How to reproduce this signal ?

[Geoff54]

Hi everyone,

I'm trying to develop a shaker's installation. I have realised the sine excitation and now i'm working on the random vibrations. Initially, the shaker installation works without Labview. The  original device produce this type of signal which 

And for my Part, with my daq device ( NI USB 6211 )  i generate this type of signal :

As you can see these two types of signal are different even if they have the same RMS amp and frequency range ( 10-200 hz )  and i want to know why.. I'm attaching my vi at the bottom of this post.

Sorry for my bad english if you don't understand something I will try to reformulate my request.

Best regards

[Bob_Schor]

There are many ways to produce pseudo-random movements.  It depends (a lot) on what you want to accomplish, and whether you want "random" or "unpredictable", and also whether you are producing it using analog or digital outputs.

From your images (making some assumptions about time bases, which I couldn't discern), I'd say the main difference between your waveforms is the spectrum -- the second one seems to have a lot more of the higher frequencies than the first.

What is it that you are shaking?  What is the purpose of the random waveform?  What are the relevant parameters (such as the frequency with which the output points are generated, and the upper and lower frequencies desired in the stimulus waveform)?  Do you want something that is truly "random", or is "unpredictable" good enough?  [I've successfully used a sum-of-sines waveform, which is easy to generate and is unpredictable, as a physiological stimulus].

Bob Schor

[Geoff54]

Thank you for your answer, 

My goal is to  reproduce a random signal for testing on embedded electronics like radio or GPS . So, i have to define a profile and then produce a random signal which matches the profile.

My program works like that :

1°)

2°) Once defined my spectrum, I measure its spectral density and selects said vibration "reference". I also determines the high and low limits with the variable "bound" that will increase or decrease the accuracy of our system (by increasing or decreasing the difference between the reference and the upper and lower limits)

3°) Generate a uniform white noise with my analog output to put across my analog output according to the gain of the amplifier.

4°) I click later on start to go to the window of "vibration control", my reference and my boundaries are drawn.

5°) Subsequently generating the PSD response which must be adjusted on the reference curve with the amplifier gain ( White curve ) 

6°) And after you clik on " enable control " and show driing PSD to generate final random signal ... ( yellow curve ) 

I let you my project below. if you want to test my VI you have to run " Main vi " and " Sous main " at the same time before starting parametrage. You just need one AO link to one AI.

Now here has the power spectrum of the original installation that I need to reproduce

For the same profile :

and the Spectral Density :

BEST regards

[Bob_Schor]

Very nice -- you seem to have the methodology that you need well understood and well developed.  I took a look at the Zip file you attached, but there are a number of missing VIs (from CLE USB NOIR\PROGRAMMES\Sine et Random ( CRIO ) and Variable globale.vi, so I didn't try to run it.

What is the problem that you are having?  From your description of what the program should do, I would think that it should work.  Is there a particular step that seems to fail?

Bob Schor

[Geoff54]

Excuse me, i have forget the main part of my Project. I'll send you the add tomorrow morning ! I think I'm on the right track, but the problem is at the level of the analog output. Indeed, the signal is hatched and I do not know why. With the other part of the program that certainly appear lighter! When I generated my orders on the shaker vibrates it but I hear a kind of resonance and my accelerometer sends me a less accurate signal than the original installation.
I show you my problem tomorrow if you do not mind !

Greetings

[Geoff54]

Hello again,

here is additional vi. I think you can run the program now. tell me if you need another Vi.

Best regards

[Geoff54]

I measure the power spectrum during the phase of "response" that here.

And look at the spectrum of the signal :

As you can see i have a difference between original signal and labview signal definitely the frequency range.

[Henrik_Volkers]

some thougths about your setup:

What you send to the amplifier AND what is generated by the shaker AND what is measured by the accelerometer migth be different thinks 😉

apply a continous chip and watch the resulting accelerometer output with a spectrogram , I'm shure you will see all kind of harmonic distortion...

For a small band sine analysis no big problem, (and a predistortion can be applied) but for noise this can make differences. 

And if you have multi resonant DUTs on your table ..... 

I would go with a repeating pseudo random signal (if possible)  and a predistotion to get (close to) the wanted acceleration.

[Geoff54]

i Think you don't understand my problem ! For me the main problem is my frequency range. As you can see i would produce a signal between 10 and 200 HZ but i don't know how can i cut my signal ? Actually my system shake but i hear some noise with my shaker due to the high frequency. I have to generate a signal in a range frequency can you help me?

greetings

[Henrik_Volkers]

harmonic distortion creates tones at higher frequencies,  create a sine at 200 Hz and capture with 100kSPS ..... have a look at the spektrum

nonlinearities in the amplifier and the shaker (and the sensor/amplifier) will create unwanted higher frequncy content.

if you use a pseudo repeating noise, you can make a fft on one periode (no window), and by starting by the lowest freqeuncy bin and interating up, correct the output to a wanted spektrum.