Understanding Low-Pass Filters

ডিভাইস

Portable balancer & Vibration analyzer Balanset-1A

€1,975.00 + ভ্যাট (যদি প্রযোজ্য হয়)

অংশ

Vibration sensor

€90.00 + ভ্যাট (যদি প্রযোজ্য হয়)

অংশ

Optical Sensor (Laser Tachometer)

€124.00 + ভ্যাট (যদি প্রযোজ্য হয়)

ডিভাইস

Balanset-4

€6,803.00 + ভ্যাট (যদি প্রযোজ্য হয়)

অংশ

Magnetic Stand Insize-60-kgf

€46.00 + ভ্যাট (যদি প্রযোজ্য হয়)

অংশ

Reflective tape

€10.00 + ভ্যাট (যদি প্রযোজ্য হয়)

ডিভাইস

Dynamic balancer “Balanset-1A” OEM

€1,735.00 + ভ্যাট (যদি প্রযোজ্য হয়)

A low-pass filter (LPF) is a frequency-selective signal-processing element that lets vibration components below a chosen cutoff frequency pass through while attenuating components above it. In vibration analysis it performs three jobs that an analyser cannot do without: anti-aliasing (stopping false frequencies from appearing in digital data), noise reduction, and isolating the low-frequency region for focused study. It is the mirror image of the high-pass filter, and the two are the building blocks of every other signal filtering scheme.

Low-pass filters are arguably the most widely used filters in vibration instrumentation. One sits in front of the converter in every digitising system as a mandatory anti-aliasing filter, and the same function is offered as an analysis tool for smoothing data, stripping high-frequency noise, and concentrating on low-frequency phenomena. Understanding how they shape a signal is therefore essential to trusting any spectrum you read.

1. Filter Characteristics

Cutoff Frequency (f_গ)

• Definition: the frequency at which the filter response has dropped to −3 dB, i.e. 70.7% of the passband amplitude.
• Below f_গ (passband): frequencies pass with minimal attenuation.
• Above f_গ (stopband): frequencies are progressively attenuated.
• Transition band: the region around f_গ where attenuation steadily increases.

Filter Order and Roll-Off

The order of a filter sets how sharply it transitions from passband to stopband:

• 1st order: 6 dB/octave (20 dB/decade) — gradual roll-off.
• 2nd order: 12 dB/octave (40 dB/decade) — moderate.
• 4th order: 24 dB/octave (80 dB/decade) — steep.
• 8th order: 48 dB/octave (160 dB/decade) — very steep.
• Higher order: a sharper transition and better stopband rejection, at the cost of more phase shift and a longer transient response.

Filter Response Types

The same cutoff and order can be realised with different mathematical shapes, each trading flatness, sharpness, and phase behaviour:

• Butterworth: maximally flat passband with no ripple.
• Chebyshev: a sharper cutoff, accepting ripple in the passband.
• Bessel: linear phase, which means minimal waveform distortion — the right choice when the shape of the time waveform matters.
• Elliptic: the sharpest possible transition, with ripple in both passband and stopband.

2. Primary Applications

Anti-Aliasing (most critical)

This is the function no digitiser can omit. Without it, frequencies above the Nyquist limit fold back and appear as false peaks — the phenomenon of aliasing.

• Purpose: block frequencies above the Nyquist frequency (half the sample rate).
• Requirement: it must act before analog-to-digital conversion — software cannot remove an alias after the fact.
• Typical cutoff: 0.4–0.8 × (sample rate / 2).
• Steepness: typically 8th order or higher for good aliasing rejection.
• Consequence of neglect: inadequate anti-aliasing creates false spectral peaks that mimic real faults.

নয়েজ রিডাকশন

• Removes high-frequency electrical noise.
• Filters out sensor-cable noise.
• Smooths data for trending.
• Improves the signal-to-noise ratio for the low-frequency components of interest.

Frequency Range Limitation

• Focuses the analysis on the frequency range of interest.
• Example: a 0–100 Hz analysis for low-speed machinery.
• Removes irrelevant high-frequency content.
• Reduces data-processing and storage requirements.

Integration Preparation

• Applied before integrating acceleration to velocity.
• Removes very high frequencies — noise that integration would otherwise amplify.
• Typical cutoff: 1000–5000 Hz depending on the application.
• Prevents the noise amplification that plagues uncontrolled integration.

3. Selecting the Cutoff Frequency

Anti-Aliasing Applications

• Rule: f_গ = 0.4 × sample rate (conservative) to 0.8 × sample rate (aggressive).
• Example: a 10 kHz sample rate gives f_গ = 4000 Hz.
• Criterion: stopband attenuation greater than 60 dB at the Nyquist frequency.

Analytical Applications

• Set f_গ just above the highest frequency of interest.
• For low-frequency analysis (0–200 Hz): f_গ = 200–300 Hz.
• For unbalance only (the 1× component): f_গ = 5–10× running speed.
• Always leave margin for the filter transition band.

নয়েজ রিডাকশন

• Identify the noise frequency range from the spectrum.
• Set f_গ to pass the signal frequencies while rejecting the noise frequencies.
• Balance noise removal against signal preservation.

4. Effects on Measurements

Amplitude Domain

• Passband: minimal amplitude change, typically less than 0.5 dB.
• Stopband: strong attenuation, 40–80 dB or more.
• Overall level: the filter reduces the overall vibration reading if significant high-frequency content was present.

Time Domain

• The waveform is smoothed as high-frequency variations are removed.
• Sharp edges and spikes are rounded.
• Transient response (filter ringing) can affect the waveform shape.
• Phase distortion can alter how the waveform is interpreted.

Frequency Domain

• The spectrum shows reduced amplitudes above the cutoff.
• High-frequency peaks are diminished or eliminated.
• The noise floor is lowered if the noise was high-frequency.

5. সাধারণ সমস্যা এবং সমাধান

Inadequate Anti-Aliasing

• Symptom: false low-frequency peaks in the spectrum.
• Cause: high frequencies folding back below Nyquist.
• সমাধান: use a steeper filter, increase the sample rate, and verify the filter is actually functioning.

Cutoff Too Low

• Symptom: valid high-frequency signals are attenuated.
• Example: bearing fault frequencies reduced by an overly aggressive LPF.
• সমাধান: increase the cutoff frequency or use a gentler filter slope.

Filter Artifacts

• Ringing: oscillations in the time domain caused by a sharp filter cutoff.
• Phase distortion: waveform shape changes arising from phase shifts.
• সমাধান: use a Bessel filter for critical waveform applications where phase linearity matters.

6. Complementary Filters

Low-Pass vs. High-Pass

• Low-pass: passes low frequencies, blocks high.
• High-pass: passes high frequencies, blocks low.
• Complementary: used together to form a band-pass filter.

Band-Pass Filter

• A combination of high-pass plus low-pass stages.
• The resulting band-pass filter passes only frequencies within a specified band.
• It rejects content both below and above that band.
• This is the front end of envelope analysis, where a band around a bearing’s structural resonance is isolated before demodulation.

7. Where the Low-Pass Filter Fits in the Field

On a digital field instrument the low-pass filter is mostly invisible — it does its anti-aliasing work silently inside the acquisition chain — yet it underpins the trustworthiness of every reading. A portable two-channel analyser such as the ব্যালানসেট-১এ bandlimits each accelerometer channel before sampling, so the FFT it computes for balancing and diagnostics is free of aliased peaks across its working range. With the spectrum clean, the analyser can resolve the 1× amplitude and phase needed to balance a rotor and report a true residual unbalance, rather than chasing a ghost frequency created by poor filtering.

Low-pass filters are fundamental components of vibration measurement systems, serving essential functions from anti-aliasing protection to noise reduction and frequency-range selection. Understanding their operation, choosing the cutoff frequency correctly, and appreciating their effect on the measured signal are crucial for accurate analysis and for avoiding measurement artifacts in digital data acquisition.

---

← প্রধান সূচিতে ফিরুন