impact testing

[JPiel]

I am trying to perform an impact test using labview express with the sound and vibration toolkit 5.0. I have located a couple of examples on the NI developer zone. Some will not run and some do not perform as expected. I can not seem to get it to average impact responses in order to use the coherence function. Can someone provide some assistance?

[EricS.]

Hi JPiel and welcome to the NI Discussion Forums!

In order better assist you could you provide some more information?  What examples did you find on the Developer Zone that didn't work?  And how exactyl did they not perform as expected?

What hardware are you using to perform the impact tests?

Also, are you using LabVIEW Signal Express, or are you programming with Signal Analysis Express VIs in LabVIEW? 

[JPiel]

Eric, I am using the SVA Impact Test example that came with the sound and vibration toolkit version 5.0. I am using 2009 Signal Express with a usb -9233. The problem I am having is understanding how to view the averaged FRF after performing a number of impacts. Is the averaged FRF to be viewed in the Step Setup or in the Data View? In data view, I find that depending on when the impact occurred relative to how the block of data has been recorded, I can sometimes get a split time block of the transient, thereby creating poor or incorrect FRF's. In addition, it also appears that the phase and coherence graphs aren't what they should be as seen below. Any advice or a better program would be greatly appreciated.                                                                                                                                                                                                                                                                                                        

  

[dsb@NI]

Copied from another post http://forums.ni.com/ni/board/message?board.id=231&message.id=6720:

Here is a previously posted example program for (LabVIEW) that implements a software trigger for impact test.

<link no longer exists>

The software trigger makes it so that the 9233, 9234, 4431, 4432, and other devices with no hardware analog level triggering can all be used for impact test.

The phase graph looks like the response and reference signals have different timestamps. Use the Frequency Response configuration parameter 'ignore time delay' to ignore or compensate for differences in the timestamps. If the phase still isn't what you expect, recommend that you check the triggering VI to verify that both signals get the appropriate amount of pretrigger samples in order to stay aligned in time.

Closer inspection of your screenshot suggests that perhaps the inputs to the Frequency Response Step are not aligned. It looks like the acquired data is input to the stimulus and the triggered data is input to the response. Both stimulus and response inputs should be triggered signals.

 

Coherence is only valid when you look at averaged results. I cannot think of a way to batch process a set of logs to generate an average as the processing will restart every time you switch logs and Run. If this specific data is critical, then we can process the logs in LabVIEW and have better control over the averaging.

In the latest version (2009) of the Sound and Vibration Assistant, the Impact Test example project does what you are requesting. The default work area runs continuously and uses a software trigger to return data blocks so that you never get a 'split time block of the transient.' The advanced work area goes further and allows you to configure limit test rejection criteria in order to exclude overloads and double hits when measuring the averaged FRF.

[JPiel]

Eric, thanks for the reply. I did not realize that I should be choosing both channels as the input signal for the trigger. Therefore, I wasn't getting a triggered response channel and this may very well be the reason for the phase issues. I will try what you suggested and see if that is the difference. I am also looking into upgrading the SVA toolkit. Thanks again. 

[JPiel]

Doug or Eric, I have one other question. How come when I set the sampling rate to 1K in the Acquire step setup, is my fmax still 1K? Shouldn't it be 500 or 512Hz? Thanks.

[dsb@NI]

sample rate is coerced to a rate supported for your device.

turn off autoscale on frequency axis and set range of interest based on the valid frequency range of your measurement.

[KarstenG]

Hi everyone,

were are using a USB-4432. We want to use it for a special kind of impact hammer test.

I tried the example "svxmpl_impact_test_daqmx.vi" and it worked without any problems. But we need more.

I will try to explain what we need.

We have several measuring positions. On each position we want to perform a hammer hit and have the FrequencyResponseFunction computed 

So far, the example works fine. Now, I want to check the signal and if its fine I want to press some kind of "Accept hit"-Button and have the hit stored. If not I want to repeat it. Then, I want to do a second hammer hit, compute its FRF and have the Coherence of both FRFs computed. If both hits are fine I will proceed to the next position.

Maybe this could be solved with two indexes one for the number of hits and one for the number of measurement positions. The results of each hit could be stored in an array. This way I would be able to change the index and repeat any of the hits without having to start over from the first position. The "Accept Button" might be in a case sequence.

How do you think about it? Are case sequences, indexes and an array a good idea? Do you know of any example that shows the principle of storing measurement results and performing computations between them.

I appreciate your ideas and help

Karsten

i

.

[EricS.]

Hi Karsten,

That is definitely possible to do.  I think the easiest would be to setup a state machine.  The states would be "Test", "Save", and "Stop"

At the end of each iteration you would wait for user input (potentially a 3 option dialogue "Accept", "Re-test", "Stop").  If you select "Accept" then move to the "Save" state which would process the data from the previous loop (through a shift register), and add the desired data to an element in an array, and its index in the array would refer to the position for the measurement.  If you select "re-test" it will just repeat the "Test" state until you have data that you want to save.

This would still increment through the positions.  If instead you wanted to be able to pick and choose which position to retest, you could do that also.  Then you you could have an index variable also that you would modify before saving the data.  For example, the first thing it asks in the "Save" State could be a prompt for what position the test was for and it will save the data appopriately. 

For more information about state machines refer to the tutorials below: 

 Application Design Patterns: State Machines

 Exercise: State Machines

[dsb@NI]

please refer to related post:

http://forums.ni.com/ni/board/message?board.id=231&message.id=1258&requireLogin=False

example code is posted - feedback appreciated