How-to
Connecting Citect SCADA to a Modbus RTU gateway over MODNET
Short answer
Yes — MODNET is Citect's driver for Modbus over TCP/IP, and a true RTU-to-TCP converter presents standard Modbus TCP on its Ethernet side, so the combination works. Success then rides on two things: the gateway's operating mode, and carrying the RTU slave address and register offsets into Citect correctly, proved by reading one register whose value you can see at the device.
On this page
Yes — and the decision hinges on one distinction. MODNET is Citect's driver for Modbus over TCP/IP, and a Moxa RTU-to-TCP gateway that presents standard Modbus TCP on its Ethernet side is exactly what that driver is documented for. Configure the gateway as the converter, point one MODNET I/O device at it, carry the RTU slave address as the unit, and prove the link by reading a register whose value you can see at the device. Two cases take a different path: a Moxa ioLogik remote-I/O unit is a Modbus TCP device in its own right with Moxa's own Citect instructions, and a link that tunnels raw RTU frames inside TCP is not Modbus TCP — the documented guidance for that mode says do not use MODNET.
What MODNET actually is #
The name reads like a legacy PLC protocol, but the driver's documented scope is Modbus over TCP/IP: the standard Modbus function codes carried in TCP sessions on the standard port, 502. Asking whether MODNET will talk to a Moxa is therefore one question — does the gateway's Ethernet side present standard Modbus TCP? A true converter does. It terminates the TCP session itself, rewrites each request as an RTU frame on the serial wire, and returns the response in TCP.
The distinction that fails installations is conversion versus encapsulation. Some setups tunnel raw RTU frames through a TCP stream; that is Modbus RTU over TCP, not Modbus TCP, and for that mode the advice is to leave MODNET out of it. A converter that answers real Modbus TCP requests is the case this driver exists for. Its parameters, including the Ethernet TCP/IP special options that live in citect.ini, are listed in the driver reference for your installed version.
If the Moxa product is an ioLogik remote-I/O unit rather than a gateway, stop here: it speaks Modbus TCP directly, Moxa publishes Citect installation instructions for it, and those instructions own the addressing. Everything from the known-value test onward still applies.
What to confirm before you configure #
Four facts about the gateway decide the whole setup, so collect them from the product documentation before opening the Citect editor.
| Confirm on the gateway | What it decides |
|---|---|
| Exact product and model | Converter versus ioLogik remote I/O, and whose installation instructions own the setup |
| Operating mode | Whether Citect can open the TCP session as master and have RTU requests issued on its behalf |
| Serial settings | Baud rate, parity and stop bits must equal the RTU device's settings, identically on both ends |
| Downstream register map | Which register numbers hold which values, their data types and their scaling |
Configure the link, step by step #
- Record the exact Moxa model before touching software. If it is an ioLogik unit, follow Moxa's Citect installation instructions; if it is an RTU-to-TCP converter, continue here.
- Give the gateway a static IP on the same subnet as the Citect station, and set its Ethernet side to the mode where it listens for a Modbus TCP master. Citect is that master; the gateway should not need to initiate anything toward Citect.
- Set the gateway's serial side to the RTU device's exact settings. If the meter runs 9600 baud, even parity and one stop bit, the gateway serial port gets the same three values — one mismatched parameter produces timeouts, not partial data.
- In the Citect project, create the I/O device on the MODNET driver, addressed to the gateway's IP and TCP port 502.
- Give the I/O device the RTU slave address as its number, so every request carries the right unit identifier — the same decision made when setting the Modbus slave ID for a Citect I/O device.
- If the I/O device reports
Cannot find serial driverswhen it starts, the driver pack is missing from the installation — see why MODNET reports it cannot find serial drivers and install the pack the driver reference names. - Create one variable against a register you can verify at the device, such as a totalizer or a value on the local display, before building the full tag list. The next section works the address and the scaling.
- Compare the Citect value with the device display. Only when that one register matches to the digit should you roll out the remaining variables.
- Leave writes disabled until the read side is proven, and before the first write to anything that can start or move the machine, confirm a safe state with the people who own it — then apply the habits in safe Modbus writes to PLC output registers from SCADA.
Get the register address and scaling right #
Most RTU device manuals number holding registers in the 4xxxx series, where 40001 is protocol address 0. The offset arithmetic is one subtraction: protocol address = register number − 40001. Reading holding registers is function 03.
Worked example with illustrative values: a meter documents its total at holding register 40011, scaled 0–10000 counts for 0–100.0 %.
- Address: 40011 − 40001 = 10.
- Scale: 100.0 % ÷ 10000 counts = 0.01 % per count.
- The gateway returns 4732 counts, so the engineering value is 4732 × 0.01 %/count = 47.32 %, which must match the meter's display to the last digit.
- A block read of 40010 through 40012 starts at 40010 − 40001 = 9 for a quantity of 3 registers.
The raw address goes on the variable; the 0.01 % per count scale goes in the variable's scaling fields. A 32-bit value occupies two consecutive registers, so when a number reads absurdly large or zero the two words are swapped, and the driver exposes a word-order setting for exactly that. Keep each read inside one contiguous block of the map — a request that steps across unmapped space is the cause behind Modbus exception 02 when a read crosses a register map gap.
Prove the link before you trust it #
A session that opens proves transport only; matching data proves the configuration. Before the hardware is even in the panel you can rehearse the Citect side of this by following how to simulate a Modbus TCP SCADA link on one PC.
When the single-register test disagrees, isolate in this order: the gateway's operating mode, the unit identifier on the I/O device, then the register offset and word order. Each level has its own symptom. Mode faults look like timeouts or refused sessions; a wrong unit comes back as a timeout or a gateway exception because the RTU side never answers; offset faults return plausible-but-wrong values. One register, checked at the device, separates all three in minutes — and it is the test that licenses every tag you build afterward.
Questions asked
Is MODNET just an old protocol for specific PLCs?
No. In Citect the name refers to the driver for Modbus over TCP/IP, not to a private PLC protocol, so it is the right driver whenever the far side is standard Modbus TCP. What it is not is a tunnel for Modbus RTU frames inside TCP; the documented guidance for that encapsulated mode says to use something else. Judge the driver by its documented scope, not by how old the name sounds.
Does this work with a Moxa ioLogik unit instead of a converter?
Yes, but skip the conversion logic. An ioLogik remote-I/O unit is itself a Modbus TCP device, so MODNET talks to it directly and the unit ID is the ioLogik's own address, not an RTU slave behind a gateway. Moxa publishes Citect installation instructions for these products; follow those for the exact variable addressing, then apply the same known-value test before trusting the data.
Citect connects to the gateway but the values are wrong — where is the fault?
A working connection proves only transport, not data mapping, so work through three checks in order: the gateway's operating mode, the unit ID on the I/O device, and the register offset in the variable. Word order is the fourth suspect for 32-bit values split over two registers. Reading one register whose value you can see at the device separates these faster than reconfiguring the whole link.
Where does the RTU slave address go in Citect?
It travels as the unit identifier of every TCP request the driver sends, so it is set at the I/O device, not per variable. Keeping one I/O device per distinct RTU slave address makes the mapping legible when several meters share one serial line behind one gateway. If reads fail for only one device on a multi-drop line, that device's address is the first thing to re-check.