In this measurement column, we address frequently asked questions received by our customer support center and provide answers. In the previous installment, we explained how to set up the impulse hammer to activate the trigger during hammering measurement.
In addition to the "trigger not engaging" issue we discussed last time, common questions we receive regarding hammering measurements include "natural frequencies not being produced," "the expected frequency response function not being achieved," and "low coherence function." This time, we will introduce methods for addressing these issues.
Measurement of frequency response function by hammering
Figure 1 shows an example of the connection used when measuring the frequency response function by hammering.
The target object is struck with an impulse hammer, and the frequency response function is measured from the force signal from the impulse hammer and the acceleration signal from Accelerometer.
-
Figure 1: Example of connection between impulse hammer, acceleration detector, and analysis device.
A/D overload and voltage range
If the signal from the impulse hammer or Accelerometer exceeds the voltage range of the analysis device, an A/D overload (input overload) will occur, and the correct results will not be obtained. In hammering measurements, multiple strikes are performed and the average result is calculated, so if an A/D overload occurs even once, the correct results will not be obtained.
If A/D overload occurs, the LED on the input channel of the analysis device will light up red. Perform several strikes and adjust the voltage range, etc., so that A/D overload does not occur on any channel. When using an impulse hammer, place the head at a constant height and let it drop by its own weight to apply a consistent impact force.
If a power supply unit or sensor amplifier is connected between the impulse hammer or Accelerometer and the analysis device, please also check that there is no input overload in those components.
If the analysis device has a function such as "A/D overload cancellation," turning it ON will automatically discard data when an A/D overload occurs. During hammering, the trigger will not engage if the impact is too weak, but the A/D overload cancellation function will activate if the impact is too strong. By monitoring the A/D overload LED on the analysis device, you can determine if an A/D overload occurred due to an excessively strong impact.
The "A/D Over Cancel" setting is configured in the DS-3000 series via [Input/Output Settings Menu] ⇒ [Sample Condition Settings] ⇒ [A/D Over Cancel]. For the CF-9000 series, it is configured via [HOME] ⇒ [Input] ⇒ [Sample] ⇒ [Over Cancel].
CCLD (Sensor Power Supply)
CCLD stands for Constant Current Line Drive, a method of supplying power to the preamplifier built into sensors such as impulse hammers and Accelerometer. When CCLD is OFF, a signal that is much smaller than normal (less than 1/100th) will be output. If the Y-axis scale of the power spectrum or time-axis waveform is set to AUTO, it may be difficult to notice that the signal is small, so set the scale to Default or Manual, or check by reading the amplitude value with the search cursor.
Depending on the analysis device, a "disconnection detection function" may activate and turn off the CCLD when the sensor is removed. If you start the analysis device without a sensor attached, the CCLD will automatically turn off, and even if you attach the sensor afterward, the CCLD will not turn on.
If the trigger does not activate, or if the amplitude of the power spectrum or time-domain waveform is smaller than normal, or if the coherence function is low, please check whether you are using a CCLD type sensor and whether the CCLD setting is turned ON. For the DS-3000 series, you can check this in the [Input/Output Settings Menu] ⇒ [Input Settings Dialog]. For the CF-9000 series, you can check this in the display at the top of the screen (Figure 2) or in the dialog box that appears when you operate [HOME] ⇒ [Input] ⇒ [Input Cond].
-
Figure 2: Example of channel status display for the CF-9000 series.
Window function
In frequency response function measurements using hammering, a rectangular window is used as the window function. Force windows and exponential windows may also be used, but Hanning windows and flat-top windows are never used.
The default setting for analysis devices is often the Hanning window. If the power spectrum or frequency response function does not produce the expected waveform, or if the coherence function is low, please check the window function setting.
Averaging process
For frequency response function measurements using hammering, the power spectrum averaging mode is used.
The analysis device may be in a different mode after being used for other purposes, so please confirm that it is in power spectrum averaging mode.
Double Hammer and Coherence Function
In frequency response function measurements using hammering, the coherence function is used to verify that the measurement is correct. If there is a high correlation between the impact force waveform from the impulse hammer and the vibration waveform detected by Accelerometer, the coherence function will show a value close to 1. The coherence function cannot be obtained without averaging it at least twice.
Figures 3-1 and 3-2 show graphs of the time-domain waveform, power spectrum, frequency response function, and coherence function under normal conditions and during double hammering. CH.1 is the impulse hammer, and CH.2 is the acceleration detector. Double hammering refers to a phenomenon where the hammer head strikes the object two or more times. In such cases, accurate measurement results cannot be obtained.
Looking at the time-domain waveform of CH.1 in Figure 3-2, we can see that a double hammer is occurring. Even if a double hammer is difficult to discern from the time-domain waveform, it can be identified by examining the power spectrum and coherence function.
A correct power spectrum for a striking force waveform is one that smoothly slopes downward to the right, as shown in the CH.1 power spectrum in Figure 3-1. With a double hammer, for example, the power spectrum may become wavy, as shown in the CH.1 power spectrum in Figure 3-2. Also, dips may appear during the downward slope. The coherence function will be lower than in the normal case.
After several successful strikes, a mishit, such as a double hammer, will suddenly worsen the coherence function. Check the coherence function after each strike and pay attention to any changes. Alternatively, you could try intentionally striking twice or shifting the striking position to observe the changes.
-
Figure 3-1 Graph of frequency response function measurement (normal operation)
-
Figure 3-2 Graph of frequency response function measurement (double hammer case)
Cable breakage or sensor malfunction
If you cannot obtain accurate measurement results, there may be a malfunction.
If you turn the trigger OFF, start the analysis device, and lightly touch the cables or connectors, and observe a large signal or noise in the time-domain waveform, there may be a break in the cable. Also, if the signal magnitude is smaller than previous results, or if there is a lot of noise in the time-domain waveform or power spectrum, there may be a malfunction in the sensor or analysis device.
However, it can be difficult to determine if a graph is normal, so we recommend saving graph images and measurement data from when measurements are taken correctly. Also, when comparing data, always compare the numerical values, not just the shape of the graph. In cases of malfunctions or errors in settings or wiring, the graph shape may be the same, but the values may be smaller.
summary
This time, we introduced several items to check when you don't get the correct results when measuring frequency response functions by hammering. If you're not familiar with the process, try checking all of these items. You should be able to understand the trends in how the graph changes when you make mistakes in settings or operation, or when you miss a beat, through repeated experience.
(Excerpt from the email newsletter issued on June 18, 2015)
