Magnetic Field measurements: Weird FFT signals I cannot explain.

[tboh]

Hi all!

First off: I don't know which forum to post this. I'm no grad student or PHD so academic forum seemed wrong, and it's not really just math, since it was the result of a measurment and the signals can come from environment, etc.

Ok. I measured the magnetic field using a sensor (HMC2003 of Honeywell) that gives me an analog output voltage dependent on the magnetic field. I have used a Agilent 34411A Multimeter to save the data on the computer.

Most important: Sampling rate was 1 data per second! 86400 measurements were made. Of the result I got I used the FFT algorithm (traing also different Windows just in case) of Origin 8.6. The result I got is shown in picture below.
http://postimage.org/image/j8jra66kl/

So my problem is basically this distribution of discrete peaks under an envelope. What can produce such a signal in my measurment? Also the number of peaks is different for the two envelopes at 0.2Hz and 0.4Hz and the difference of frequency the same, so the second should be no harmonic wave of the first. I found no signals of that frequency around my measurment location.

So at all those experienced physicists out there: What can cause these signals? Physically? Or some mathematical property?...

 

[lewando]

Could be higher frequencies (above 1Hz, that you are not filtering) ailiasing as a frequency component in your band of interest.

 

[tboh]

I also analysed the spectrum in other frequency ranges. The only real peaks I found for the harmonic waves of 50Hz, the power line frequency here. They should not be visible in a 1Hz sampling frequency... Especially not like that :)

 

[lewando]

Some thoughts:

Maybe it is real. What is the thing that you are measuring? A refrigerator? The Earth? Do the spurs go away when you remove the source (good luck if Earth)?

Maybe the magnitude of these spurs is not significant. What would be the field strength of these spurs (in gauss, not "dB") ?

You should be using a battery to power the sensor.

Too bad you can't power your Agilent with a battery.

Maybe it's the proximity of your PC? Try data logging without being connected to PC.

You should still be using an anti-aliasing low-pass filter on the output of your sensor.

 

[tboh]

I'm measuring the magnetic field in a lab. So basically everything, from Earth to some remaining signals from neighbor-experiments.

The strength of those signals is not so high with 250muGauss. But still its not completely insignificant (high precision experiment) and I don't know where it comes from. Identifying a source would help because I have to discuss the signal.

50Hz Signals from sensor are also there with Battery, I tested.
And I never heard of an anti aliasing low-pass before ^.^ Should it be used in general for measurements like that with sampling rates unter 1kHz?

 

[lewando]

You should do some level of AA filtering, depending on your band of interest and what amount of non-interesting signals are present outside your band of interest. For background read:
http://www.ni.com/white-paper/3016/en#toc4

Try transporting your measuring system to a "quiet" place away from your lab. If you see the same thing, then the source of the mystery spurs is likely your measuring system, itself.

 

[Xitami]

2^(100 / 6) = 104032
(~6 dB per bit in A/D)
noise