Accelerometer - G's - Vel - disp 2

[cjquijano]

I am hoping someone can help me with a brain block.

I am collecting accelerometer data in volts. Using the scaling factor we received with the accelerometer (10mV/g), I can get G's. My question is how do I convert to velocity and displacement?

Thanks for your help.

Christopher Quijano

 

[tomirvine]

You need to integrate acceleration in order to calculate velocity.

You then need to integrate velocity in order to calculate displacement.

The integration can be carried out in either the time domain or the frequency domain.

Highpass filtering is usually needed if the integration is done in the time domain.

Furthermore, the acceleration signal should be multiplied by 386 to convert G to in/sec^2, prior to integration.

Sincerely, Tom Irvine

http://www.vibrationdata.com

 

[JimTaylor]

A couple of thoughts.

1. Your instrument should be able to do the integration to velocity and maybe displacement. Beware of double integration to displacement - it adds a lot of noise.

2. To convert a single frequency mathematically use:
V=6.28fD and A=6.28fV where D=displacement in thousants of an inch, V=velocity in inch/sec and A= acceleration in inch/sec-squared. f = frequency in hertz.

 

[GregLocock]

Be careful with that. B&K seem to be recommending that you don't integrate with their otherwise wonderful 2635 charge amps, and I'd agree. It seems OK with sinusoidal inputs, but it can demonstrably go very nasty on real world signals.

Obviously integrating your spectra on the FFT analyser will be OK (with certain provisos), if that's what you meant by 'instrument'

Cheers

Greg Locock

 

[JimTaylor]

Greg is correct. You can't use the mathematical method shown for anything but a pure, single frequency sine wave. There is no simple calculation for a real world signal.

I ment FFT analyser. I should have said so to prevent confusion. Thanks.

Jim

 

[cjquijano]

Thanks everyone for you help. What I am trying to do is measure the vibration on either side on an industrial engine isolation mount. I have an accelerometer hooked to a battery powered current source, and that signal is connected to a National Instruments DaqCard. I am using the Labview software to show the raw signal in volts and an averaged power spectrum. I wish I had all the other fancy equipment to do this test with, but I work for a construction equipment manufacturer, so I'm lucky to have what I got. My goal is to be able to compare the frequencies and displacements on either side of the isolation mount.

Well, thanks again for your help,

Christopher Quijano
MSA

 

[JimTaylor]

Labview should be able to do the integration for you. It's pretty good software.

 

[alvesp]

You need to mount a 'Frequency Response Function' on Labview. Not the simplest of tasks, I must add.
I did it after spending substantial time at it.

You wish to know what is the transmissibility of vibration at each frequency within a range of interest. Depends on where your engine is going to be working and how quiet you wish for it to be around it.

 

[kaykay]

Displacement, velocity and accelertion are related to each other as a function of frequency. It must be understood that the readings be of the same type or they eill not agree. In general it is difficult to get any two readings pricisely agree with each other.
You may lok into

http://www.reliabilitydirect.com

D=Displacement, P-P,mils
V=Velocity, 0-P,in/sec
A=Acceleration, 0-P,g's

D = 19.10 X 103 X (V/CPM)
D = 70.4 X 106X * (A/CPM2)
V = 52.36 X 10-6 X D X CPM
V = 3.87 X 103 X (A/CPM)
A = 14.2 X 10-9 X D X CPM2
A = 0.27 X 10-3 X V X CPM

 

[MikeyP]

Frequency domain integration works fine as long as your signals do not suffer significant leakage when Fourier transformed. Obviously you can apply a Hanning window to prevent this, but this will alter the original time signal. The best way to do frequency domain integration is to have a time signal which is in the form of a burst (ie the response time signal starts at zero and is allowed to decay to (near) zero within the acquisition length.

Another thought... If you have measured the acceleration transmissibility (a dimensionless frequency domain function) won't this be the same as the displacement transmissibility? Or is there something I'm missing? it is late after all!

Regards

Michael

 

[Irwin]

It looks that is very simple! (Just 2 integration process!) I tried this with Cats software and I was surpriced! I tried in several way to integrate to filtered to move the function. I think the best way if you make a simple test. (Forexample a simple mass - spring system) You pull down with an exact distance and try to measure with your system and with a simple watch and calculate what should be there? Do it with each accelerometer! And after that maybe you have a chance!
(I am not sure that anybody else can integrate back each case the velocity and the displacement! You should know what would be the result!!!)
Also an important point: the frequency range of your accelerometer and a type of your accelerometer! Pleses see the characteristic of your test bench!

We made an another test: A measurement company measured accelerations and in the same points displacements. They gave us the data and we could not fit the datas! I am intereseted in your case, will you get correct data or not!

Ervin

 

[Rob45]

Here's another hint:
For your engine mounts, you want the resonant frequency to be as low as possible to achieve the best isolation. It might be necessary to install travel stops to limit travel during start-up/shutdown, when the engine rotational speed will equal the resonant frequency, but this approach will otherwise give you the best isolation.

 

[micjk]

To anyone using a CSI 2120 consultant analyzer and RBMware software. I've collected data of some equipment, downloaded it to RBMware and ran an exception analysis report based on percentage of fault. The report was generated with mixed values i.e. acceleration and velocity. Is this because the fault percentage was based on these values, if so how can I convert them to be the same, either acceleration or velocity. I've tried several formulas but the results seem to be out of whack.
Thanks in advance,
Roy Gariepy
Maintenance and Reliability Tech
Dorlastan Fibers
Bayer Corp.