5 Checking overloads

High-quality test-based models require good signal quality of the measured data. Overloads are detrimental to the signal quality. Thus, you have to make sure to spot and exclude all overloads from the data of your test-based model.

Types of overloads

When the input signal is higher than what the data acquisition system or sensor can measure, there is an overload. There are two types of overloads that can affect your measurement:

• Quantization overloads: when the measured voltage exceeds the defined quantization range of the acquisition system.
• Sensor overloads: when the voltage measured by the sensor exceeds its measurement range.

Overloaded data are bad for your measurement, so you have to spot them directly while you are measuring. This is not always straightforward, so you can measure data with overloads without noticing. For this reason, you must also check your data after measuring all the excitations of each VP. Integrating this check into your measurement process ensures high-quality test-based models. (link)

Quantization overloads

You have a quantization overload when the voltage measured by the acquisition system exceeds its defined quantization range. These overloads are spotted and marked by DIRAC and automatically excluded.

Reacting to quantization overloads

To avoid (or limit) quantization overloads, you must increase the quantization range. We recommend to set it to 10 V.

To change the quantization range you have to:

1. Open the DAQ window.
2. Go to 4 Channel settings.
3. Change the range to 10 V at each channel.

Sensor overloads

If sensors have a measurement range that is smaller than the quantization range, they can be overloaded before the system detects an overload. Overloaded sensors lead to bad data, so you must detect and delete them. You have to check if you get sensor overloads while measuring. However, they can be difficult to detect, so you must check the data also as soon as you finish all the measurements of a VP.

Spotting sensor overloads while measuring

The sensors closer to the impact location are the ones that will most likely present overloads. When overloaded, the sensor channel will measure a high voltage. The signal will then slowly decrease until it reaches the zero range. The slow exponential decay of an overloaded signal in the time domain corresponds to a curve with ski-slope shape in the frequency domain, that you can visualize in the Graphing area. When a channel is overloaded, it will turn orange on the Hardware card.

To spot senor overloads while measuring:

1. Select the sensor that is closest to the excitation.
2. Impact with the hammer at that location.
3. Check if one or more channels are overloaded in the Hardware card.
4.  Check if the FRF presents the ski-slope shape in the Graphing card.
5. Wait for the signal to go back to the zero range in the Scope card.

Reacting to sensor overloads while measuring

When after an impact you notice that the sensor got an overload, before doing the next impact, you have to:

1. Delete the measurement.
2. In the Hardware card, right-click on the overloaded channel and select Reset overloads.
3. Wait for the signal in the Scope card to go back to the zero range.

If you keep experiencing overloads when impacting the same spot, consider:

• Using a hammer with a softer tip.
• Impacting with less strength.

Spotting sensor overloads after measuring

It could be hard to spot all sensor overloads while measuring. Once you finished measuring all FRFs of one VP, check if you missed any overload. You can spot the sensor overloads by looking at the magnitude of the FRFs in Analyze.

To spot sensor overloads in Analyze, you have to:

1. Select the Measured preset.
2. Set the mode to Magnitude.
3. Set the averaging to Maximum.
4. Set the frequency to a low range (e.g. 20 Hz).
5. Zoom into the magnitude scale to show the maximum values.
6. Spot the dark blue cells and click on them.
7. In the graphing area, check if the curve has the ski-slope shape.

Reacting to sensor overloads after measuring

After spotting sensor overloads, you must delete the affected data. You then have to impact again at all excitation locations where you had overloads.

To avoid overloads you can:

• Use less sensitive sensors: for the same excitation, it will output a signal with a lower voltage.
• Use less energy when impacting, without compromising on the coherence.
• Do a roving sensor measurement campaign. By impacting close to both the indicator and target sensors you can impact with lower energy and avoid overloads.