One runtime, one box
Acquisition, DSP and the anomaly model run in the same runtime on the same IPC, rather than splitting signal conditioning, analysis and alarm logic across separate modules.
Motors · pumps · fans · compressors · gearboxes
Catch bearing failures weeks before they cause downtime
Local FFT, envelope analysis and AI stress scoring on the machine itself — no cloud round-trip needed to raise an alarm.
0cloud round-trips to raise an alarm
Built for the reliability engineer
The challenge
Built for the reliability engineer.
- Bearing wear, imbalance and misalignment are the leading causes of unplanned rotating-equipment failure.
- Vibration analysis is often outsourced to periodic manual inspection rounds — gaps of weeks between readings miss developing faults.
- Full DSP-capable systems are traditionally expensive, so many plants only instrument their most critical assets.
How the system solves it
Only the capabilities that apply to this use case.
- 01
IEPE vibration
25.6 kS/s · raw waveform on IPC
- 02
Edge thresholds
RMS · Peak · Kurtosis → relay
- 03
On-device DSP
FFT · envelope · RPM detection
- 04
AI Stress Index
0–100 · Watch / Warn / Danger
Alarm to I/O survives without DSP or AI — basic protection never depends on analytics
- Metrics & Alarms tier
- RMS, Peak and Kurtosis thresholds drive an immediate physical alarm — tower light or relay — with no AI or DSP service required.
- High-Speed DSP tier (optional, same hardware)
- Raw waveform capture, on-demand FFT, spectrogram, envelope analysis, band-pass filtering and automated RPM detection.
- Edge AI tier (optional)
- Autoencoder-based Machine Stress Index (0–100) with Watch / Warning / Danger staging, trained on the asset's own normal-operating baseline.
- Modular on one runtime
- Start with basic alarms on a CX9240-class controller and add DSP or AI later on the same install where CPU headroom allows. Headroom per tier is confirmed during sizing, so you know before you buy whether a later upgrade fits.
Why it wins here
Architectural differences, not slogans.
Protection survives the analytics
The alarm-to-I/O path does not depend on the AI or DSP layer being present or healthy — basic protection still works while advanced analytics are disabled or the model is retraining.
Your data stays yours
Raw waveforms live in the edge database and export over OPC UA, MQTT, Modbus or HTTPS. No proprietary cloud is required to read your own signals.
Deployment path
Variant
Wired / TwinCAT-EtherCAT
Hardware tier
Mid-tier IPC (C6015 or C6025)
Why
Enough headroom for DSP, with room to add AI if the budget allows. EtherCAT DAQ terminals daisy-chain along the machine line instead of home-running every cable to a cabinet.
Proof & validation
Measured on your machines, published with your sign-off
Figures are published only after a customer pilot is measured and signed off — lead time from first detection to end of life, false-alarm rate across a run, and downtime avoided — alongside the tested configuration. Until then, you get a written test plan, not a borrowed statistic.
Explore the other use cases
Remote & hard-to-wire assets
Monitor equipment kilometers from the control room — without a wiring project
Read use caseMachine protection & interlocks
Trip the machine before damage happens — deterministically, locally
Read use caseMulti-site fleet standardization
One condition-monitoring platform, right-sized for every site in your fleet
Read use caseBrownfield retrofit
Add condition monitoring without ripping out what's already there
Read use case