40001 · Signal path
How it works
The same path for every instrument: a camera on the process, the number
computed on the device, written to a register your PLC already reads.
-
1 cam · FIELD OF VIEW
Camera
A fixed camera watches the surface your operators watch by eye — the froth in a cell, the foam on a tank.
-
2 edge · INFERENCE
On-device inference
The surface is measured on the device, at the plant — computer vision and machine-learning models running at the edge. Nothing is sent off site.
-
3 40001 · WRITE
Modbus registers
The measurements are written to Modbus TCP holding registers — over a dozen physical characteristics: size, coverage, velocity, texture, depth, state.
cov % → 40001
-
4 plc · READ
Your PLC / DCS
The control system reads another register and decides what to do. It behaves like a level transmitter.
No server. No cloud. No GPU rack.
- What does an X Vision instrument measure?
- The surface an operator watches by eye — froth in a flotation cell, foam on a process tank — turned into continuous numbers. X Float alone measures over a dozen physical froth characteristics: bubble-size distribution, velocity, stability, texture, depth, colour and process state. Colour is only one of them.
- Does it need cloud connectivity?
- No. Inference runs on the device, on-premises. The instruments are offline-capable and nothing leaves site.
- How does the measurement reach the control system?
- As Modbus TCP holding registers. The PLC reads another register and decides what to do with it — the instrument behaves like a level transmitter.
- Does installation require a shutdown?
- No. The instrument is a self-contained camera that needs only power and a network drop. It mounts over the launder or tank in minutes and commissions in hours, with no process shutdown — and it moves to another cell or tank just as easily. There is no fixed sampling rig to build and no routine maintenance in normal operation.
- Does X Float use AI?
- X Float's core is classical computer vision — over a dozen physical measurements taken straight from the image and stereo depth. Machine learning is added for higher-order tasks such as site-calibrated grade correlation, not as a black box. X Foam, by contrast, uses a neural segmentation model to separate foam from water by colour.