Topologies

Typical gateway connection patterns

This section helps choose the site pattern before configuration. It covers RS-485 lines, TCP devices, Modbus TCP server, NAT, MQTT, multiple gateways, dispatching, and UMEC Space for distributed sites.

How to choose a topology

Need Recommended pattern Pre-commissioning check
Read several local RS-485 devices. One RTU 1 line. Devices have unique slave IDs, shared baud/parity/stop bits, and correct A/B/GND wiring.
Separate different speeds, floors, cabinets, or equipment groups. RTU 1 + RTU 2. Each line has its own serial profile and device list.
Add a network Modbus TCP device. TCP channel in polling mode. The device IP is reachable from the gateway network, the port is open, and unit ID is explicit.
Send telemetry to an external system. UMEC Space or Generic MQTT. Local polling is stable and MQTT secrets are not stored in public places.
Expose RTU devices to SCADA over Ethernet. Modbus TCP server listener + route map or NAT. Routes do not conflict and repeated slave IDs are handled by explicit NAT rules.
Dispatch several sites or zones. Multiple gateways + one SCADA or UMEC Space account. Every gateway has a unique site name, clear IP/hostname, and separate credentials.

Pattern 1. One local RS-485 segment

The simplest pattern: the gateway polls RTU 1 devices, the operator uses Control and dashboards, and logs are stored locally. External integration may be disabled.

flowchart LR
    O[Operator laptop] -->|HTTP| G[UMEC Gateway]
    G -->|RTU 1 / RS-485| D1[Meter]
    G -->|RTU 1 / RS-485| D2[Sensor]
    G -->|RTU 1 / RS-485| D3[Controller]
    G --> S[(Storage / Logs)]

Pattern 2. Two independent RS-485 lines

Use two lines when devices are physically separated, use different serial parameters, or one group must remain available if the other segment has a problem.

flowchart LR
    G[UMEC Gateway]
    G -->|RTU 1: 9600 8N1| A1[Ventilation]
    G -->|RTU 1: 9600 8N1| A2[CO2 sensors]
    G -->|RTU 2: 19200 8E1| B1[Meters]
    G -->|RTU 2: 19200 8E1| B2[Relay modules]
    UI[Web UI] --> G

Pattern 3. Local RTU devices plus TCP device

The TCP channel is used for network Modbus devices and integration endpoints. Do not treat it as transparent RS-485: it has its own polling, not-polling, and disabled modes.

flowchart LR
    G[UMEC Gateway]
    G -->|RTU 1| R1[RS-485 devices]
    G -->|TCP bus3| T1[Modbus TCP device]
    G -->|TCP tunnel| T2[Remote Modbus endpoint]
    G --> C[Control / Dashboards / Automation]

Pattern 4. Local site plus MQTT

MQTT is enabled after local operation is verified. The gateway keeps polling locally and sends prepared telemetry outward. Losing the broker should not stop local Control and scenarios.

flowchart LR
    D[RTU/TCP devices] --> G[UMEC Gateway]
    G --> UI[Local web UI]
    G -->|MQTT / Wi-Fi or Ethernet| B[(Broker)]
    B --> P[SCADA / Cloud / Analytics]

Pattern 5. SCADA reaches RTU through Modbus TCP server

Here the gateway accepts Modbus TCP requests and routes them to RTU lines. This is useful when the upper system speaks Modbus TCP while field devices remain on RS-485.

flowchart LR
    S[SCADA / PLC] -->|Modbus TCP| L[Gateway listener]
    L --> G[UMEC Gateway route map]
    G -->|RTU 1| A[Slave ID 1..20]
    G -->|RTU 2| B[Slave ID 21..40]
    G --> N[NAT / strict map]

Pattern 6. NAT for repeated addresses

NAT is useful when different segments contain the same slave IDs, but the external system must see them as distinct addresses. Do not use NAT as a substitute for clean addressing: every rule must be explicit and verifiable.

flowchart LR
    S[Modbus TCP client] --> G[Gateway NAT]
    G -->|source unit 101 -> RTU1 unit 1| A[Device A / unit 1]
    G -->|source unit 102 -> RTU2 unit 1| B[Device B / unit 1]
    G -->|priority order| R[NAT rules]

Pattern 7. Multiple gateways for dispatching

Large sites often work better with several gateways placed close to equipment: one in the metering cabinet, one near ventilation, and one in a remote pump room. SCADA or a dispatching server receives data from every gateway, while local diagnostics remain available in each zone.

flowchart LR
    subgraph Site[Site]
      G1[Gateway A / Metering cabinet] --> M1[Meters]
      G2[Gateway B / Ventilation] --> V1[AHU / CO2 / dampers]
      G3[Gateway C / Pump room] --> P1[Pumps / sensors]
    end
    G1 -->|MQTT or Modbus TCP| D[Dispatching]
    G2 -->|MQTT or Modbus TCP| D
    G3 -->|MQTT or Modbus TCP| D
    D --> O[Operator / engineer]
Rule Practical meaning
Name gateways by installation place. Examples: shop-01-metering, shop-01-ahu, warehouse-pump-room.
Do not mix physical zones without a reason. When a line fails, the service engineer knows where to look: cabinet, AHU room, or pump room.
Separate local access from external data collection. The site operator uses the specific gateway UI, while the dispatcher sees the combined picture in SCADA or UMEC Space.

Pattern 8. Distributed sites in UMEC Space

UMEC Space fits deployments with many sites: stores, warehouses, pump stations, ventilation rooms, and metering cabinets. Each gateway collects data locally, while the platform shows the site list, communication status, alarms, trends, and aggregated dashboards. Customers operate sites in UMEC Space Dashboard, with mobile access through UMEC Home or UMEC Business.

flowchart LR
    subgraph Stores[Distributed sites]
      A[Store 1 / Gateway]
      B[Store 2 / Gateway]
      C[Warehouse / Gateway]
      D[Pump station / Gateway]
    end
    A -->|MQTT over TLS / site credentials| U[(UMEC Space)]
    B -->|MQTT over TLS / site credentials| U
    C -->|MQTT over TLS / site credentials| U
    D -->|MQTT over TLS / site credentials| U
    U --> M[Dispatcher]
    U --> E[Service engineer]
What to configure Expected operating result
Unique site and gateway name. The platform clearly shows where data came from and which site has an alarm.
Stable local polling before cloud enablement. If the internet fails, the site keeps working locally and publishing resumes after recovery.
Separate credentials and secrets per site. Compromise of one site does not expose the whole fleet.

Worked examples

Store

The gateway collects refrigerated display temperatures, ventilation state, energy meters, and alarms. The dashboard shows temperature, consumption, and active warnings.

Ventilation unit

RTU 1 connects the AHU controller, CO2 sensors, and temperature sensors. A scenario can raise a warning on high CO2 or fan failure.

Metering cabinet

RTU 1 or RTU 2 polls meters. Stable line parameters, unique addresses, and regular log preservation matter most.

SCADA bridge

The gateway accepts Modbus TCP requests from SCADA and routes them to RS-485 devices. NAT is used only where slave IDs repeat.

Screen set for topology verification

Screen What should be visible Ready when
Setup Line, serial profile or TCP endpoint, device list, channels, and commands. Devices are saved, addresses do not conflict, and parameters match equipment documentation.
Control Fresh values, line state, available commands, and selected chart channels. There is no mass offline/stale state and commands were checked on safe loads.
Dashboard Key site indicators: temperature, energy, equipment state, and alarms. The panel is understandable to the operator without opening Setup.
Logs Scenario Log, Modbus Log, and Storage files with evidence. You can prove when a device responded, when an error happened, and what action the operator took.

Common RS-485 mistakes

Mistake Symptom Action
A/B swapped All or almost all devices on the line do not respond. Swap A/B at one line end and repeat the Inspector or Control check.
No common GND Responses are unstable, especially under load or on a long cable. Check common reference/ground according to the equipment requirements and exclude potential difference.
Wrong termination A short line works, but a long line has random timeouts. Place terminators at the bus ends, not on every branch.
Wrong baud/parity/stop bits The device is wired but never responds to any register. Compare the equipment manual with the Setup line settings.
Duplicate slave ID Responses look random or values belong to the wrong device. Assign unique addresses within the line or move devices to separate lines/NAT mapping.

From box to first dashboard

Connect power and Ethernet, open http://192.168.4.2, sign in as administrator, and change the password.
In Settings set the site name, network settings, time, and integration mode.
In Setup configure RTU 1: baud, parity, stop bits, timeouts, and the device list.
Add read channels: name, register/function, data type, units, and polling interval.
Open Control and confirm values are fresh without persistent offline/stale state.
Create a dashboard: add widgets for main values, use clear labels, and make the dashboard default.
Check Logs: Modbus Log should prove exchange, Scenario Log should be empty or contain understandable events.

Security checklist

Action Why it matters Minimum rule
Change the password after first login. Bootstrap credentials are known to installation and service teams. Use a unique site password and store it in the customer-approved password store.
Isolate the service network. Web UI and Modbus TCP should not be reachable from guest or public networks. Place the gateway in a technical VLAN or behind a router with explicit access rules.
Handle MQTT secrets carefully. Secrets expose telemetry and sometimes control topics. Do not publish screenshots with secrets; use separate credentials for separate sites.
Restrict Modbus TCP access. A wrong SCADA command can change equipment state. Allow only trusted IPs and review the writable-command list.