How products are combined by industry. Tell us your equipment conditions and we will review the configuration and reply.
Machines without communication functions, such as presses and hydraulic equipment, require an operator to walk the floor and check run status and alarms. Replacing the machine or modifying the PLC is a heavy cost, and in many cases the manufacturer's warranty is at stake.
Signal flowExisting control panel contacts→EdgeIO controller→Modbus TCP · MQTT→MES · OEE
Example acquisition and control signals
| DI | Run, stop, alarm and e-stop contacts, 8 to 16 points |
|---|
| AI | Hydraulic pressure and motor temperature, 4 to 8 points (4~20mA) |
|---|
| Pulse | Production count (branched from the existing counter output) |
|---|
Points to review
- The existing control circuit is left untouched; only a branch off the lamp and relay contacts is taken.
- Check first whether the contact is dry or 24V-driven, so the input type (dry/wet) can be decided.
- Isolated inputs are used so that taking the branch does not affect operation of the existing circuit.
Injection molding machines use different communication specifications depending on the maker and the model year. Fitting a separate gateway and PLC to every machine raises the per-unit cost and adds that many more points to manage.
Signal flowInjection molding machine→BL192Pro→MQTT→Cloud · MES
Example acquisition and control signals
| Pulse | Shot count (clamp signal or the existing counter) |
|---|
| AI | Clamping force, barrel temperature, cycle time |
|---|
| DI | Run, alarm, mold change in progress |
|---|
| DO | Stack light, call buzzer |
|---|
Points to review
- Which signal the shot count is taken from determines data reliability. The clamp-complete signal is the most stable.
- Local storage is configured so the counter is retained after a power failure or restart.
- Calculating cycle time in local logic before sending it reduces the load on the upstream system.
Adding a new machine to a SCADA system already in operation normally requires developing a dedicated driver. The development repeats every time another type of machine is added, and it ties the site to the SCADA vendor.
Signal flowTextile machinery · field sensors→BL191 (OPC UA server)→OPC UA→Existing SCADA
Example acquisition and control signals
| AI | Speed, tension, temperature |
|---|
| DI | Run and stop-reason contacts |
|---|
| Modbus | Reading existing inverters and instruments |
|---|
Points to review
- Designing the OPC UA node addresses per machine in advance means the SCADA screens do not have to be rebuilt as machines are added.
- The SCADA polling interval and the controller update interval must match, or values become unstable.
- The OPC UA security policy (certificates, signing and encryption) has to be agreed with the upstream system first.
Even a single line mixes equipment makers and controllers, which makes it hard to collect output and downtime data on the same basis. When the acquisition method differs per machine, the basis for calculating OEE is not consistent either.
Signal flowLine equipment group→Gateway (per line)→Modbus · OPC UA · MQTT→MES · OEE
Example acquisition and control signals
| Count | Good and defective quantities |
|---|
| DI | Running, stopped, stop-reason code |
|---|
| Time | Time synchronization between machines (for calculating downtime periods) |
|---|
Points to review
- Decide first whether stop reasons come from the machine as a code or from operator input.
- OEE is time-based, so if the clocks on the machines are not synchronized the totals do not line up.
- Separating the OT and IT networks and placing the gateway at the boundary makes the security segmentation clear.
- A stopped line is expensive, so switch and link redundancy is reviewed at the same time.
An existing building automation system is built on BACnet, while the air handling units and instruments being added are often Modbus. Because the specifications differ, they cannot be shown on the BAS screens and end up managed separately.
Signal flowAir handling units · instruments (Modbus)→BA190Pro→BACnet/IP→Existing BAS
Example acquisition and control signals
| RTD | Supply and return air temperature (PT100 / PT1000) |
|---|
| AI | Humidity, differential pressure, flow |
|---|
| AO | Valve and damper position control (0~10V / 4~20mA) |
|---|
| DI / DO | Fan and pump status, start and stop |
|---|
Points to review
- BACnet object numbers (instances) must be divided into ranges in advance so they do not collide with existing devices.
- On segments using BACnet MS/TP, the baud rate and MAC address have to match the settings of the existing network.
- Whether control is also handed over or the system only monitors changes how write permissions are designed.
Power grid operations systems accept only IEC 60870-5-104 or IEC 61850. Field energy meters often use DL/T645 or Modbus, so they cannot be connected as they are.
Signal flowEnergy meters · protection relays→Protocol gateway→IEC104 · IEC61850→Grid operations system
Example acquisition and control signals
| Measurement | Voltage, current, active and reactive power, power factor, accumulated energy |
|---|
| Status | Circuit breaker and disconnector open/closed status |
|---|
| Event | Protection operation, alarms (with time stamps) |
|---|
Points to review
- The ASDU addressing scheme has to be agreed with the grid operations side first. Changing it later means changes on both sides.
- Time stamps on events matter, so a time synchronization method (NTP or similar) is part of the configuration.
- Substation equipment is hard to take out of service, so adding only a data layer while keeping the existing equipment is the suitable approach.
Pump stations and small treatment plants are hard to reach with a wired network and have no staff on site. A fault can go unnoticed until the next inspection round.
Signal flowPump station · treatment plant→4G RTU→MQTT · SMS→Control server · staff mobile phone
Example acquisition and control signals
| AI | Level, flow, pressure (4~20mA) |
|---|
| RTD | Water temperature |
|---|
| DI | Pump run, overload, flooding, door open |
|---|
| Relay | Remote pump start and stop |
|---|
Points to review
- Local logic is put in place so pump control keeps working on site even if communication is lost.
- Alarms must still go out during a power failure, so a backup battery and power-failure detection are configured together.
- Alarm recipients and the escalation order have to be defined in advance so that night-time events are covered.
- Transmission interval and event-transmission criteria are set separately to keep data charges in check.
Sending vision inspection images to the cloud for a decision adds the round-trip delay and accumulates link costs. If the decision cannot keep up with line speed, the inspection itself becomes the bottleneck.
Signal flowCameras · sensors→EdgePLC with NPU→Result only is transmitted→MES · cloud
Example acquisition and control signals
| Image | USB / Ethernet camera |
|---|
| DI | Trigger (product arrival detection) |
|---|
| DO | Good/defective sorting output, reject actuator |
|---|
| AI | Vibration and current for anomaly detection (when an IEPE module is used) |
|---|
Points to review
- The NPU performs with INT8 quantized models. Model size reduction has to be considered from the training stage.
- Decision speed depends on model size and input resolution, so it is selected to match the line tact time.
- Lighting conditions have the largest effect on accuracy. Fix the lighting conditions before selecting hardware.
- A remote deployment path (BLRAT or similar) is included in the configuration so models can be replaced on site.