This is what lands in your technician's hand. Not a spectrum to interpret — the machine, the component, how urgent, how sure, and eighteen days to do something about it.
No line shutdown to install · No inbound firewall rules · On-premise available
One vendor. One contract. No integration project.
Four stages, one vendor, one contract. No second product to buy to make the first one useful, and no integration project standing between the sensor and the person who fixes the machine.
A sealed, battery-powered unit magnetically or stud-mounted on the bearing housing. Three modalities in the same device, so a diagnosis has corroboration instead of one noisy channel.
Doing DSP on the sensor protects battery life. Doing the diagnosis on the sensor would freeze your fault model in firmware. We split it deliberately — so a model improvement reaches every asset you own overnight, without a truck roll.
One sensor. One receiver. Nothing else.
A sealed sensor on the machine and a rugged receiver on the wall. Nothing else — no wiring, no panel space, no PLC tags.
Every diagnosis arrives as failure type, affected component, severity and confidence — plus a recommended action. Severity decides who gets told and how loudly. The alert links to the asset, never to a dead-end inbox.
See the productEleven fault modes. Two honest gaps.
Most vendors will tell you they detect “anomalies”. Your reliability engineer needs to know which fault modes, sensed how, on which equipment. This is the whole list, including the gaps.
| Fault mode | How it is detected | Equipment | Status |
|---|---|---|---|
| Outer-race defect | Vibration — band energy at BPFO + harmonics | Motors, pumps, fans, gearboxes | Detected |
| Inner-race defect | Vibration — BPFI + shaft-rate sidebands | Motors, pumps, fans, gearboxes | Detected |
| Rolling-element defect | Vibration — BSF + cage-rate modulation | Motors, pumps, fans, gearboxes | Detected |
| Cage / train defect | Vibration — FTF, sub-synchronous energy | Motors, pumps, fans, gearboxes | Detected |
| Imbalance | Vibration — dominant 1× radial, stable phase | All rotating equipment | Detected |
| Misalignment | Vibration — elevated 2× with axial component | Coupled drives | Detected |
| Mechanical looseness | Vibration — harmonic series, raised noise floor | All rotating equipment | Detected |
| Overheating | Temperature — absolute trend and rate of change | All rotating equipment | Detected |
| Cavitation | Vibration — broadband high-frequency energy | Pumps | Detected |
| Gear-mesh defect | Vibration — GMF + shaft-rate sidebands | Gearboxes | Detected |
| Unexpected stop / idling | Magnetometer — running state and shaft speed | All rotating equipment | Detected |
| Electrical faults rotor bar, eccentricity, demagnetisation | Requires stray-flux ESA — out of scope for the current magnetometer | Motors | Not yet |
| Lubrication starvation as a distinct diagnosis | Currently surfaces as a bearing-mode escalation, not its own class | All rotating equipment | Not yet |
One application. Four roles. No separate analyst tool.
One web application, four roles, no separate analyst tool. Asset-first by design — people come in to look at a machine, so every alert links straight to the asset it came from.
Grouped by the plant areas you define and sorted by what will actually stop production. An asset inherits the worst severity of any sensor on it, so nothing hides behind an average.
See it on your own assetsOne line each. An abstaining model is shown honestly as unknown and routed to an engineer, rather than being quietly hidden or dressed up as a real fault.
How we decide severityTwo boxes. One outbound connection. Nothing to open.
This is the part your IT group reviews, and it is on this page rather than behind a sales call.
Isolated by tenant. Audited by default.
We would rather tell you precisely which controls exist today than show you a wall of badges. Here is the posture, in the language your security review will use.
Documented API. Live streams. No lock-in.
TrackBeat is the condition-monitoring layer. It is not trying to replace your CMMS, your historian or your ERP — it is trying to give them a reason to raise the right work order.
An OpenAPI-described REST interface over assets, sensors, readings, insights and alerts — the same API our own application uses — plus a WebSocket feed for live readings and MQTT at the transport layer. No private endpoints.
Over 1,100 shared equipment templates carry the right geometry, speed and bearing data the moment an asset is commissioned, with bearing fault frequencies resolved automatically from the designation. CSV import means a few hundred machines is a spreadsheet, not a fortnight of typing.
RoadmapTurnkey connectors are in development. Today the documented API supports the same integration, built by you or by us during onboarding.
Week zero to day ninety.
Condition monitoring has to learn your machines before it can judge them. We would rather set that expectation now than promise insight on day one and lose your confidence in week three.
We pick a line together, list its assets by criticality, and agree in writing what success looks like — faults caught, at what lead time, at what false-alarm rate.
Sensors mount externally on bearing housings; the receiver goes on a wall. No shutdown, no wiring, no controls work. Your technician commissions each sensor in the app.
The platform observes each asset across its real duty cycle — every shift, load state and changeover — to learn what normal looks like on your floor, not on a test rig.
Insights start firing. Technicians log what they found when they opened the machine, and each label feeds retraining. This is where the false-alarm rate drops.
We measure against the week-zero document: what we caught, how early, what we missed, what we called wrong. That evidence drives the plant-wide rollout decision.
Every claim here points at the part of this page that proves it.
Anyone can write that they are transparent. These are the specific decisions that cost us something — and each one links to the section of this site where you can verify it rather than take our word.
Urgency is anchored to the published ISO 10816/20816 velocity zones — A, B, C, D — not to a proprietary 0–100 health score only we can interpret.
When a signature is out of distribution we return unknown and route it to an engineer, rather than snapping to the nearest familiar fault.
Electrical faults and lubrication starvation are listed as not yet detected, in the same table as everything we can.
Users are free across all four roles. A monitoring platform only works when the manager, the engineer and the technician are looking at the same screen.
We hold no SOC 2 or ISO 27001 attestation today, and the security section says so in those words rather than showing a badge wall.
Your numbers. Your arithmetic. Not someone else's ROI study.
Every vendor in this category will quote you a return figure from a commissioned study of a model customer you have never met. We would rather hand you the calculator and let you put your own plant into it.
That last one is the honest variable. Not every failure gives warning — a seized bearing from a lubrication error can go in hours. Rotating-equipment faults that develop over days are what this platform is for, and you should set that share yourself rather than let us set it for you.
Quoted against your asset list, not your headcount.
We do not charge per seat, because a monitoring platform only works when the whole plant is looking at the same screen. Pricing is quoted against your asset list.
One production line, fixed scope, fixed price, and a written success definition agreed before we start.
A full site under monitoring, with the receiver tier sized to your asset density and layout.
Several plants under one organisation, or a deployment that has to live inside your own perimeter.
Bring your worst-behaving production line. We will scope it, install it, agree the success criteria in writing before we start, and review the results against them honestly — including what we got wrong.