Signal detection with cross correlation

[Elargus]

Hi
I am working on a project that consist of measuring the spectral noise density of a resistor and I need some help with that.
Presently this is how I've been going about it.
1- I have the voltage noise out the resistor going through two low noise preamplifier in parallel (gain of 10000 with a bandwidth from 1hz to 300khz).
2- take the output of the low noise preamplifiers and put it into two differential analog inputs of a PCI 6250

3-I used daqassistant to create the two channels. the sampling rate is set to 1MS/S and I take 10000 samples
4- From labview 7.1, I use the crosscorrelation express VI and get the cross correlation of these two inputs. This should enables me to get rid of the noise signalfrom the low noise amplifiers since they are uncorrelated and leave me with only the square of the noise signal from the resistor
5- then FFT the output of that to get the spectral noise of the resistor.

Now this is the problem I am having; the resulting spectral noise is not correct; I know this because knowing the resistance of the resistor I can compute the thermal noise out it and it should be: 4KTR(A^2) where A is the gain of my preamplifer. From the cross correlation with a resistor of 10K I get a noise signal around 0.01 mean while I am expecting  1.66e-8. I am therefore trying to track down where this error is comming from. I noticed after reading some threads on crosscorrelation that its output is normalised. Could this be my problem? Is there any way to solve it?
Sorry for being this long but I thought I should give as much details as possible.
Thanks

[Kolos]

Hi Elargus!

Im working now in a project, whats a bit similar to this one... Can you post/send your vi?

Thanks:Kolos 

[mark@diagnosticbiosensors.com]

Kolos and Elargus,

I am also trying to develop a test station for measuring small changes in resistance.  I would like to calibrate my system by injecting a known physical noise source like a 10kOhm resistor. 

I'm trying to simulate the data handling in LabView, but I'm having trouble finding out what the units are when using "white noise" for example.

Mark

[Kolos]

Hello!

I don't know the white noise thing...

[Elargus]

Hi Kolo,

I am sorry I saw your message late... I finished that project long time ago and i hope I find the vi. 

[Elargus]

Hey mark,

i think it depends on the kind of operation you are doing in your cross correlation. In my case the unit was Voltage^2/Hertz since the noise from a resitor is S=4KTR(A^2)

[Kolos]

Did you find it(your vi)?

Can you tell me, what formula did you use?

Did you used something like this:

 ?

Thanks: Koli 

[Elargus]

Yes that is the formulae i used with the difference that the sommation is from 0 to infinity. i found one of the first Vi i built for this project it is still very rudimentary but the basic idea is laid out. Here I used the express Vi Convolution and cross correlation. The result of the cross correlation is given in the time domain so it is necessary to carry out an FFT afterward. If my memory serves me correctly, it is important to extract a portion of the final signal to stay away for example from regions where you have undesired noise signals such as 1/f noise and there things get even more complicated. In case you need that too, let me know, I will look again and see if I can find the Vi where this feature was added. In any case le tme know how it goes.

Alex

[vel725]

Am going to do my Post graduate project regarding Flow measurement using cross correlation ultrasonic transreceivers… In my project, i have to find delay between the two signals from pair of ultrasonic receivers using cross correlation method.

I need some guidance of You all… could u please help me…..