Online and Offline Sine Data Reduction are applications in the Simcenter Testlab Environmental portfolio that are used to process time data from a Sine Control vibration test into spectra. A unique tracking filter approach is used for the time to frequency conversion because it cannot be done effectively with traditional Fast Fourier Transform (FFT) techniques due to the constantly changing excitation frequency.
This article reviews sine data reduction processing and how it is performed.
Article Contents 1. Sine Data Reduction Overview 1.1 Fourier Transform: Amplitude and Harmonics 1.2 FFT Based Processing versus Tracking Filter 2. When is Sine Data Reduction Used? 3. Sine Reduction Data Acquisition Oversampling 4. Online Sine Data Reduction Application 4.1 Getting Started with Online Sine Data Reduction 4.2 Channel Setup 4.3 Sine Reduction Setup 4.3.1 Channel Parameters Panel 4.3.2 Acquisition Panel 4.3.3 Sweep Panel 4.3.4 Measurements Panel 4.3.5 Throughput Recording Panel 4.4 Sine Reduction Acquisition 4.4.1 Run Name Box and Arm Button 4.4.2 Status Panel 4.4.3 Sweep Panel 4.4.4 Recording Panel 4.4.5 Actions Panel 4.5 Procedure for Running Sine Reduction Acquisition 5. Offline Sine Data Reduction Application 5.1 Getting Started with Offline Sine Data Reduction 5.2 Processing Pane 5.2.1 Adjusting the COLA Signal Settings 5.2.2 Harmonic Order Setting 5.2.3 Acquisition 5.2.4 Measurements 5.3 Save Panel
1. Sine Data Reduction Overview
Often the purpose of sine vibration testing is to determine the response of a Device Under Test (DUT) to a fundamental excitation frequency. However, it is rare that a DUT exhibits a pure sinusoidal response strictly at the forcing frequency. Most of the time a DUT responds both at the forcing frequency and harmonics of that frequency, creating a “noisy” response.
As seen in Figure 1 (top, blue trace), the input to the shaker from the Simcenter SCADAS hardware is a pure sine wave, but the response of the control accelerometer contains significant noise (bottom, red trace).
Figure 1: Pure sine input to shaker from SCADAS output channel (top, blue) vs. raw control channel response (bottom, red) and filtered control channel response (bottom, green).
To compensate for this and achieve results that report the response of the DUT at each excitation frequency, swept sine shaker controllers implement a time-domain based amplitude estimation method that incorporates a tracking filter. The tracking filter is essentially a band-pass filter that is applied at the frequency of excitation that suppresses the influence of harmonics and noise on the resulting spectra.
The effect of the tracking filter application can be seen in Figure 1 (bottom, green trace). If the raw, unfiltered data had been used for control, the drive signal would have been reduced, potentially undertesting the DUT in this case. More information on amplitude estimation methods in swept sine testing can be found here: Sine Control: Amplitude Estimation Methods.
Note: If studying the harmonic response of a system to a fundamental frequency, Total Harmonic Distortion (THD) is a calculation Simcenter Testlab can perform. More info on THD and its causes can be found here: What is Total Harmonic Distortion (THD)?
1.1 Fourier Transform: Amplitude and Harmonics
The other (and possibly more compelling) reason to use a tracking filter technique for swept sine testing is that FFT-based (Fast Fourier Transform) analysis is neither fast enough nor capable of adequately capturing the amplitude response of the DUT for controlling a swept sine shaker test.
FFT-based measurements require that a minimum time sample be recorded to calculate a result. This minimum time is inversely related to the frequency resolution of the resultant FFT. In the example shown in Figure 2, a 0.5 second segment of a sine sweep was processed into a FFT of two Hertz resolution:
Figure 2: FFT-based spectrum processed from 75.5 to 76 seconds of the recorded time data. Drive input is shown in blue, response and harmonics in red. A harmonic cursor has been applied. The sine tone sweeps from from 99.3 Hz to 101.6 Hz over the half second.
In that 0.5 second time, the sine tone has swept from 99.3 Hz to 101.6 Hz. Not only are harmonics evident, but the changing frequency of the sine tone during the FFT leads to a “smearing” of the frequency content. The FFT processing provides a somewhat misleading picture of how the DUT is reacting during the sweep.
This scenario gets even worse when attempting to get finer frequency resolution. In Figure 3, a spectra is calculated from the same sweep with 0.25 Hz resolution. This requires a four second time sample. During this time, the frequency has swept from 99.3 Hz to 119.5 Hz.
Figure 3: FFT-based spectrum processed from 75.5 to 79.5 seconds of the recorded time data showing a reduction in amplitude. Drive input is shown in blue, response and harmonics in red. A harmonic cursor has been applied. The sine tone sweeps from from 99.3 Hz to 119.5 Hz over the four seconds.
This spectrum shows how something as trivial as changing the frequency resolution can result in a totally different result! Even though this was a 1 g sine sweep, the amplitude reported in this spectrum is less than 0.5 g! This difference in amplitude is similar in concept to how leakage can affect spectra from non-periodic signals. Because the frequency is constantly changing, the amplitude is distributed over several frequency lines in the result, distorting the true amplitude of the sine wave.
Even with parameters optimized to calculate a best-case result, the inadequacy of FFT-based spectral processing compared to a tracking-filter processing is evident, as shown in Figure 4.
Figure 4: Tracking filter-based spectra (blue traces) vs. FFT-based spectra (magenta traces) of the control channel (top) and the DUT response (bottom).
While Figure 4 shows that the FFT result starts to approximate the clear result obtained with a tracking filter, the FFT-based result starts to fall apart as the sweep reached higher frequencies. Additionally, the tracking filter data can produce a considerably finer resolution in the calculated spectra.as shown in the zoomed view in Figure 5.
Figure 5: Zoomed in view of tracking filter-based spectra (blue traces) vs. FFT-based spectra (magenta traces) of the control channel (top) and the DUT response (bottom). Bottom trace is displaying markers at spectrum data points illustrating the finer frequency resolution that can be obtained with tracking filter processing.
The tracking filter provides a far better picture of how the DUT is responding at any given frequency. It does so with far less headache than iterating FFT processing parameters to get the best possible result. Of course, without tracking filter processed data to compare to, it is difficult to judge what a “best possible result” might look like!
2. When is Sine Data Reduction Used?
Sine Data Reduction is used when a secondary data acquisition system records data during a swept sine test. The two most common reasons for using a secondary system are:
The controller does not have enough input channels to support the data acquisition needs of the test plan, or does not have the appropriate signal conditioning for all sensor types (e.g. strain gauge conditioning).
Data redundancy is desired.
These scenarios are common in large satellite proto-flight or qualification testing, where hundreds of sensors may be used to measure force, strain, acceleration, etc. Figure 6 shows a common dual system setup.
Figure 6: Typical dual-system setup for a satellite test. The main system (left) is used to perform the sine sweep on the test article (middle). The secondary reduction system (right) follows the test (via the COLA signal) and processes data during the test, alleviating the control system resources from having to process data while controlling the sine test.
In these cases, Online Sine Data Reduction is often used to get real-time results that can be used to monitor the safety of critical components and shut down a test if needed. In the cases of Online Sine Data Reduction, a COLA (Constant Output Level Analog) signal is output from the main system into the Tachometer input on the secondary system. The COLA signal allows the secondary system to instantaneously calculate the drive frequency that the controller is outputting to the shaker.
Note that a COLA (Constant Output Level Analog) signal is a sine wave with a constant amplitude, but at the same frequency as the controller drive signal.
The secondary system applies the tracking filter in the same way the main system does during the sine sweep. Figure 7 shows a diagram of an Online Sine Data Reduction setup.
Figure 7: Online Sine Data Reduction setup. The COLA output of the main sine controller is connected to the sync input of the reduction system.
The COLA signal is output from the main controller to tachometer input on the secondary or reduction system.
It is not absolutely necessary to reduce the data with the secondary system during the test. The time data can be recorded during the sweep to process later.
In many cases, all that is obtained is raw time-domain data which will need to be converted to frequency domain data for analysis. The need for this is present in many other industries besides aerospace. For example, in cases where additional signal condition is required such as for strain gauges, or in cases where the secondary data acquisition system is operated by a different department within a company using a shared vibration test lab. This makes data processing a critical part of the process. Only confusion and costly delays can result if the recorded data and the controller data are either processed differently or gives an erroneous depiction of the DUT dynamics.
Note: It is absolutely critical the data obtained from the secondary measurement system contain sweep frequency data from the controller to synchronize with the Sine Reduction process. Ideally this is from the COLA output. Worst case the control channel or drive signal can be used for this (more on that below), but it must be split from the controller output and fed into the secondary data acquisition system to get good results. More information on using a control signal in lieu of a COLA signal in the article: Creating COLA Channel from a Control Signal.
3. Sine Reduction Data Acquisition Oversampling
Oversampling is important when taking data for Sine Reduction. Standard recording speed for throughput in many Swept Sine Controllers is typically set to the required Nyquist frequency to avoid aliasing at the highest frequency of the sweep.
For example, a Swept Sine test in the Siemens Testlab software with a max frequency of 2000 Hz would record the time data at a rate of 6400 Samples/sec. However, Sine Data Reduction requires that data be recorded at a relatively high sample rate. Online Sine Data Reduction, for example, uses a minimum oversampling factor of 60.
Figure 8 shows the difference in COLA processing at a 1x (blue trace), 2x (magenta trace), and 10x (cyan trace) oversampling factor (base value of 6400 Samples/sec).
Figure 8: Processed COLA data in frequency vs. time used for Offline Sine Reduction. The upper picture is a zoomed in view of the lower picture showing that under-sampled data can lead to poor results.
Using under-sampled data can lead to issues with the sweep data too, especially when coupled with an under-sampled COLA signal (Figure 9).
Figure 9: Poor results in sweep control spectra due to under-sampled time data.
Fortunately, there are some tools in Simcenter Testlab to mitigate under-sampled time data. The COLA up-sampling feature can help to clean up noisy COLA signals, see “Adjusting the COLA Signal Settings” below for details. It is important to note that under-sampled data may still yield unsatisfactory results, and require other adjustments such as increasing the number of periods in the measurements. Best practice is to record time data with at least 10 times oversampling. This will reduce any amplitude error to less than 5%.
This section of the article discusses the set-up and execution of a secondary Sine Reduction system. Running "Simcenter Testlab Online Sine Reduction" acquisition software requires that this secondary system receives an input COLA signal (as discussed above) from an active Swept Sine control system as shown in Figure 10.
Figure 10: The Simcenter Testlab Online Sine Reduction software works on a secondary reduction system (orange, right), while the sine control test is performed by the main system (left).
The reduction system can not only process data from the sine test in progress, it can monitor levels and shutdown the main controller if vibration exceeds critical values.
4.1 Getting Started with Online Sine Data Reduction
To open Online Sine Reduction, double click on the “Online Sine Reduction” shortcut in the Testlab Environmental folder (Figure 11).
Figure 11: To start "Online Sine Reduction", double click on the icon in the Simcenter Testlab Environmental folder.
This will open the Online Sine Reduction Workbook. Within the program either an existing project file can be opened, or a new project file can be created.
For more information on getting started with Testlab, project files, and the Documentation and Navigator worksheets, please see the following resources:
After the Documentation and Navigator worksheets, the input channels must be configured.
4.2 Channel Setup
The Channel Setup worksheet needs to be configured with the sensor coupling, sensitivities, etc. that are connected to the Simcenter SCADAS hardware.
For sine reduction, there are two things that need to be addressed in addition to the typical channel setup for an environmental test. As noted above, the COLA channel is required to process the sine data. The COLA channel from Output2 of the main SCADAS unit controlling the tests must be input into Tacho1/sync input of the Secondary SCADAS unit performing the Sine Data Reduction. See Figure 7 for the hardware setup.
After making the connection, Tacho1” must be enabled in the channel setup of the reduction system as shown in Figure 12.
Figure 12: Sine Data Reduction channel setup.
An additional setting, “COLA Hysteresis” is available to adjust for noisy incoming COLA channels. This setting is only used in rare circumstances, as the incoming COLA channel should be nearly perfect.
After the Channel Setup is complete, navigate to the Sine Reduction Setup worksheet (bottom of screen). This is where all the parameters related to sine data reduction acquisition and processing are input (Figure 13).
Figure 13: The Sine Reduction Setup workbook contains five areas to setup a sine reduction test.
The five main areas in the sine reduction setup are:
4.3.1 Channel Parameters Panel
Channel Alarms, Aborts, or Alarm/Abort Spectra can be set in the Channel Parameters Panel. On each channel the upper and lower alarm, upper and lower abort can be set. When activated, the corresponding check value can be entered in engineering units; this check value will be used for the selected alarm/abort estimators. Alternatively, these ‘constant value’ alarms and aborts can be replaced by frequency-dependent abort profiles.
To activate this feature, the add-in “Sine Reduction Abort Profiles” must be loaded and the checkbox “Abort Profiles” in the Advanced Setup dialog must be selected. The columns in the channel parameters panel will be greyed out so that the alarm values are no longer active and that the upper and lower aborts can now be defined as profiles (Figure 14).
Figure 14: The Channel Parameters Panel in the Sine Reduction Setup Worksheet.
The “Sine Reduction Abort Profiles” add-in requires an automation license. Whenever the signal on a channel is outside the alarm/abort limits set here, a warning will be displayed during the measurement run.
When an alarm is transgressed, this is signaled in the “Last alarm” status feedback in the acquisition worksheet. An entry is also added to the logfile of the run. The first transgression of an abort will also launch the “SineReductionAbortActions.bat” batch file. A default version of this batch file is installed in the central configuration folder and will do nothing more than play a sound. This file can be copied to the local (or group) configuration folder and its contents can be customized to launch the required actions needed to react to an observed abort condition.
With the “Sine Reduction Abort Profiles” add-in loaded and the advanced setup parameter “Abort Profiles” active, the above behavior is overruled. The constant value alarms and aborts are replaced by frequency-dependent abort profiles within the given frequency range. These abort profiles can be globally rescaled by applying the up/down button with a given dB value.
The mechanism of launching a batch file upon transgression of an abort is replaced by launching an executable which monitors the abort condition upon arm time. To configure which executable is launched, an entry is available in the TestLabEnvironmental.ini file. The default (located in the central configuration directory) contains the following entry:
The default AbortProfilePrg.exe (located in the central bin directory) will do nothing more that display “none” if no aborts are transgressed and “abort” otherwise. In both cases (standard alarms/aborts or abort profiles), the Online Sine Data Reduction application will keep on measuring when an abort condition occurs. By replacing the contents of the batch file or by creating an alternative executable (using windows automation) one can trigger any action that is required to signal the abort condition to the operator and/or to the system that is actually in control of the excitation to the structure.
4.3.2 Acquisition Panel
In this panel, all acquisition parameters relating for the sine data reduction test are set (Figure 15):
Figure 15: The Acquisition Panel in the Sine Reduction Setup worksheet.
Freq. resolution: This is the frequency resolution of the acquisition. Remember that Sine data processing is a time domain method, and the frequency resolution set here will not affect processing (unlike FFT-based processing), it will only affect how many data points are stored in the result spectra.
Min. frequency: This represents the lowest frequency to be used in the acquisition. All spectral lines below the minimum frequency will be removed. This should be set to the minimum frequency of the Main Sine Control unit’s reference spectrum.
Max. frequency: The highest frequency to be used for the vibration channel group in the acquisition. All spectral lines above the maximum frequency will be removed. This should be set to the maximum frequency of the Main Sine Control unit’s reference spectrum.
Spectral spacing: This should be based on whether the Main Sine Control unit will perform a logarithmic or linear sweep.
Number of Periods: This can be set to Fixed or Tabulated. Again, this should be set up to mimic the settings of the Main Sine Control unit.
Fixed: When the “Fixed number of periods” is selected, the same number of periods per estimate will be used over the entire frequency range. If the entered number of periods per estimate is insufficient for the selected maximum frequency, the status of the verify setup will be red.
Tabulated: When “Tabulated number of periods” option is selected, the number periods per estimate used over the frequency range can be set. This allows the settings to be optimized according to the needs of the test, for better performance or smoother estimates. The default settings for the Tabulated number of periods are chosen such that optimal performance is obtained. The number of periods editor allows you to define a varying number of periods per estimate over the used frequency range. The editor (Figure 16) is opened by pushing the Define... button.
Figure 16: The Tabulated Number of Periods Editor.
The Advanced Acquisition settings (Figure 17) allow different averaging modes to be selected for the amplitude estimation. This determines how Simcenter Testlab handles intermediate measurements between the set frequency resolution lines.
Figure 17: The Advanced Acquisition settings dialogue.
This is the same in a swept sine test - Simcenter Testlab is constantly measuring each period of the sine signal and calculating an amplitude. The available averaging modes are Exponential, Stable, and Peak hold.
Exponential (default selection): Yields an averaging result to which the newest measurement has the largest influence while the older ones are gradually forgotten. The amplitude An is calculated as follows:
An = [1-w]A(n-1) + wAn
Where w is the exponential constant from 0 to 1.
Stable: Also called “linear” averaging, all amplitude measurements have the same influence on the final averaged value. The amplitude An is calculated as follows:
An=((n-1)/n) A(n-1) + An/n
Peak hold: Compares all intermediate samples and reports the largest amplitude of that set in the next recorded frequency line.
In the Sweep panel the system can be set to automatically detect sweep reversal if multiple sweeps are programmed for the main sine control unit (Figure 18).
Figure 18: The Sweep panel of Sine Reduction Setup workbook.
When the Automatic sweep reversal detection checkbox is selected, the software automatically detects when a sweep reversal occurs, and stores the data for the different sweeps in separate folders.
There are "Advanced..." settings for the sweep (Figure 19).
Figure 19: The Advanced settings menu of the Sweep panel.
Some details on the settings:
Advanced… settings: In the Advanced Sweep parameters, how many sweeps are skipped between saving the different sweeps can be set. This is useful during long duration tests, when measuring a huge amount of data or when a very short sweep period is used. By default the stored data corresponds to a snapshot of the online data. When alarms and/or aborts are active, this means a blockset will be saved containing the measured data and the active alarm and abort profiles.
Save primary blocks only: Selection allows only the measured data to be saved (without the extra alarm and abort curves).
4.3.4 Measurements Panel
In the Measurements Panel, any of the five options (Harmonic/RMS/Peak/Average/THD/Measure FRF), either on its own or in combination, can be selected as shown in Figure 20.
Figure 20: Measurements panel of Sine Reduction Setup workbook.
They represent the method of estimating the amplitude for the frequency spectra that will be acquired on the channels specified in the channel list.
Harmonic: This is a filter method that works completely in the time domain. It offers the best estimate for the amplitude of the fundamental frequency and provides excellent harmonic rejection.
RMS: The RMS estimator calculation filters out the effect of quickly changing peak values.
Peak: This estimator simply looks for the maximum amplitude of the sample time signal.
Average: This method filters out the effect of quickly changing peak values and takes into account the complete signal.
THD: This method provides a measure of the total harmonic distortion.
Measure FRF: Checking this box means that the frequency response function will be computed for the selected channel. Select the channel from the dropdown list of available channels to use as a reference for computing the FRF.
In the Measurements panel, there is an "Advanced..." button with the following settings (Figure 21):
Figure 21: The Advanced Setup menu of the Measurements panel.
By switching on the Absolute checkbox, the peak estimator is no longer working on the chunk of data used for the other estimators – which includes the necessary low-pass filtering – but instead works on a constant sample rate (the sample rate of the parallel throughput recording). Typically used for low frequency tests, this allows Testlab to detect peaks that would otherwise be filtered out. Note: When this checkbox is activated, Throughput Recording needs to be activated too.
By switching on the Fixed BW checkbox, the RMS estimator is no longer working on the chunk of data used for the other estimators – which includes the necessary low-pass filtering – but instead works on a constant sample rate (the sample rate of the parallel throughput recording) and thus on a fixed bandwidth. This allows to capture the energy of phenomena that would otherwise be filtered out. The fixed bandwidth RMS estimator can be used as notch estimator to protect the structure. Note: When this checkbox is activated, Throughput Recording needs to be activated too.
4.3.5 Throughput Recording Panel
It is highly recommended to record the raw time data during the sine reduction test, especially for a high value test article. This can be done in the Throughput Recording panel (Figure 22).
Figure 22: The Throughput Recording panel.
Checking the Activate recording box enables the recording of the raw time data during Online Sine Reduction acquisition. This option requires the Add-in “Time Recording During Online Sine Reduction” to be activated.
More settings are available under the "Advanced..." button of the Throughput Recording panel (Figure 23):
Figure 23: The Advanced Throughput Recording settings dialogue.
The Advanced settings dialogue allows an oversampling factor to be set for the recording. By default, Oversampling factor is set to 1 and this corresponds to the sampling required to obtain alias free data at the max frequency set in the Acquisition panel.
4.4 Sine Reduction Acquisition
To run a Sine Reduction acquisition, navigate to the Sine Reduction Acquisition worksheet (Figure 24).
Figure 24: Sine Reduction Acquisition worksheet.
There are a number of options for configuring how data is displayed during the test which will not be covered in this article, but any data will be displayed in the area labeled “responses viewer”.
The main areas to note for Sine Data Acquisition are the Run Name box, Arm button, Status panel, Sweep panel, Recording panel, and the Actions panel:
4.4.1 Run Name Box and Arm Button
In the upper right of Sine Data Acquisition, the current measurement run name is shown next to a Arm button (Figure 25):
Figure 25: The Run Name and Arm Panel.
Run Name Box: As soon as the Sine Reduction Acquisition worksheet is entered, a new run name is given, and the system needs to be re-armed before it is ready to begin a test. To change the name of the run, delete the current name and enter a new one.
Arm Button: This button initializes the system using the current settings. While it is possible to change some parameter settings after “Arming” and before starting the next run but these changes will not be used as settings for this run, they will only be valid for the next run. Most settings however are locked and unable to be changed while the system is armed for acquisition.
Generally it is best practice to click “Stop” and before making any changes and then re-arming.
4.4.2 Status Panel
In the Status panel (upper right of Sine Reduction Acquisition worksheet) there are several fields that provide information on the progress of the run (Figure 26).
Figure 26: The Status panel of Sine Reduction worksheet.
Details shown in the Status panel:
Status: This field displays the current state of the run. When the frontend is armed it will indicate that the system is Armed. Once the run is underway, it will display a series of messages indicating the current process or which action button has been hit.
Frequency: This field shows the current frequency value as calculated based on the incoming COLA signal.
Last alarm: Shows the last frequency at which alarm levels were exceeded.
Last abort: Shows the last frequency at which abort levels were exceeded.
Progress indicator: The test progress is shown in the form of a bar.
4.4.3 Sweep Panel
The sweep panel (Figure 27) contains information on the sweep number (i.e., the number of sweeps performed) and the sweep save setup override.
Figure 27: The Sweep panel.
Settings and information shown:
Sweep (number indicator): This indicator shows the number of the current sweep. The sweep number can be increased or decreased using the arrow buttons nest to the sweep number indicator. This may be necessary if the main control unit is stopped and restarted so that sweep numbers in the secondary system match the sweep numbers in the main system.
Save checkbox: When this box is checked, the current sweep will be saved. The box can be checked to save the sweep and uncheck it when you do not. This is only necessary if the settings made in the Advanced Sweep setup parameters need to be overridden.
4.4.4 Recording Panel
The Recording panel provides information on the progress of the time data acquisition (Figure 28).
Figure 28: The Recording status panel.
Status: The status field informs you on the actual status of the throughput recording. Possible states are:
Throughput not available: The Time Recording During … add-in must be activated using the Tools>Add-ins menu.
Throughput deactivated: The Throughput setting is switched off in the setup worksheet.
Waiting for start: Time recording is not yet running, but it will start together with the test.
Active: Time recording is running.
Stopped: Time recording is stopped.
Time: The time field provides information on how long the throughput recording measurement has been active for the current test, and an estimation on how much recording time is left with the current amount of free disk space.
4.4.5 Actions Panel
The Actions panel contains buttons to start, stop, and pause the test (Figure 29):
Figure 29: The Actions panel.
Buttons do the following:
Start: This starts the Sine Reduction Acquisition run as defined in the setup.
Stop: This aborts the run. The results for the run are available in the project and can be viewed in the “Batch Printing” or “Navigator” worksheet.
Interrupt: This button causes the test run to be interrupted. The run can be resumed by clicking on the continue button.
Continue: This enables you to resume the run if the Interrupt button has been used.the continue button can be used to resume the sine sweep.
Save & Reset: This button will trigger a save of the online data in a folder of the current run and subsequently reset all online data. The recorded time data is not affected by this action.
4.5 Procedure for Running Sine Reduction Acquisition
Before arming the system, make sure to connect a valid and active COLA signal to the first channel of the Syscon module. Without a valid COLA signal being present at the first input of the Syscon module, Arming will be impossible. See Figure 7 for images of the connection set-up.
Press on the Arm button to initialize the system so it is ready to begin a test. The status becomes Armed and this can take a few seconds.
Once the status panel at the top of the worksheet shows “Armed”, the test can be started by clicking the “Start” button.
During the test you can select which responses you would like to view in the Responses display panel. Note: Status now changes to “Measuring” and the action buttons can now be used to interrupt and continue the test.
When data acquisition is finished, click the “Stop” button.
5. Offline Sine Data Reduction Application
Simcenter Testlab Offline Sine Data Reduction is used to post-process time data acquired during a sine control sweep. Processing parameters can be adjusted to better follow the COLA signal, average differently to overcome noise, or calculate additional functions.
5.1 Getting Started with Offline Sine Data Reduction
To open Offline Sine Data Reduction (OSDR) program, double click on the “Offline Sine Data Reduction” shortcut in the Simcenter Testlab Environmental folder (Figure 30).
Figure 30: Offline Sine Data Reduction is started from the Simcenter Testlab Environmental folder
In Navigator worksheet, locate the Sine Throughput file, right click, and select “Add to Input Basket” to add the dataset to the input basket (Figure 31).
Figure 31: Adding Throughput data to Input Basket in the Navigator Worksheet.
Move to the Time Data Selection worksheet. Click the “Add” button to select the time data for processing (Figure 32). If data is already selected in the Time Data Selection Worksheet, click the “Replace” button to replace the existing data.
Figure 32: Location of Add and Replace buttons in the Time Data Selection Worksheet.
With the data added or replaced, the list of time traces should appear in the Data Set table in the upper left, as seen in Figure 33.
Figure 33: Time Data Selection Worksheet with Data Set added.
If interested in viewing (with zoom, etc.) the time history data before performing offline data reduction, check out the following knowledge articles:
Next, navigate to the Offline Sine Data Reduction Worksheet. Add the Data Set to the Trace List by clicking the “Make Trace List” button (Figure 34) in the upper left of the workbook.
Figure 34: Adding the Data Set to the Trace List in Offline Sine Data Reduction.
The Data Set should then appear in the Trace List. Traces can be displayed as desired by checking the respective box in the “View” column. The COLA signal must then be selected in the “COLA trace” drop down. A smooth frequency vs. time should be overlayed by default in the Strip Chart display.
The “View COLA traces” checkbox can be used to disable or enable this function (Figure 35).
Figure 35: View COLA traces checkbox location.
If the COLA trace (in red) is now shown overlaid with the COLA signal (green) in the upper right of the display, click the "View COLA traces" checkbox ON.
It is possible to process only a segment of the time data. By default, the selected processing segment (if defined) is used from the Time Data Selection worksheet, but a segment can be selected in the Offline Sine Data Reduction worksheet as well. With the processed COLA trace overlaid on the visualized traces, it is easy to select the desired segment from the source data to process.
Use the “Define segment” button in the Trace List pane to apply this segment selection (Figure 36).
Figure 36: Location of “Define segment” button in the Trace List pane of the Offline Sine Data Reduction worksheet.
The processing pane section (lower left of the Offline Sine Data Reduction workbook) has settings for processing the COLA signal into a frequency as shown in Figure 37:
Figure 37: The Processing Panel is located in the lower left of the Offline Sine Data Reduction workbook.
This panel has several key processing settings located in the COLA, Acquisition, and Measurements tabs.
5.2.1 Adjusting the COLA Signal Settings
If the COLA signal displayed is not a very smooth trace (an examination of the detailed view is recommended, see Figure 38), the COLA settings may need to be adjusted to improve the processed COLA signal.
Figure 38: Strip Chart Display of under-sampled COLA signal before COLA Settings are applied.
To access the COLA Settings menu, click the “COLA Settings…” button in the Processing section of the Offline Sine Reduction Worksheet as shown in Figure 39.
Figure 39: Accessing the COLA Settings Dialog.
The COLA Settings dialog contains several settings to help improve the COLA frequency-time trace:
Upsampling factor: The conversion from a recorded sinusoidal signal to a frequency versus time trace relies on the correct determination of the crossing between the signal and the cross level. When faced with a relatively low sample rate in the traces, up-sampling of the signal can improve the detection.
Cross level: This sets the level with which the COLA signal will be compared to determine the period of the underlying sine signal. To obtain precise estimations of the crossings, the level should be such that the crossings occur where the slope of the COLA signal is steepest. The COLA output of the DAC of a Testlab Sine Control system is a sine with zero mean. On such a COLA signal the most appropriate value for cross level is 0 (zero).
Lower tolerance and Upper tolerance: These parameters specify a hysteresis tolerance band around the cross level. The used value is not very critical (it is required to be smaller than the amplitude of the sine signal in the COLA) except when faced with a noisy COLA trace.
Hold off rate: This parameter defines the minimum elapsed time between two crossings expressed as a percentage of the previous period length. Just like the hysteresis tolerance, this parameter is only important when faced with a noisy COLA trace.
Smoothing factor: This parameter smooths out the evolution of the resulting frequency-time trace. The value is the size, expressed in number of crossings, of the linear smoothing window applied on the period between crossings before converting to frequency. A moderate value for this smoothing factor can be required when faced with a noisy COLA trace to avoid oscillations in the frequency-time trace.
By adjusting the upsampling factor, the COLA trace in Figure 37 was cleaned up significantly as seen in Figure 40.
Figure 40: Same COLA data as Figure 31 with upsampling factor of 10 applied.
5.2.2 Harmonic Order Setting
The amplitude and phase information of the fundamental frequency can be extracted by setting the harmonic order to "fundamental". This can also be done for harmonics of this fundamental frequency by setting the harmonic order to 2-10. Calculating harmonic orders is only supported for the harmonic estimator. When a harmonic order different from the fundamental order (first order) is set, all estimators except for the harmonic estimator need to be deselected (see Measurements Processing below). The available harmonic orders are the first to the tenth order. If a higher harmonic order is calculated, the name of the harmonic order is extended. The order number is indicated between brackets behind the function class.
The Acquisition tab of the Processing section has settings for the resulting spectrum as shown in Figure 41.
Figure 41: The Acquisition Processing tab has settings for the spectral result.
The Freq. resolution, Min freq., Max freq. and Spectral spacing parameters determine the abscissa of the spectra on which the estimated amplitude and phase information will be reduced.
Spectral spacing: Set to Linear or Log.
Frequency resolution: Can be entered in Hz (linear spectral spacing) or in lines per Octave (log spectral spacing).
Minimum and Maximum frequency: Determines the frequency band. The frequency lines in abscissa of the resulting spectra that do not receive a value (or an average of values) because no estimate is generated for the band around that line will be removed from the spectra. Additionally, depending on the frequency band present in the selected segment it is possible that the processed spectra will only cover a sub-range of the frequency range that is defined here. If the selected segment contains frequencies outside of the band defined here, the resulting spectral lines will be discarded from the results.
Number of periods: The estimation process will cut a chunk of samples from the source traces to perform its calculations. As the processed COLA trace indicates the current frequency, the number of samples corresponding to the “number of periods” is used. A fixed value can be chosen for the complete frequency band. If the entered number of periods per estimate is insufficient for the selected maximum frequency and sample rate, a warning will be given when the processing is started (Figure 42).
Figure 42: Warning pop-up when the number of periods is insufficient for selected max frequency and sample rate.
Alternatively, a tabulated version can be defined. The "Define..." button will access the Number of Periods Editor where this table can be entered (Figure 43).
Figure 43: Tabulated Number of Periods Editor
The Advanced Acquisition dialog sets the parameters that influence how the averaging of multiple estimates to the same spectral line is done. It is scheduled by clicking on the "Advanced..." button in the Acquisition tab of the Processing panel (Figure 44).
Figure 44: Click “Advanced…” to access Advanced Acquisition dialogue.
The following parameters can be set:
Averaging mode:When two (or more) amplitude/phase estimates are generated on frequencies close to a spectral line in the defined grid (abscissa), these estimates are averaged. This parameter determines whether this averaging is stable (linear), exponential or peak-hold.
Exponential constant:This parameter sets the exponential weighting factor when the exponential averaging mode is chosen.
The types of spectral functions to be calculated during the processing are selected in the Measurements tab shown in Figure 45.
Figure 45: Measurements Processing tab.
Parameters that can be set include:
Estimator: The checkboxes control which estimator(s) will be active, resulting in spectra being calculated and stored. At least one estimator must be activated. Total Harmonic Distortion (THD) can also be calculated. For more information about THD see: What is Total Harmonic Distortion (THD)?
Phase reference:The phase information of the harmonic estimates on a trace will be referenced to the estimates for this channel.
Measure FRF:Selecting this checkbox will result in the generation of single reference FRFs with the selected channel as reference.
5.3 Save Panel
Once all the processing parameters have been set, the “Save” panel (Figure 46) is used to initiate processing.
Figure 46: The Save panel in the lower left of the Offline Sine Data Reduction.
There are two settings:
Run name: Specify the name of the run where the results are saved. Run names must be unique within each section. If a run name is entered that already exists, it will be appended with a number to prevent duplicated run names. If multiple runs are saved the number will increment with each additional run.
Save… button: This starts the processing and saves the resulting spectra in a run under the active section using the name set in the run name field. During the processing, a progress indicator will po up to provide feedback. The pop up includes a “Break” button. When the Break button is pushed, the calculations will stop, and no results will be saved.