Understanding High-Pass Filters

Devices

Portable balancer & Vibration analyzer Balanset-1A

€1,975.00 + VAT (if applicable)

Parts

Vibration sensor

€90.00 + VAT (if applicable)

Parts

Optical Sensor (Laser Tachometer)

€124.00 + VAT (if applicable)

Devices

Balanset-4

€6,803.00 + VAT (if applicable)

Parts

Magnetic Stand Insize-60-kgf

€46.00 + VAT (if applicable)

Parts

Reflective tape

€10.00 + VAT (if applicable)

Devices

Dynamic balancer “Balanset-1A” OEM

€1,735.00 + VAT (if applicable)

A high-pass filter (HPF) is a frequency-selective signal-processing element that allows vibration components above a specified cutoff frequency to pass through while attenuating components below the cutoff. In vibration analysis, high-pass filters remove low-frequency vibration (from unbalance and misalignment) so the analyst can focus on high-frequency content (from bearing defects, gear mesh, and electrical sources), and they eliminate sensor mounting-resonance artefacts and DC offsets. It is the mirror image of a low-pass filter.

High-pass filters are fundamental to envelope analysis, anti-aliasing systems, and general signal filtering, enabling extraction of diagnostic information from a chosen frequency range while rejecting unwanted low-frequency components that would otherwise mask or overwhelm the signals of interest.

1. Filter Characteristics

Three parameters define how any high-pass filter behaves: its cutoff frequency, its slope, and its underlying design type.

• Cutoff frequency (f_c): the frequency at which the filter response drops to −3 dB (70.7% of the passband amplitude). Below f_c frequencies are progressively attenuated; above f_c they pass with minimal loss. The cutoff is chosen according to the application and the frequency content of interest.
• Filter slope (roll-off rate): the rate of attenuation below the cutoff, expressed in dB per octave or dB per decade. A 1st-order filter rolls off at 6 dB/octave (20 dB/decade) — a gentle slope; a 2nd-order at 12 dB/octave (40 dB/decade) — moderate; a 4th-order at 24 dB/octave (80 dB/decade) — steep. Higher orders give a sharper transition and better rejection but are more complex to implement.

The filter type determines the trade-off between sharpness and fidelity:

• Butterworth: maximally flat passband response.
• Chebyshev: sharper cutoff, but with ripple in the passband.
• Bessel: the best time-domain behaviour, with minimal phase distortion.
• Elliptic: the sharpest transition of all, but with ripple in both passband and stopband.

2. Applications in Vibration Analysis

Bearing defect detection

This is the most common application. A cutoff of typically 500–2000 Hz removes low-frequency unbalance and misalignment vibration, leaving the high-frequency impact signals generated by bearing damage. It is the first stage in envelope-analysis processing, which then demodulates those impacts to reveal the bearing fault frequencies.

Integration to velocity or displacement

When integrating acceleration to velocity or displacement, an HPF set at 2–10 Hz removes the DC offset and very low frequencies that would otherwise integrate into large drift errors. This step is essential for accurate low-frequency integration.

Sensor mounting-resonance elimination

An accelerometer mounting resonance — typically 3–10 kHz for a magnetic mount — can distort readings. A high-pass (or band-limiting) filter removes this artefact so the measurement represents true machine vibration rather than a sensor effect. Sound sensor mounting practice complements the filtering.

DC-offset removal

A high-pass filter with a very low cutoff (0.5–2 Hz) strips out the DC component of a signal. This is necessary for proper signal processing, preventing FFT errors and integration drift.

3. Practical Implementation

Analog versus digital filters

Analog high-pass filters are hardware circuits inside the signal-conditioning chain. They operate in real time, handle anti-aliasing and sensor conditioning, and have fixed characteristics once designed. Digital high-pass filters are software-based and applied in post-processing; their cutoff and order are adjustable, and they can be applied or removed after data collection. Modern analysers offer multiple digital filter options so the same record can be examined several ways.

Selecting the cutoff frequency

For bearing analysis, set f_c below the lowest bearing fault frequency — typically a 500–1000 Hz cutoff. This removes 1×, 2× and gear-mesh components while passing the bearing fault frequencies (typically 50–500 Hz) and their high-frequency modulation. For integration, set f_c at 2–5× the lowest frequency of interest: too low allows drift, too high attenuates valid low-frequency components, with 2–10 Hz being typical for general integration.

4. Effects on Measurements

A high-pass filter changes the signal in three ways the analyst must keep in mind:

• Amplitude effects: frequencies below the cutoff are reduced, very low frequencies essentially eliminated, and frequencies well above the cutoff left unaffected; the transition region shows a gradual reduction rather than a hard edge.
• Phase effects: all filters introduce a frequency-dependent phase shift, which can alter the shape of the time-domain waveform. Bessel filters minimise this phase distortion, which matters when waveform timing is being interpreted.
• Waveform effects: the filter removes low-frequency baseline variations and centres the time waveform around zero, which can change its apparent character. It is therefore important to know what filtering was applied when interpreting a waveform.

5. Combining High-Pass Filters with Other Filters

High-pass filters rarely act alone. Pairing a high-pass with a low-pass produces a band-pass filter: the HPF blocks low frequencies, the LPF blocks high frequencies, and the combination passes only a middle band — exactly the selectivity needed to isolate a specific frequency range. In a full multi-stage processing chain, anti-aliasing (low-pass) is applied before digitisation, a high-pass removes DC, and a band-pass conditions the signal for envelope analysis; this sequential filtering builds up complex signal conditioning from simple stages. Where a single narrow component must instead be rejected, a notch filter is the complementary tool.

6. High-Pass Filtering in Field Measurement

In day-to-day field work, the right high-pass setting is what makes a faint bearing defect visible beneath dominant rotor vibration. A portable two-channel analyser such as the Balanset-1A measures the broadband signal needed for both balancing and diagnostics, and applying a high-pass stage before envelope analysis lets the engineer separate early bearing defects from the large 1× unbalance response on the same machine. Understanding high-pass characteristics — cutoff frequency, filter order, and the effects on amplitude and phase — is therefore essential for sound bearing analysis, reliable signal integration, and any task that calls for frequency-selective measurement.

---

← Back to Main Index