are somewhat consistent in a ll readings) and assume it to be 0 degrees for the higher harmonics because of lower accuracy. What is ur suggestion about it???

Best regards,

haroon

[Haroon2k9]

Hi Lynn,

          Thanks for the response. I am taking the FFT using the linear scale rather than dB. Although the amplitude is only around 5 % of fundamental for 43rd harmonic. But the problem is phase is not consistant even for the harmonics which have an amplitude of around 70 % of fundamental, may be because of high frequency which increases in the error of transduscer.

I am really sorry, i really don't know what phase means for me 😞 , i am just giving a current signal to spectral analysis VI to get the amp and phase spectrum.

If i dont measure the phase spectrum, the final simulation in my project will have an error of around 11 % which I don't want.

Your suggestions will be very helpful.

Best regards,

Haroon

[JÞB]

A better measurement would be THD+N

The clamp range is 0.1A-24A which leaves you with an approximate 24dB dynamic range and a noise floor of about 100mA.

in any case treat all compontents that are less than 100mA as noise and use the formula:

THD+N= Irms (2) +Irms (3) +Irms (4)...Irms (5) + NOISE/Irms Fundamental

This should provide a fair treatment of the real fidelity of the sample.

Message Edited by Jeff Bohrer on 05-27-2010 03:32 PM

[Haroon2k9]

thnx for the reply. may i please know, how did u estimate the dynamic range and noise floor and what is the information i get from this 😞 ???  What if the load is only of 10 W working at 230 V??? the current will be pretty less than 100 mA.

Best regards,

Haroon

[JÞB]

---

Haroon2k9 wrote:

thnx for the reply. may i please know, how did u estimate the dynamic range

The current clamp manual lists that the 24 amp range has 0.1 to 24A limits:10 log (24 /0.1 ) = 23 and change

and noise floor:

the miniumum 100mA

and what is the information i get from this 😞 ???  What if the load is only of 10 W working at 230 V??? the current will be pretty less than 100 mA.:

You would need more sensitive equipment

Best regards,

Haroon

---

Glad to help! but please use whole words.  we are triing to keep this a readable forum, even for those posters who don't speak txt

Message Edited by Jeff Bohrer on 05-27-2010 07:36 PM

Message Edited by Jeff Bohrer on 05-27-2010 07:37 PM