Overall Vibration

[MWarlow]

Hi,

Ive posted a few questions on Vibration recently and am now at the point where i have a greater understanding on what is required.

My setup is an accelerometer , connecting to a sinal conditioner that outputs 50mv/g for every 5pc/g.

This is then coming into Labview 8.5 using a NI9234 card, and i have created a VI as attached

It all seems to work fine in that my acceleration windows shows my spectrol analysis of my acceleration in G, i then integrate it to get my velocity (Vibration) spectrol analysis.

And again this all ties up, if i put in avibration of 5g , i see 5g and i also get the correct vibration at that frequency.

The window at the bottom on the VI is my speed, and id like to see the vibration at the current speed...to do this i have used the extract single tone sub vi, pass the frequency of the speed signal in and again this seems to work fine..Lastly i would also like to do overall vibration descibribed by the customer spec as "a calculation of square root of the sum of the squares of all vibration levels within the specified range (50-1000 Hz).". This is also described as broadband vibration and we currently just choose that broadband type in some equipment a called a VM600 that we currently use, but would like to replace with the NI kit.

Now i assume? that in the sound and vibration measurement suite under vibration levels thats what these do...i cant really find alot of info on them, but in the example i have used the peak one, which "seems" to be adding up all the peaks....although i cant find any info to back this up..does anyone know from the statement above what i should be using to do this?...and possibly a description in the same termonology as the above as to what the others do....it would be very much appreciated...lastly i would only need to do it between 50 and 1000Hz , again how would i do this to prevent inclusion of any vibration outside these freqencies....any help clearing these up would be appreciated..as i think im very nearly there to what i want..just these last few issues...

Again, any help and guidance on the issues and perhaps confirmation of what i have done in the vi attached is correct would be so much appreciated

Many thanks for reading and help you can offer

Best Regards

Mike

[dsb@NI]

Mike,

Attached is a screenshot that shows how to compute the total vibration level between 50 and 1000 Hz. It does not matter whether you choose to perform the integration in the time- or frequency-domain. If you stick to time-domain integration, make sure that you allow time for the integration filter to settle. After you compute the spectrum, use the SVFA Power in Band VI to measure the total energy in the frequency range of interest. The Power in Band VI always returns units in EU^2 rms. Convert this to peak magnitude units by multiplying by 2 and taking the square root.



You should not use the Peak Vibration Level VI because the peak measurement does not do averaging, but instead keeps track of the single sample with the maximum absolute value. Because the peak level measurement is based on the value of a single sample, the peak measurement would not be suitable for your application because it would be too sensitive to noise.

Much of the above information is included in the user manual for the Sound and Vibration Measurement Suite.
http://digital.ni.com/manuals.nsf/websearch/C25A17EB97228E64862573720033EFAB

Message Edited by dsb@NI on 08-25-2008 10:52 AM

[MWarlow]

Hi,

Many thanks for the detailed reply, i have implemented what you have shown and it definitely seems to remove the noise facor, and allows me to window my broad band vibration so a big thankyou.

One question though if i may, and its just to settle my mind.

You state that to convert (mm/s)^2 rms to mm/s you x2 and then square root this....could you explain this please as i thought mm/s rms to peak mm/s would be that but the initial value is ^2 also?...apologies is this is a very basic question

Regards

Mike

[dsb@NI]

Yes, if your input value is in units of [engineering unit (EU)] and you set dB On to False, the Power Spectrum VI returns a spectrum in EU^2 rms. The unit label output of the VI will also indicate the units of the spectrum. In some Sound and Vibration examples, the unit label is used to customize the y-axis scale label text which hopefully minimizes confusion about the units of the displayed result.

[MWarlow]

Hi Doug, sorry i meant clarify the equation that if its (mm/s)^2 rms you simply x2 and squarerroot to get peak?..just wondered where that comes from

Ta

Mike

[dsb@NI]

The power spectrum can be treated as the energy in the signal as a sum of sinusoids. Power in Band sums the power by taking the sum of the squares of all bins and returns power in band in units of (mm/s)^2 rms. For a sine wave: peak level = sqrt(2) * rms level.  One could convert the spectrum to peak units before performing the power in band computation, but it was simpler to scale the end result.

[MWarlow]

Hi Doug, again thanks for the prompt reply.

What happens to the ^2 (from the power in band in units of (mm/s)^2 rms) though  - are we sayin the peak broadband is (mm/s)^2 ?? and that is the units that we say overall vibration in?

Thanks once again,

Mike

[dsb@NI]

I don't know what your requirements are in terms of reporting units. I read your posts as a need to report the velocity vibration in units of (mm/s pk). With the Sound and Vibration Toolkit, you have the tools to measure the acceleration, velocity, and displacement vibration in whatever units are required for your application. I again recommend that you use the unit label outputs of the SVT VIs to verify your units along the code path and to give you confidence to apply the the conversions your application requires.

Please feel free to send me an email at doug.bendele@ni.com so that we may start a more interactive discussion.

[pallen]

Hello Doug,

My name is Patrick Allen and I am currently working on a vibration problem that may be similar to one you've provided a solution for here. 

The test stand I am monitoring currently has four accelerometers on it and we are looking to obtain overall vibration levels in ips (in/sec)

I have spent some time with the sound and vibration toolkit and have successfully been able to capture a signal.  The code pictured below is captured on just one channel.  I am not sure though that this signal provides the overall  vibration level throughout a frequency range. 

The other difficulty I am having is how to log this data in a meaningful way to a spreadsheet.  The test stand we are running, logs data from a few dozen thermocouples as well as a dozen other analog channels to a file at 50Hz. 

At first, I thought the SVL Running RMS Level.VI would provide exactly what I needed.  My data is written to a functional global in one VI, and is read from that functional global in another at 50Hz; essentially just picking the latest value that is in there.  However I have been told that this is incorrect and that we do not want RMS values.

A dynamicist involved with the project recently sent me some procedural  information that I am not sure I fully understand either. 

I was given this formula to calculate ips:  ips = g*386.04/(2*pi*f)

Then told to obtain one second of data at 50k.  Then obtain a 3200 or 6400 line FFT which would give "the fft in units of g at each frequency".  I was then to apply the above formula to each g/frequency on the fft to perform the integration and get the results. 

This information seemed very confusing to me, and I'm not sure if what I have so far is even close to what they are looking for.  Any information or guidance you could provide to me would be very much appreciated. 

[pallen]

I forgot to include a shot of my code.

 

Message Edited by pallen on 10-29-2008 10:50 AM