Accelerometer Data Conundrum: Power Spectrum, Noise, and Scaling (VI included)

[PFour]

Hey everyone,

I am acquiring voltage data from a pair of piezoelectric accelerometers and am running into three main issues. As far as I can tell, they're [generally] independent, so I'll break them down separately.

1. I unable to get my accelerometer bias values correct. The attached calibrationcert.png shows a calibration document for one of the accelerometers. At the moment, I am taking my raw AI voltage data and dividing by the sensitivity in V/m/s^2 (in this case 0.0104). My understanding was that this should scale the data correctly such that I am plotting in G's. However, it seems that each accelerometer has an additional offset, which I can't seem to find anywhere on the sheet(s). When lying flat on a table, accel 1 is consistently sitting around -7 g's and accel 2 is around -25 g's. Any ideas what I could be missing? 

2. I'm seeing a great deal of noise in my data acquisition. I am using specialized low-noise cables, yet even with the accels sitting motionless on foam pads the data jumps around pretty significantly (+/- 3 g's). Take a look at the frontpanel.png to see what I'm referring to. Assuming that this issue is somehow linked to my first problem, listed above. 

3. I would like to plot my power spectrum in real time. I've tried three slightly different methods; none of which are producing the consistent stream of populated cluster data that is needed to make the PS plot. Please take a look at these three methods in AccelerometerDataAcquision.vi

I appreciate the help! It should be noted that I was able to generate a single instance of the response PSD in another test vi (using method #1). For whatever reason, that method falls apart when I stick it in a while loop. I will attach this, as well as the data file that it reads from -- change the file path and it should be good to go. 

Main VI in question: AccelerometerDataAcquisition.vi

Cheers

[Henrik_Volkers]

Some important information is missing:

Type of accelerometers?  

Type of DAQ you are using?

How are they connected?

Why do I have to download 7MB of zip from an unknown for maybe getting that information?

First shot: You have IEPE sensors ... and they are not correctly attached and read ...

[Henrik_Volkers]

OK .. downloaded the zip..

and BINGO it's an IEPE sensor and the BIAS voltage of 11V is clearly noted 🙂

can't open your vi... and maybe you should ask the moderator to remove your zip since some data in there is clearly marked confidentiality

Again: What type of DAQ and signal conditioning are you using?

IF you DAQ supports IEPE sensors, is it turned on?

[Henrik_Volkers]

For the rest of your questions:

Seems that you havn't connected the the sensors right, so no wonder you read garbage 😉

And (not only) in signal processing you have: Garbage in -> Garbage out 😄

Hint: The manual/documentation of the sensors have a long explanation on how IEPE sensors work and how to connect them. Same with the DAQ ...

And: If you scale with SI units (V/(m/s²)) you get SI units .. not G's  (whatever that means 😉 :D)

[GerdW]

Hi,

maybe you should ask the moderator to remove your zip since some data in there is clearly marked confidentiality

Well, I would also mark code of that quality as "confidential"… 😄

[PFour]

Henrik:

1.) No worries -- there is nothing that is actually confidential in the included .zip. That's a placeholder for until we are acquiring actual data. Thanks for checking though

2.) I am using a cDAQ with NI 9215 (BNC) and these IEPE sensors [ sensor 1 , sensor 2 ].

3.) I believe I understand how they work and am pretty certain they are properly connected. Thinking my issue has to do with how I am scaling and applying the bias voltage. In general, it is.... Accel [m's^2] = Volts / Sensitivity , but where does the bias come into play? I am wanting a range of +/- 50 g's (490.5 m/s^2).

I have incorporated this by setting my maximum and minimum AI voltage to +/- 5.1V such that at 490.5 m/s^2 the unit should be reading +5.1V due to the 0.0104 V/m/s^2 sensitivity.

Also, the included documentation mostly discusses proper mounting and surface preparation techniques. The information included that discusses sensitivities and converting from volts to absolute acceleration is relatively bare-bones.

[GerdW]

Hi PFour,

but where does the bias come into play?

See the product manual for sensor1, page 10, figure 8! This explains how the bias is applied and "comes into play!"

The manual for your NI9215 does not mention any ICP/IEPE driver, so the question remains: HOW did you wire up your sensors? How do you apply the bias?

[PFour]

GerdW:

Aha! Finally makes sense -- I see what you're saying now. I was foolishly mixing up the capabilities of NI 9215 with NI 9234. Of course I'm getting garbage data because I'm not even exciting my sensors! 

I think that if I route something like this between my sensors and my DAQ I should have more luck. 

Cheers

[GerdW]

Hi P4,

well, I took a cheap current limiting diode with a standard 24V power supply, took less than 10€.

For my measurement requirements that was sufficient…