Home › Applications › Remote IO over 4G
Remote IO over 4GA contact closes at one site. A lamp lights at another.
Mirror digital inputs at one site onto relay outputs at another over 4G, using any open MQTT broker. The mapping is in the firmware at both ends — no bridging software to install, no server to maintain, no vendor cloud in the data path.
This is indication and signalling only — it is not an interlock, not an emergency stop and not a safety function, and nothing that protects people or plant may depend on it.
How it works
including what happens when it breaksWhat sits at each end
Seven mapped signals, not eight. One relay is permanently DATA VALID, because a lamp that is off because the contact is open looks identical to a lamp that is off because the link died. Wire the panel so DATA VALID holds in a COMMS FAIL lamp, and feed that lamp from a supply that survives the card. A dark panel is not a healthy panel.
When the link drops
three failures, three detection timesThe WT32i’s watchdog is configured over Modbus. If the card stops being written to, it releases its outputs by itself — without consulting the network, the broker or us. All three times below are arithmetic from your settings, and all three are measured and recorded at commissioning.
Nothing of ours runs on the broker. The master publishes status, the slave treats that topic as its command topic and executes it. That removes an entire class of failure — there is no bridge process to crash, hang or be left un-patched — and it means the link works with whatever broker the site already has.
Store & forward is off on this channel. Buffering is right for telemetry and wrong here — replaying a queued output command after a ten-minute outage would light a lamp with a state from ten minutes ago. Latest value wins; anything stale is discarded on its sequence number.
Where it earns its keep
Who can drive the outputs
Because any broker works, the broker you choose is the whole of your security boundary. Use TLS on both the 4G and LAN paths; give the master and the slave separate credentials; set topic permissions so the master can only publish its own status and the slave can only subscribe to it; disable anonymous access and the plaintext listener.
Anyone who can publish on the master’s topic can drive the slave’s outputs. That is the one sentence to take away — credentials are per-device, rotated at handover, and owned by your IT rather than by us. A convenient public test broker is fine on the bench and wrong in a control room.
What this is not
This is indication and signalling only — it is not an interlock, not an emergency stop and not a safety function, and nothing that protects people or plant may depend on it.
Your pull-cords, zero-speed switches, trip relays, dry-run protection, ESD and fire systems stay hardwired and local, exactly as they are today. This runs alongside them and tells a person at the other site what is happening. It is a process status annunciator — not a fire alarm annunciator, not listed for life-safety use, and not connected to any fire system.
Public cellular round trip is commonly around 200 ms and can be seconds when a cell is congested or a modem is re-registering. It is never guaranteed, and we do not offer a latency figure we cannot hold. What you get at handover is the measured distribution for your installation, recorded across a working day.