Troubleshooting
How to Connect an HMI to an SLC 500 Over DH-485 Serial
Short answer
Most HMI-to-SLC 500 DH-485 failures live in the cable, the serial settings or the panel driver, not in the tags. Run every station at the same baud rate (19200 is the DH-485 default), give each a unique address in the 0–31 range, use the approved cable or the documented interface-module wiring, and prove the link by reading one known PLC address before trusting a single screen value.
On this page
A third-party HMI that will not talk to an SLC 500 over DH-485 is almost never a tag problem — it is a link problem. Work a fixed order: prove the cable, prove the serial settings, prove the driver assignment, and only then read one known PLC address. On a healthy link, every station runs the same baud rate, every station holds a unique address, and the driver in the panel project points at the port the cable actually uses. Settle the cable with documents, not with a copied drawing: the panel maker's approved assembly on one end, the pin assignments in Rockwell publication 1747-IN006 on the other. If your terminal is a Rockwell panel rather than a third-party one, work the same network through the PanelView 1000 to SLC-5/02 DH-485 checklist.
Split a dead link from a bad tag #
DH-485 is token passing: the panel only speaks when it holds the token, so a fault anywhere on the segment — a damaged cable, a duplicated node address, a driver on the wrong port — looks identical at the screen to a dead processor. Sort the symptom before opening the project:
- Every value frozen and a communication error on the panel: the link. Nothing updates until the token reaches the processor, so start with cable, settings and driver, never with the tag list.
- Other screens update but one number is wrong or stale: that screen object's data-file address or data type. The network has already proven itself.
- The link died when a second device was added: a duplicated node address. Renumber before touching anything else.
- Alive but everything crawls: poll load, not wiring. The screen is asking for more data than the wire can carry in its update window.
Settle the cable with documents, not with a copied drawing #
For the panel side, order the approved assembly the panel maker sells for this exact link; HMI-CAB-C84 is that part number. The maker publishes the cable's internal pinout as a convenience, and its documents state that cables built without authorization are not supported, so a home-built copy leaves the fault and the downtime with you. Connector orientation, crossed TX and RX, and the handshake loops inside that cable are exactly the details a redrawn or mirrored diagram gets wrong with no visible clue.
On the SLC side, an RS-232 device joins the DH-485 segment through the DH485-232C interface module. Rockwell publication 1747-IN006, that module's installation instructions, provides the pin assignments for wiring the mating connector of the cable used to interface a user device to the module's ports — the vendor-documented path, and unlike the panel cable it is meant to be wired from.
Distance sets the topology. RS-232 is point-to-point and good for about 15 m (50 ft) at full rate: inside that, one processor and one panel can talk directly. Past 50 ft, or with more than two devices, they belong on the DH-485 twisted-pair segment — RS-485 wiring, one straight daisy chain, stubs kept short, shield grounded at a single point, following the multidrop rules for connecting multiple HMIs to one PLC over 200 ft of serial.
Match the settings that must agree at both ends #
Three settings decide the link:
- Baud: one rate for the whole segment. DH-485 equipment ships at 19200 baud, with 9600 and 38400 as the other settings in common use; set every station alike.
- Node address: unique within 0–31 at every station, 32 stations maximum on the segment. Processors typically come set to node 1, and a duplicated address stops the token for everyone.
- Driver: in the panel project, the SLC 500 / DH-485 driver must be assigned to the physical port the cable plugs into. A driver left on the wrong port fails exactly like a broken cable.
An estimate to set expectations: at 19200 baud a serial byte costs roughly 10 bits, so raw wire throughput is about 1,920 bytes per second. A screen reading 100 integers — 200 bytes — needs at least 200 ÷ 1,920 ≈ 0.10 s on the wire before token waits and protocol overhead, so single-digit updates per second on a block that size is the hardware working, not a fault. When screens crawl rather than die, why an HMI freezes or updates slowly over serial walks that diagnosis.
Commission the link in stages #
- Write down the panel model, the SLC 500 port in use, the cable identity, and how many devices will share the segment.
- With power off at both ends, seat the cable, power up, and confirm both connectors are fully home. No home-built assembly anywhere in the chain at this stage.
- Set every station to the same baud rate and a unique node address inside 0–31, and put the driver on the wired port in the panel project.
- Load the project and open the panel's communication diagnostics; you want a read or receive counter climbing, not a status flag alone.
- Read one known address — an element whose value you can verify at the processor itself. Make it a read, not a write: a test write into the wrong element can move a live machine.
- Only after that read returns the expected value do the screen tags earn trust. Any single wrong value from here on is a binding fault, not a network fault.
Worked example: one processor, two panels #
An SLC 500 processor sits at node 1. Panel A connects 40 ft away on the RS-232 path: 40 ft is inside the 50 ft RS-232 reach, so the direct route through the interface module is legitimate. Panel B is then added in a second cabinet 150 ft away: 150 - 50 = 100 ft past the RS-232 spec, so panel B gets no RS-232 run — it joins the DH-485 twisted-pair segment instead. The node audit reads processor 1, panel A 2, panel B 3: 3 stations on a segment that allows 32, no duplicates, so token rotation is intact. Baud is confirmed at all three stations before the project is loaded.
What each symptom points to #
| Symptom | Layer | First move |
|---|---|---|
| Communication error, nothing on the panel updates | Link | Confirm the approved cable, then 19200 baud and unique node addresses at every station |
| Died when a second device was added | Node address | Renumber every station inside 0–31; duplicates stall the token |
| Link alive, one value wrong or frozen | Tag binding | Correct that object's data-file address and data type in the project |
| Link drops when the cable is touched | Physical | Re-terminate the connectors, shorten the stubs, check strain relief |
| Alive but slow on every screen | Poll load | Shrink the blocks and slow the poll rate the screen requests |
The decisions #
- An unknown or home-built cable in the chain: replace it — the approved panel assembly on one end, 1747-IN006 wiring on the other — before trusting any other test result.
- No communication on a known-good cable: set 19200 baud and unique node addresses at both ends; that clears most of what remains.
- Link alive, one value wrong: fix the object's address and data type in the project. The network is not the fault.
- An RS-232 run beyond 50 ft, or more than two devices on RS-232: move them onto the DH-485 twisted-pair segment.
- Link alive but every screen slow: cut the poll list. The arithmetic above, not the wiring, is the bottleneck.
Prove the cable, prove the settings, read one known address — in that order, every time.
Questions asked
Is the SLC side wiring under the same restriction as the panel cable?
No — the two ends are documented differently. The panel maker publishes the HMI-CAB-C84 pinout as a convenience and its documents state that unauthorized construction is not supported, so order that assembly. Rockwell publication 1747-IN006, by contrast, exists to be wired from: it gives the pin assignments for the connector that interfaces a user device to the DH485-232C interface module. One end you buy, the other end you build to the printed numbers.
What is the difference between DH-485 and DF1 on an SLC 500 port?
DH-485 is the token-passing multi-drop network that lets panels and programming tools share one twisted-pair segment; DF1 is a point-to-point serial protocol for one device on a port. The DH485-232C interface module documented in Rockwell publication 1747-IN006 couples the two worlds, which is what lets an RS-232 panel port join a DH-485 segment.
Can a panel and my programming software share the same DH-485 segment?
By design. The token passes around up to 32 stations, so a panel at node 2 and programming software at node 3 share one segment without unplugging anything. Two things break that arrangement: a duplicated node address, which stops the token, and a device hung off the daisy chain as a long stub, which disturbs the electrical segment rather than the protocol. Renumber the stations and shorten the stub.
The link is healthy but one screen number looks scrambled — where is the fault?
In that screen object, not in the network. A healthy DH-485 link that shows one wrong value almost always has an object aimed at the wrong data-file element or reading an integer as a floating-point number. Confirm the element address at the processor, set the object's data type to match the element, then re-read the same element on the processor display to close the fault out.
Does every device on the segment have to run the same baud rate?
Yes — DH-485 runs one network-wide rate, normally 19200 baud out of the box, with 9600 and 38400 as the other settings in common use. There is no per-device rate: one station left at 9600 stalls token rotation for the entire segment, and the symptom is identical to a dead cable. Set every station alike and record the rate on the panel drawing for the next technician.