4 Checking signal and noise

Getting source characterization first-time right, can be a challenging task. To achieve good quality, it is essential to detect overloads, detached sensors, broken cables, and noise issues.

When performing source characterization, you are interested in modeling the primary excitation, which is the vibration generated by the active component. Any additional excitation not produced by the active component (e.g., noise, test-bench machinery) is a secondary excitation.

When you characterize an active component, you will always model both excitations. Secondary excitations are, of course, unwanted, but always present. So, you must check how much of the secondary excitation is modeled in your calculation and how to reduce it, if needed. To do so, you perform noise measurements (where the primary excitation is switched off) and compare them to the operational measurements.

We recommend performing noise measurements before the operational measurements and compare them with all the operational conditions you want to study. Noise measurements are needed to gain insight in the quality of the Blocked Forces and their validation. You also need to calculate Blocked Forces of the secondary excitations (Blocked Force noise) and compare them with Blocked Forces of the operational measurement.

How to: Perform noise measurements

To perform noise measurements, you need to record secondary excitations at different stages, from the most silent condition (everything switched off) to the noisiest one and then the active component in operation. At each noise stage measurement, you “add” noise and perform a measurement (of about 10 seconds).

For example, when testing an electric motor, we usually measure these stages:

  • Stage 0: Room noise. All components are switched off.
  • Stage 1: Electricity on. The electricity in the room is switched on.
  • Stage 2: Engine on. The engine is switched on.
  • Stage 3: Controller on. The controller is switched on.
  • Stage 4: Engine in idle. The engine is set to idle mode.
  • Operational condition.

How to: Visualize noise measurements

  1. Import all datasets in the Operational Data Module.
  2. Create Master channels and define the channel type.
  3. Multi-select the operationals and Plot them as Spectrum in the Graphs card.
  4. Scroll through the channels to visualize both indicator and validation channels.

How to: Evaluate the signal quality

To evaluate the signal quality, you have to compare the operationals to the noise measurements. You must check the signal at all sensor locations, evaluating both indicator and validation/target sensors.

General rules:

  • There must be a clear separation between the operational and the noise measurements in the frequency range you are interested in. Generally, there must be a difference of at least 10/20 dB between the last noise stage and the operational.
  • Noise peaks must not interfere with the operational measurements, in the frequency ranges of interest.
  • Noise peaks can shift in frequency between idle and operational conditions (e.g., during a runup).
  • If you have good signal at the indicators and worse at the targets, you can calculate Blocked Forces, but they can be validated only in limited frequency bands.

How to: React to bad quality measurements

If you have a good signal-to-noise of your raw data, you are mainly modeling the primary excitation and the contribution of the secondary ones is limited. When this is the case, you can go on and perform additional operational measurements and post-process them.

If you have a bad signal to noise, you are modeling both primary and secondary excitations. In this case, the secondary excitations are too high compared to the primary, and decrease the quality of your model. According to the issue, you can take different actions to improve the quality of the model.


If your primary excitation source is too low, you can:

  • Increase the operational excitation, if possible (e.g., different speed, load, etc.)
  • Change sensors (e.g., use more sensitive sensors).


If your secondary excitation is too high, you must understand whether it is caused by vibrational or electromagnetic noise.

Vibrational noise is any vibration generated by the other components (e.g., tire noise when analyzing the engine) or by the auxiliary machinery (e.g., dyno, pumps, etc.). When they are too high, you can:

  • Move the indicator sensors as close as possible to the source, if possible.
  • Improve the test-bench isolation, if possible.
  • Change the test-bench (e.g., re-design the bench).

Electromagnetic noise is any noise generated by sensors, cables and so on. When this noise is too high, you can:

  • Isolate cables from the noise (e.g., prefer LEMO type connectors over BNC).
  • Isolate sensors from the noise (e.g., use ground-isolated sensors, such as J sensor series from PCB).

How to: Evaluate Blocked Force quality

When you are satisfied with the signal quality of your measurements, you can proceed and start the post-processing. The first step consists of calculating the Blocked Forces of both the operational and the noise measurements and plotting them together for comparison.

The Blocked Forces of the noise measurements must be far lower than the forces measured during operational excitation. If not, the Blocked Forces of the primary excitation are polluted with secondary excitations. In general, all the guidelines provided above for raw data also apply to Blocked Forces.

  • There must be a clear separation between the operational and the noise Blocked Forces in the frequency range you are interested in.
  • Noise peaks must not interfere with the operational measurements, in the frequency ranges of interest.
  • Noise peaks can shift in frequency between idle and operational conditions.
  • If the signal of the operational measurement at the indicator sensors is polluted by noise, the modeled Blocked Forces will be of low quality. This is the case because the Blocked Forces model also includes the secondary excitations, which are very high and “mask” the primary excitation. These Blocked Forces are modeling noise and not the component.
  • If the signal of the operational measurement at the indicators is of good quality, but at the targets, it is polluted by noise, you can have good quality Blocked Forces, but they cannot be validated, or they can be validated but in limited frequency bands.

    Contact us for more VIBES

    Contact our support team or call us on +31 85 822 50 49