“Oops, I entered the wrong calibration value, and took a bunch of data! Now what do I do?”
Background
Under ideal circumstances it is always best to calibrate each transducer in your test setup before making a measurement. Calibration helps ensure you get accurate amplitudes in your data by checking the transducer sensitivity.
This sensitivity is used by Simcenter Testlab (formerly called LMS Test.Lab) to convert the voltage signal from the transducer to the correct amplitude of whatever quantity is being measured. The measured quantity (pressure, acceleration, etc) is called the “Engineering Unit,” or EU. The transducer sensitivity is provided by the transducer manufacturer, and will come in the form of voltage/EU, for example 10.0 mV/g or 53.3 mV/Pa.
Sometimes mistakes are made during test setup.
Sometimes calibration may not be possible before measurements need to be made, and a best guess is used for the sensitivity values. Perhaps a typing error is made and the wrong sensitivity is input to Simcenter Testlab. Whatever the reason for the error, in order to obtain accurate amplitude values in these cases it becomes necessary to change the sensitivity values after the fact and adjust our data to the correct level.
This article will outline a method to perform this operation. It is important to note that the raw voltage signals recorded by the SCADAS unit and saved in Simcenter Testlab remain constant before and after the following procedures… What is being modified is the conversionfactor between the measured voltage signal and the corresponding EU displayed by Testlab.
The Problem: Wrong Sensitivity used in Channel Setup
In this example let’s suppose I inadvertently grabbed the wrong kind of accelerometer for my test. I intended on using a 100 mV/g Model B accel for my testing, but accidentally used a Model A instead, which has a sensitivity of 10 mV/g. (See Figure 1)
Figure 1: Two models of accelerometers with common sensitivities.
Because I did not have time to calibrate, I continued with my Channel Setup (Figure 2), setting my “Actual Sensitivity” field to 100 mV/g. This field is telling Testlab that every 100 mV of voltage it records on Channel 3 corresponds to 1 g of acceleration. In reality I’m using Model A accelerometers, which put out only 10 mV for 1 g of acceleration.
Figure 2: Channel Setup is where transducer sensitivities are entered in Testlab.
I continued with my testing, and later when I looked at my data I noticed a big problem… The acceleration amplitude is way too low (Figure 3), only about 1/10th of what I should have measured!
Figure 3: Time history of Channel 3 acceleration measurement. Peak amplitude should read 1.4g!
Panicked, I look through the pictures I took of my test setup and I notice my accelerometer mistake. My Actual Sensitivity field is wrong for all of my tests. The lab is busy now, and I can’t re-test. Can I fix my data? The good news is: yes!
Time Signal Calculator
In general, Time Signal Calculator (TSC) can be used to perform all kinds of tricks with time data. See this article for more tips on TSC. In this example we are using Time Signal Calculator to apply a scale factor to our time histories to correct the amplitude of our data. Figure 4 shows the formula we will use to apply the correct scale factor to our data.
Figure 4: Formula to determine scale factor.
SensitivityRecorded is the incorrect sensitivity we used for our recording, SensitivityActual is the correct value for our transducer. Dividing the two sensitivities will result in a unit-less scale factor that will correct the amplitude in our recorded data.
The values from my accelerometer mix-up example are shown in Figure 5 a-c. The resulting scale factor in this example is 10x. This makes sense: I incorrectly told Testlab that the transducer was more sensitive than it really was, so to correct the amplitudes I should scale the amplitudes up.
Figure 5: Calculating scale factor to correct the accelerometer mix-up.
Procedure in Simcenter Testlab:
Step 1: Load time data into Time Data Selection worksheet by right-clicking on the .LDSF file in Navigator and selecting “Add to Input Basket”.
Step 2: Turn on Time Signal Calculator Add-in. [Tools > Add-ins…> Time Signal Calculator]
Time Signal Calculator will appear at the bottom of the Time Data Selection worksheet.
Step 3: Create a new time trace by multiplying the original data by the ratio (wrongly used cal value/correct cal value) according to the formula in Figure 4.
Step 4: Hit Calculate. The new time data appears as shown below. Orange indicating that the calculated data has not been saved yet.
Step 5: Remove original data before saving to avoid confusion. If the original data is not removed, both the incorrect and corrected data will be stored in the new time file. Select the entire row by clicking on the row number, then click the Remove Channel(s) icon.
Step 6: Save corrected time traces. Use “Save As…” to give the corrected data a new Run name. All done!
Plot showing original and corrected acceleration data.
Multi-Run Processing
This method can also be used to fix multiple runs/tests all at once. I can fix this accelerometer channel on all of my runs simultaneously by adding all of them to the Input Basket at the same time. In Time Data Selection, change the view selection to Channels Pivot to view all the runs to fix.
The view is changed to show all the channels.
Channels Pivot shows me all the channels along the top, and all the measurements/runs along the left side. In this case, I have four measurements with one channel – CH3:BODY 3:+Z – in each. Now when I run the calculations in Time Signal Calculator it will apply our scaling factor to all of the runs at the same time.
I now see my new BODY:3:+Z is calculated for all of the runs. Before I save these corrected runs, I will remove the original data as before. This time I will select the original data by clicking on the column header for CH3, then clicking on Remove Channels.
After pressing "Save As...", highlight the runs to save. Choose "Use original run name, append:" to add an extension to all the corrected runs, and then press "OK".
After performing the steps in this article, only the time histories are corrected. Any spectrums, order cuts, autopowers, etc that were previously calculated will not be corrected. The corrected time histories will need to be reprocessed to get the spectral data to have the correct amplitude.
