Configuring Balluff CMTK Vibration Monitoring for Repeatable Data

Rerouting zip ties or adding a laser sensor will not, by itself, make the Balluff CMTK demo’s vibration readings repeatable. The deciding condition is whether measurements come from the same point on the bearing housing under comparable speed and load; a clean-looking cable bundle cannot correct a changing mechanical measurement path.

Why cable rerouting and sensor swaps miss repeatability

The demo uses a Balluff CMTK box and vibration sensors to monitor motor, pump, or gearbox condition. Its graph is described as a simplified Grafana view. The enclosure is space-constrained, with horizontal routing prioritized and zip ties used rather than routing through the bottom of the Panduit over the DIN rails.

Those layout choices raise a wiring inspection question, but they do not establish that the measured vibration is wrong or that the wiring is unsafe. A visual review cannot determine cable type, sensor connection details, grounding, shielding, or whether a cable is mechanically protected. Check those items against the documentation for the installed sensor and CMTK hardware before approving the wiring.

Likewise, adding a laser measurement does not automatically improve a trend. A contact sensor measures motion at its mounting point; a non-contact reference can compare motion without sharing that contact point, but the measurements answer different questions unless their location, direction, timing, and machine operating state are aligned.

Balluff CMTK measurements and the limits of a trend graph

A vibration sensor and monitoring box turn machine motion into data that can be trended. A Grafana display can make changes visible over time, and the demo is described as capable of becoming more elaborate. A trend is useful for observing change, but the plotted quantity and its context determine what the display can support.

RMS, velocity, temperature, and alarms are useful trend fields when the installed instrumentation provides them. Record the actual sensor outputs and units rather than assuming every channel is available. A plotted value without speed, load, sensor location, or operating-state context can make a normal state change look like deterioration—or conceal a developing change.

Use the following register to define what the monitoring system can decide. Read numerical alarm limits from the applicable machine, sensor, or monitoring-system documentation; the demo description supplies no numeric limit.

Sensor mounting and measurement-point consistency

Mounting consistency is the first corrective action for repeatability. The next iteration was specifically identified as needing a consistent sensor mount and the same point on the bearing housing. This matters because the measured motion depends on where and how the sensor couples to the machine; moving the sensor changes the measurement path as well as the location.

1. Choose and document the bearing-housing measurement point for each monitored machine. Use a location that can be found again, not merely a general description such as “near the motor.”
2. Document sensor orientation and mounting method from the sensor manufacturer’s instructions. Reinstall using the same method whenever a sensor is removed or replaced.
3. Keep the cable route clear of forces that could disturb the sensor connection or mounting. Inspect tie points and connectors during a repeatability check; do not treat tie count or visual neatness as a measurement-quality test.
4. Record the sensor identity, point, orientation, and machine identity with each measurement set so future comparisons refer to the same physical installation.

If the bearing housing, frame, guard, or base may move differently, a single sensor point may not describe the whole assembly. A multi-point comparison can reveal whether components move together or whether the selected point gives only a local view.

Speed, load, and operating-state tags

Speed and load tags turn an unqualified trend into a state-aware comparison. Vibration changes with operating conditions, so a threshold applied identically during unlike states can generate misleading alarms. The demo’s next-iteration suggestions include speed/load tags and separate thresholds by operating state; implement those before interpreting a trend as a fault progression.

1. Identify the operating states that matter for the motor, pump, or gearbox being monitored.
2. Capture speed and load with the vibration reading, using the machine’s available signals or recorded operating data.
3. Group baseline and follow-up readings by comparable state. Do not compare a low-load sample directly with a materially different load or speed and call the difference a fault.
4. Assign and document thresholds by operating state using the relevant machine or instrumentation criteria. Record the source of each limit and the alarm action.

If the system cannot provide a trustworthy speed or load tag, mark the state as unknown and avoid presenting the sample as directly comparable to state-qualified baseline data. The missing context is a data-quality limitation, not a reason to invent a threshold.

Trend metrics versus waveform and spectrum evidence

Grafana is suitable for trending metrics such as RMS, velocity, temperature, and alarms when those signals are available and correctly labeled. Trend displays answer whether a measured quantity changes. They may not identify whether the cause is imbalance, looseness, misalignment, a bearing defect, or process-related vibration.

When the question is fault discrimination, preserve access to raw waveform or spectrum data. Compare like operating states and use before/after data from a known fault or known corrective condition where available. A spectrum provides frequency-domain evidence; the raw waveform retains signal behavior that a single trend statistic can hide. Do not infer a particular fault solely from an RMS or velocity change when the signal detail and operating conditions have not been examined.

Temperature is a useful additional trend only when the installation measures it. It supplies thermal context, not a substitute for vibration data or proof of a vibration fault. Keep its unit and measurement source explicit in the Grafana labels and alarm configuration.

Alternative architectures and non-contact comparisons

Other approaches named for this class of monitoring differ in integration and measurement method; they are not automatic drop-in replacements for the CMTK demo.

A non-contact reference is useful when you need to test whether the bearing housing, frame, guard, and base move together—or whether a contact sensor location tells only part of the story. Align the reference measurement with the contact measurement in location, direction, operating state, and time before comparing results. A laser measurement is a comparison method, not a substitute for repeatable sensor mounting and state-tagged data.

Repeatability and alarm verification sequence

Verify the measurement process before trusting the alarm. A repeatable baseline makes later changes interpretable; without it, dashboard precision can exceed measurement certainty.

1. Inspect the installed wiring and sensor mounting against the applicable product instructions. Record the point, orientation, connection condition, and cable routing.
2. Collect repeated measurements at the same point with the machine in the same documented speed/load state. Compare the same measured channels and units.
3. Check that Grafana displays the intended sensor, units, timestamp, speed/load tags, and operating-state label. Confirm that missing or unqualified state data is visible rather than silently treated as equivalent.
4. Review trends for change, then inspect raw waveform or spectrum when the task is to distinguish a fault mechanism. Preserve before/after records for a known condition or correction when available.
5. Exercise the configured alarm logic against the intended operating states and documented limits. Confirm that each threshold maps to the right state and that the displayed alarm corresponds to the configured channel.

A repeated reading that changes substantially under nominally matched conditions sends you back to mounting, location, operating-state tagging, and data handling before you tune alarm thresholds. If repeated data are stable but the signal indicates a developing condition, retain the waveform or spectrum and investigate the machine condition using the appropriate diagnostic process.

Frequently asked questions