Machine Health Monitoring
Machine Health Monitoring (MHM) gives you continuous vibration-based condition monitoring for rotating industrial equipment. Batteryless Eversensors attached to your machinery measure vibration every ~60 seconds and automatically alert your team when levels cross the thresholds you define — catching developing problems before they become failures.
Supported Sensors
| Sensor | Notes |
|---|---|
| MVM Eversensor 1 (Silverglade 1.0) | First generation vibration sensor |
| MVM Eversensor 2 (Silverglade 2.0) | Adds MaxLevels data and electrical measurements |
| MHM 1.0 | Fluke MVP-based platform |
Each sensor reports every ~60 seconds.
What the Platform Measures
| Measurement | Unit | Description |
|---|---|---|
| Vibration velocity | mm/s | Overall vibration intensity — primary alarm metric |
| Vibration acceleration | mg | High-frequency shock events |
| Spectral peaks (X/Y/Z) | Hz | Frequency-domain analysis for fault characterization |
| Electrical field | µT (micro-Tesla) | Electromagnetic monitoring (MVM Eversensor 2) |
| MaxLevels | — | Peak vibration data between readings (MVM Eversensor 2) |
Alarm System
You configure vibration thresholds per machine. When a reading crosses a threshold, the platform updates the machine's alarm state and notifies your team.
Alarms fire on state transitions, not on every crossing. If a machine is already in an alarm state, it won't generate a duplicate alert — only a change in severity level triggers a new notification.
See Alarm Thresholds & States for configuration details.
Process State Detection
The platform automatically determines whether each machine is running or stopped, based on vibration level. This gives you utilization context alongside your alarm data — for example, you can distinguish between a low-vibration alarm and a machine that was simply off.
- Default ON threshold: 0.508 m/s vibration velocity (configurable per machine)
- Minimum event duration: 5 minutes (filters out transient noise)
Full Time Waveform (FTWF) Data
For advanced vibration analysis, the platform captures full time waveform data with frequency-domain (FFT) decomposition. This lets vibration analysts diagnose specific fault types — bearing wear, imbalance, misalignment — from the spectral signature.
The Spectral Analysis & Waterfall Charts view in Insights provides a visual interface for this data. FTWF data is also available programmatically via the REST API at /v2020-07/machines/{id}/timeseries.
See Sensor Mounting & Troubleshooting for how to physically mount a sensor and resolve common installation issues.
Sensor Health Monitoring
Each sensor's health is tracked in three states:
| State | Meaning |
|---|---|
| Good | Data arriving normally |
| Sleep | Sensor is conserving energy (low VCAP charge) |
| Offline | No recent data received |
Data Retention
| Data | Retention |
|---|---|
| Raw sensor readings | ~1.5 years (548 days) |
| Sensor aggregates | ~6 months (180 days) |
| Alarm history | ~1.5 years (548 days) |