IMDHMIControls LibraryShop main store

Troubleshooting

Replacing a KoyoNet Citect link with Do-more Modbus TCP

SCADA and data logging Published 7 min read

Short answer

Use the H2-DM1E as a Modbus TCP server and point Citect's ModNet driver at it, mapping measurements into MIR, setpoints into MHR, and status and command bits into MI and MC. The working KoyoNet link stays only as a temporary read bridge, because its octal addressing, integer-only V-memory and reported 768-word window are exactly what you are leaving.

On this page
  1. What the legacy path actually gives you
  2. Sizing the KoyoNet V1400 to V2777 window on this retrofit
  3. Where the Do-more Modbus server can see your data
  4. The migration in order
  5. When a migrated tag misbehaves
  6. Skip the OPC server on this job
  7. Questions asked

Use the H2-DM1E as a Modbus TCP server and point Citect's ModNet driver at it. The CPU's single Ethernet (RJ45) port already carries the Terminator remote I/O, the drives stay on Modbus RTU, and Do-more CPUs with on-board Ethernet run the Modbus TCP server on that same port, so the SCADA link needs no gateway and no OPC server — only a mapping rung and one new I/O device. The Modbus server on a Do-more answers from four memory areas, MI, MC, MIR and MHR, and from nothing outside them, so the real work is choosing which signal lands in which area and writing the copy logic. Keep the working KoyoNet link as a read-only bridge while you test, then delete it.

What the legacy path actually gives you #

KoyoNet does talk to the DL server inside the Do-more, and that is worth having during the changeover: it proves the SCADA PC, the licence and the tag database all still work. What it costs, measured on that same working link, is a list of five limits:

  • Octal addressing returns for every SCADA tag, because the DL-compatible memory follows the DirectLOGIC octal convention: after V1417 comes V1420, since 8 and 9 are not octal digits.
  • No bit-of-word casting for Citect tags, so status bits packed into a word cannot become plain Boolean tags.
  • The DL server interface defines V-memory as integers only.
  • Negative values travel as two's-complement patterns, which a tag declared unsigned displays as large positive numbers.
  • The reachable window for the Citect driver was V1400 to V2777 — a reported limit of this working link, not a published specification, and the next section sizes it.

None of these are transport faults. A different driver will not fix them, because each one lives in how the data is addressed and defined.

Sizing the KoyoNet V1400 to V2777 window on this retrofit #

Inputs, each with its source:

  • Window: V1400 through V2777, the V-memory the Citect driver reached on the working link, as reported on that installation.
  • Address base: octal, the DirectLOGIC convention the DL-compatible view inherits.
  • Data width: one V-memory location is one 16-bit word.

The rule in words: convert each end of the window from octal to decimal, subtract, and add one so both ends are counted.

  • V1400 in octal = (1 × 512) + (4 × 64) + (0 × 8) + 0 = 768 decimal
  • V2777 in octal = (2 × 512) + (7 × 64) + (7 × 8) + 7 = 1535 decimal
  • 1535 − 768 + 1 = 768 words

The result: 768 words of 16 bits for the entire Citect database, with any 32-bit value costing two of them.

The decision: above 768 words of live data — drive setpoints and feedback, Terminator I/O status, alarms — the KoyoNet path cannot carry the job and the migration is mandatory. Below that count the link can carry commissioning, but the integer-only definitions, the unsigned traps and the packed bits all remain, so finish the Modbus move for the permanent installation either way.

Where the Do-more Modbus server can see your data #

On Do-more CPUs the Modbus servers are sandboxed to four memory areas — MI, MC, MHR and MIR — and cannot see anything outside them. Internal Do-more memory is strongly typed; these four areas are loosely typed words and bits, like the DL205 memory the old project used, so your mapping code casts values on the way in and the Citect tags declare their type on the way out.

Do-more area Modbus table Direction from SCADA Use it for
MI Discrete inputs Read-only Status bits
MC Coils Read-write Command bits and mapped Booleans
MIR Input registers Read-only Measurements and feedback
MHR Holding registers Read-write Setpoints and commands

Coils sit on binary boundaries, which is the tidy representation you were after: a status bit becomes a plain Boolean tag instead of bit 6 of an integer word. Keep measurements in MIR and commands in MHR so the direction of every value is visible in its address, and name one owner for each register in the data contract.

The migration in order #

  1. Back up the Citect project and the Do-more program before touching any communications configuration.
  2. Write the data contract before any code: one table of tag, area, address, data type, direction and scaling. That table is the cross-reference future technicians were missing.
  3. Add the mapping rung: copy and cast internal values into MIR and MHR words and MI and MC bits, with a block-copy instruction such as MEMCOPY in a periodically executed rung. Change nothing else in the control program.
  4. Clamp and validate everything SCADA writes — range-check each MHR setpoint before it can reach a drive write, and pulse MC commands — because a proven read path proves nothing about a safe write path.
  5. Enable the Modbus TCP server on the Ethernet port, entering the IP address, subnet mask and gateway the Do-more H2 hardware manual lists for the port. Leave the RTU drive network and the Terminator I/O untouched so the new link is the only variable you are testing.
  6. Add the MODNET I/O device in Citect against the PLC's address; the driver-and-port layout follows the same pattern as connecting Citect SCADA to a Modbus gateway over MODNET.
  7. Set the device's slave (unit) ID to the server's ID, following setting the Modbus slave ID for a Citect I/O device. If MODNET answers a TCP setup with a missing-serial-drivers message, why MODNET reports it cannot find serial drivers covers that fault.
  8. Bring up exactly one read-only test tag at one known MIR address and compare the raw word in Citect with the same word in the Do-more memory view. Settle the address-base question with that register before generating the rest: reference-based and offset-based conventions differ by one, and an off-by-one base returns a believable neighbour. You can rehearse this whole test offline by simulating a Modbus TCP SCADA link on one PC.
  9. Cut the read tags over one by one with KoyoNet configured but unread, then move command writes last, after the clamps have proven themselves. If a second Citect project polls the same device during the change, resolving #BAD errors when multiple Citect projects poll one MODNET device is the page for that fault.
  10. Delete the DL-server path once the installation has soaked; two live paths to the same data are a maintenance trap, not a redundancy.

When a migrated tag misbehaves #

Symptom Likely cause Fix
Question marks or no quality Transport: timeout, wrong address, server not enabled Check reachability and the server configuration first
Good quality, wrong value Address base off by one, reading a neighbour Re-test the one known MIR register
Negatives show as large positives Tag declared unsigned Declare the raw type signed 16-bit
Bits unreadable or awkward Still packed in a word Move them to MI or MC
Writes never land Clamp rejected the value, or the mapping rung is not scanning Check the rung execution and the limits

Question marks on the tags are almost always why HMI values show question marks when PLC comms time out, so clear the transport before touching any tag. A tag that goes BAD on its configuration points at the type instead: BAD raw data type fix for a MODNET tag walks that correction.

Skip the OPC server on this job #

You do not need one. ModNet is already a Citect driver, the Do-more already serves Modbus TCP on the port in use, and every read and write this retrofit needs is a word or a bit. An OPC server would add a licence, a process to keep alive and a third place to diagnose, and it would not remove the mapping work, because the server still sees only the four Modbus areas. The decision: ModNet to the native Modbus TCP server, KoyoNet dark through commissioning and deleted after it, and OPC reserved for a future consumer that cannot poll Modbus at all.

Questions asked

Do you need an OPC server between Citect and the Do-more?

No. The Do-more CPU serves Modbus TCP on the Ethernet port it already uses, and ModNet is the Citect driver for that protocol, so the pair covers every read and write of words and bits this job needs. An OPC server would add a licence, a process to keep alive and a third place to diagnose, without adding a capability the mapping cannot give. Revisit OPC only if a later requirement is OPC UA to systems that cannot poll Modbus.

Why does a negative value arrive as a large positive number?

The register holds a 16-bit pattern, and the sign lives in the interpretation. A stored value of -2500 is the two's-complement pattern 65536 - 2500 = 63036, so a tag declared unsigned displays 63036. Declare the Citect raw data type as signed 16-bit to match what the mapping rung wrote, and the same pattern reads -2500. The migration does not remove two's complement; it removes the guesswork by letting you set the type per tag.

Can ModNet read the Do-more's internal data directly?

No. On Do-more CPUs the Modbus servers answer only from MI, MC, MHR and MIR; internal strongly typed memory is invisible to a Modbus client. Every value Citect needs must be copied and cast into those four areas by the mapping rung. That is also the discipline that pays off later: the mapping table becomes the documented interface a future technician can read without a cross-reference.

How do you fix the ModNet address base before building hundreds of tags?

Resolve it with one register, not a manual. Put a distinctive known value in a single MIR word, read it from Citect, and see which address returns it: reference-based and offset-based conventions differ by one, and only a live read settles which one your driver uses. When that register reads correctly, freeze the convention in the mapping table and generate the remaining tags from it. An off-by-one base returns a believable neighbour, so never trust good quality alone.

Should the KoyoNet driver stay as a fallback?

Through commissioning, yes, but dark: leave the I/O device configured and unread while ModNet carries the plant, so you can fall back within minutes if a fault appears. Keep its write paths disabled the whole time, because a proven read says nothing about a safe write. After the soak period, delete the DL-server path entirely; a second live path to the same data is a maintenance trap, not a backup.

Keep reading

  • How-to

    Connecting Citect SCADA to a Modbus RTU gateway over MODNET

    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.

    SCADA and data logging

  • How-to

    How to simulate a Modbus TCP SCADA link on one PC

    Yes — one PC can run the SCADA client and a Modbus TCP server simulator at once. Decide which side you are proving, keep exactly one client and one server in the arrangement, prove the server exists with an independent Modbus client, then align object type, offset and word order before trusting any value; a connected link proves transport only.

    SCADA and data logging

  • Troubleshooting

    Reading a CLICK PLC float over Modbus TCP in an HMI

    F428673 is the right address: the F prefix reads the two Modbus registers as the IEEE 754 float the CLICK program wrote to DF1 (24.9645 here), while L returns the raw 32-bit pattern (1103607636) and U returns one 16-bit half (46932). The screen showed 25 because an integer tag or whole-number format rounded it. Move the project to 3.99w and bind F with a decimal format and a floating-point variable.

    SCADA and data logging