PyXis vibration filter settings matter because a vibration sensor can capture far more frequency content than every machine needs for every diagnostic task. A useful condition-monitoring point is not created by bandwidth alone; it is created by matching the sensor, mounting, machine speed, expected fault frequencies and analysis filters to the actual asset.
PyXis is a three-axis industrial vibration sensor with a specified measurement bandwidth from DC to 6 kHz at the -3 dB point. Its Modbus configuration includes adjustable high-pass and low-pass cut-off frequencies, allowing engineers to reject irrelevant low- or high-frequency content before interpreting condition metrics, harmonic arrays and diagnostic indicators.
What high-pass and low-pass filters do
A high-pass filter suppresses frequency content below its cut-off region. A low-pass filter suppresses content above its cut-off region. Used together, they define the frequency band that is most relevant to the machine and the fault modes being monitored.
For condition monitoring, this matters because not every vibration component represents useful machine information. Very low-frequency motion can be influenced by structural movement, process variation or sensor handling. Very high-frequency content may include local impacts, electrical interference, mounting effects or frequencies outside the diagnostic objective.
The goal is not to make the signal look quieter. The goal is to preserve the frequency content required for the maintenance decision while reducing content that does not contribute to that decision.
Start with the machine, not the filter value
Before choosing filter settings, document the machine context. PyXis supports machine-profile information including motor running frequency, RPM context, gear ratio, gear-tooth count and bearing geometry. These parameters help relate measured vibration to known mechanical frequencies.
A practical configuration review should include:
- Running speed: identify normal and minimum/maximum operating RPM.
- Drive train: note direct drive, belt drive, gearbox stages and gear-tooth counts where applicable.
- Bearing geometry: record the bearing information required by the released diagnostic profile.
- Expected faults: decide whether the monitoring objective includes imbalance, looseness, misalignment, gear effects, bearing impacts or another condition.
- Mounting: document sensor location, mounting method and X/Y/Z orientation.
- Operating state: capture normal load and speed ranges before establishing a healthy baseline.
Changing the filter without preserving this context can make historical trends difficult to compare. The filter configuration should therefore be treated as part of the asset record, not as an invisible sensor setting.
How the filters interact with PyXis measurements
PyXis publishes multiple condition-data families over Modbus RTU. These include three-axis acceleration, velocity and displacement metrics; RMS, peak, peak-to-peak and crest factor; harmonic frequency and amplitude arrays; per-axis severity and fault states; RPM and 1X order context; and firmware-dependent advanced diagnostic indicators.
Filtering can influence how these values are interpreted because it changes which portions of the machine spectrum are emphasized or rejected. That is why filter settings should remain consistent when comparing condition trends over time.
For example, an engineer investigating low-order running-speed behaviour may want to preserve the frequencies around the machine’s rotational components. An engineer investigating higher-frequency impulsive behaviour may require a different analysis window. One universal filter configuration should not be assumed to be ideal for every motor, pump, fan, gearbox or conveyor.
Do not confuse sensor bandwidth with the chosen analysis band
PyXis has a specified DC-to-6 kHz measurement bandwidth. That defines the sensor’s measurement capability; it does not mean every application should analyse the entire band with identical settings.
The configured analysis filters are an engineering choice inside that measurement capability. A narrower application-specific band can be useful when it is deliberately selected around the frequencies of interest. Conversely, making the band too narrow can remove information needed to identify a developing condition.
A good commissioning record therefore distinguishes between:
- the sensor’s available measurement bandwidth;
- the configured high-pass and low-pass cut-offs;
- the machine speed and mechanical-frequency context;
- the baseline date and operating condition; and
- the firmware/register-map revision used to configure the sensor.
A practical PyXis filter-commissioning workflow
- Install the sensor rigidly. Mount near a bearing housing or structural load path on a clean, flat surface. Record the X/Y/Z orientation.
- Verify communications and identity. Confirm stable Modbus RTU polling and record device code, hardware revision, firmware revision and serial number.
- Enter machine context. Configure the appropriate machine profile, motor frequency and applicable gear/bearing information.
- Observe the machine under known healthy operation. Use representative speed and load conditions rather than a single unusual operating point.
- Select the analysis band deliberately. Set the high-pass and low-pass values according to the released register map and the frequencies required by the monitoring objective.
- Check the resulting condition data. Review X/Y/Z metrics, dominant harmonic values, RPM/1X context and any enabled diagnostic indices for engineering plausibility.
- Establish the baseline. Capture representative healthy operation across the normal speed and load range.
- Document and freeze the configuration. Store filter values, firmware version, machine profile, mounting information and baseline date with the asset record.
Three filter-setting mistakes to avoid
1. Removing low-frequency content simply because it looks large
Large low-frequency vibration can sometimes be exactly the condition that matters. Before raising a high-pass cut-off, confirm whether the content is structural motion, process behaviour, running-speed vibration or an artefact that can legitimately be rejected.
2. Reducing the low-pass cut-off without checking diagnostic objectives
Higher-frequency vibration can contain information related to impacts, gear behaviour or bearing condition. Reducing the upper analysis band may improve signal focus, but it can also remove useful diagnostic evidence.
3. Changing filters after baseline capture without resetting the comparison logic
A trend is meaningful only when the measurement configuration is comparable. If mounting, orientation, filters, machine profile or firmware changes, record the change and determine whether a new baseline is required.
Use the released Modbus register map
The PyXis datasheet identifies high-pass and low-pass cut-off frequency as writable analysis-filter parameters, but detailed engineering integration must use the register map that matches the installed firmware.
Do not infer engineering units, scaling, signedness, enumeration values or feature availability from a register name alone. The released firmware-specific register map and release note are the authoritative sources for implementation.
This is especially important when commissioning multiple sensors over time. Recording the firmware revision and register-map revision makes later maintenance far more reliable than relying on an undocumented configuration copied from another machine.
Filter settings are part of condition-monitoring traceability
Condition monitoring works best when every alarm can be traced back to the measurement configuration that produced it. PyXis supports identity, machine context, filtering, thresholds, baseline information and processed condition outputs through its Modbus data model. That makes it possible to keep the sensor configuration aligned with the maintenance record.
For related engineering guidance, see what the PyXis 6 kHz vibration bandwidth means in practice and how harmonics, order tracking and machine speed improve vibration interpretation.
Planning a PyXis deployment?
Review the Tachyonic product range, use the product selector and check the technical resources before design freeze.
Need help defining the monitoring band for a motor, pump, fan, gearbox or conveyor? Send the machine speed range, asset type, mounting location and maintenance objective for an engineering review.
Technical basis: PyXis 222x Series datasheet TTN-DS-2026-002, Rev 08, issued 22 July 2026, and the matching firmware-specific Modbus register documentation. Confirm the hardware, firmware and released register map before final commissioning or field changes.
