No cabling
Mount the sensor, pair it with the gateway, done — retrofit without shutting the line.
Deployment · wireless
LoRaWAN VibraSens sensors report to a ChirpStack network server running on your own Beckhoff IPC — with real raw-waveform capture and edge AI, just duty-cycled. One host runs the wireless stack, never both.
A host runs the wired stack or the wireless stack — one logic set per IPC. Mixed plants run one host of each; both publish into the same HMI and the same northbound protocols.
Retrofit coverage across assets that were never worth trenching for — with the same edge intelligence.
No cabling
Mount the sensor, pair it with the gateway, done — retrofit without shutting the line.
LoRaWAN
Long-range ChirpStack network server running on your own edge host.
Raw waveform
VibraSens wireless sensors still capture raw data — duty-cycled, fewer points per capture.
Lowest cost/point
No acquisition card per channel, so more assets fit the same budget.
Sensor to plant system, with the network server inside your own plant.
Battery-powered wireless accelerometers, magnet or stud mounted on the asset.
Long-range radio link across the plant — no home-run cabling, no conduit work.
The network server runs on your edge host, so payloads never transit a vendor cloud.
Same Linux stack: alarms, DSP, AI scoring, storage and egress. Physical I/O optional.
The autoencoder runs on every reporting interval; when it flags an asset, pull the logged captures and diagnose with the full post-processing suite.

Long-horizon logging on the node — compression-first storage keeps months of history, so the evidence is still there when the AI points at it.
Point it at the machine. The autoencoder learns that asset's own normal — no expert spectra setup.
Machine Stress Index climbs out of band. Watch → Warning → Danger, locally, in under a second.
Open the captures around the event and run the post-processing that answers why.
Post-processing toolkit — on demand, on the recorded data
FFT spectrum
harmonics, BPFO / BPFI lines
STFT waterfall
how the spectrum evolved
Power spectral density
broadband energy
Demodulation
envelope, early bearing defects
Cepstrum
sideband families, gear mesh
Spectral kurtosis
where the impulsiveness lives
Time domain
the raw waveform itself
Identical services to a wired node — the difference is data density, not features.
RMS, Peak and Kurtosis per reporting interval, checked against ISO 10816-3 bands at the edge.
FFT, STFT, power density, envelope demodulation and spectral kurtosis on duty-cycled raw captures — fewer points than wired, same tools.
The same autoencoder Machine Stress Index (0–100) with Watch / Warning / Danger stages, trained on the IPC.
Local time-series database, outage buffering, and northbound OPC UA, MQTT, Modbus TCP or HTTPS.
Same HTTPS web HMI, RBAC, edge health, live logs and offline update with rollback.
Optional Azure aggregation — $24 per sensor per year, and entirely optional.
Wireless is not a lesser product — it is a different data budget.
Raw waveform
Yes — duty-cycled captures on demand or on schedule
Deep DSP
Yes — fewer points per capture than a wired channel
Continuous streaming
No — reporting intervals, not a permanent stream
Physical I/O reaction
Optional, slower than the sub-second wired path
Acquisition hardware
Gateway only — no per-channel DAQ card
A wireless sensor costs about half the compute of a wired channel, so the same IPC covers twice the assets.
with DSP + AI metrics only
| IPC | Metrics only | DSP + AI | Role |
|---|---|---|---|
| CX9240 | 48 | 29 | Compact DIN-rail embedded PC |
| C6015 | 96 | 57 | Ultra-compact industrial PC |
| C6025 | 144 | 86 | Modular industrial PC |
| C6030 | 192 | 114 | High-performance industrial PC |