IMDHMIControls LibraryShop main store

Troubleshooting

Recovering alarm data from an undocumented SLC 5/04

PLC programming Published 8 min read

Short answer

An upload from an SLC 5/04 returns the running logic and data files but no symbols or descriptions, because the processor never stores them. Recover alarm meaning from the operator interface tag database, any older backup and the electrical drawings, trace each alarm backward through the logic, then map only explained addresses into the gateway.

On this page
  1. What an upload returns, and what it never returns
  2. Start from the operator interface, not the gateway
  3. Reserve a DH+ node address before the gateway joins
  4. Trace the remaining alarms backward through the logic
  5. Reading alarm bit B3:12/4 through a Modbus gateway
  6. Sort each symptom before you blame the network
  7. Map only what you can explain
  8. Questions asked

You will not have to poll registers blind and hope. An upload from an SLC 5/04 hands you the running logic and the data files, and nothing else: the processor does not store address symbols or descriptions, so a clean upload proves you can read the controller, not that you understand the machine. Recover the meaning from three sources you may already own — the operator interface's tag database, any older program backup in the maintenance archives, and the electrical drawings — then trace each alarm backward through the logic until you can say what sets it, what latches it and what clears it. Map only explained addresses into the DH+ gateway. The tools are slated for replacement within a couple of years, so keep the whole recovery read-only.

What an upload returns, and what it never returns #

An SLC 5/04 processor holds the executable program and the data-file structure, and nothing about the words that went with them. Address symbols, descriptions and rung comments live in the offline database on the programming machine; the controller has no place to keep them, so an upload delivers ladder files, data files and configuration, and no documentation with them. A clean upload therefore answers one question — can I read this controller? — and leaves the real question, do I understand this machine?, untouched.

Before you touch the processor, hunt for backups: maintenance archives, the machine documentation folder, stored commissioning files, whatever the defunct vendor left on a handoff disk. Any old file helps even when the logic has drifted, because the upload-with-old-file method opens the older project, uploads the current program into that same file, and lets the old descriptions stand on every address the two versions still share. Verify each borrowed comment against the live logic and the screen before you trust it; a comment sitting on a repurposed address is worse than none.

RSLogix 500 is licensed software, so plan to buy a license or to take a temporary one from the local distributor, and have that settled before your window opens. The programming connection on this family is serial; a converter cable such as 9509-USBCBL-CP3 is the kind of part to confirm with the distributor against your processor's actual port. Then upload only — never download to a running processor — and save the uploaded file under its own name, leaving every older project unchanged.

Start from the operator interface, not the gateway #

The shortest route from an operator-visible condition to a PLC address runs through the interface project, so open that before you study a single rung. List what the receiving system needs, then pull the address behind each item:

  • Every alarm definition with its displayed text; the text becomes that address's starting description.
  • Numeric displays and their engineering units, with the address behind each one.
  • Status animations, with the ranges that drive them.
  • Pushbuttons, marked as writes: a button's address is a command or an acknowledgment, not a state to monitor.
  • Expressions and scaling, wherever a displayed value is computed from more than one PLC address.
  • The target controller recorded beside every address, since an interface may talk to more than one device.

Export the tag database if the interface software allows it, and keep the export beside the uploaded project. If the interface supplies every address and its meaning, the logic work shrinks to verification against machine states. If only a running interface exists with no editable project, its screens still tell you what to hunt for in the ladder. The same recovery exists on other platforms; the notes on recovering Concept 2.6 PLC symbol descriptions cover that identical problem on a different controller family.

Reserve a DH+ node address before the gateway joins #

DH+ carries up to 64 stations at addresses 0 through 63, so the node question has a bounded answer. Browse the network with your programming software's driver, write down every station that answers, and put the gateway on any number outside that list; the position itself carries no meaning to the protocol. Match the gateway's baud rate setting to the stations already running before it joins, and write the finished address map into the panel drawings so the next person inherits it. When the receiving side is a Logix controller that reads the same values itself, the worked message path in CompactLogix MSG to SLC 5/04 through a DH+ gateway shows the addressing from that end.

Trace the remaining alarms backward through the logic #

Electrical drawings give you a second reference independent of the software. Label the I/O from the prints and the terminal identifications, then verify each label online during normal cycles; where the prints never existed, the strip-by-strip work in recovering coil feeder wiring without drawings rebuilds the same knowledge from the cabinet. From there the trace is a fixed sequence:

  1. Find the logic that drives a known alarm or status address; the cross-reference report lists every rung that reads or writes it.
  2. Walk upstream: the initiating condition, any delay timer, the latch, the reset path, and any suppression or enable contact.
  3. Treat a timer preceded by permissive contacts as a common alarm pattern to search for, not a rule; other alarms latch instantly or carry separate acknowledge and reset bits.
  4. Separate the active alarm bit from alarm history and acknowledgment bits, and map the one the receiving system actually means.
  5. Re-walk the program after each pass; shared latches and suppression conditions reinterpret earlier finds.

Forcing an unknown input or commanding a fault to watch a bit move can actuate the machine, so keep the recovery read-only through normal cycles and leave any deliberate test for a controlled window under an approved procedure.

Reading alarm bit B3:12/4 through a Modbus gateway #

The inputs, and where each comes from:

  • Alarm address B3:12/4: an example address in the real SLC format — binary file B3, word 12, bit 4 — the kind of entry an interface tag export hands you. Treat the word and bit numbers as example inputs, not findings from your machine.
  • Word width: one B3 word is 16 bits, which is the documented SLC 500 data-file layout.
  • Gateway map: your map assigns B3:12 to one Modbus holding register, so each read returns that whole 16-bit word at once.
  • Example read-back: 2293, an illustrative value used here to show the arithmetic.

The rule in words: bit n of a 16-bit word carries the weight 2^n, so an alarm on bit 4 is active exactly when the word's value masked with 2^4 equals 2^4.

The arithmetic: 2^4 = 2 × 2 × 2 × 2 = 16. The example read returns 2293, which is 08F5 in hexadecimal and 1000 1111 0101 in binary with bit 11 the leftmost digit. Bit 4 is the fifth digit from the right and reads 1, so 2293 AND 16 = 16.

The result: on the example read, the alarm is active.

The decision, with its threshold: map B3:12 as one register and test every alarm in that word with its own mask — bit 0 weighs 1, bit 4 weighs 16, bit 15 weighs 32768 — and a masked result equal to the mask means active while a masked result of 0 means that alarm is clear. The example word carries seven set bits (0, 2, 4, 5, 6, 7 and 11), so it is seven alarms sharing one register; never read it as an alarm code of 2293. If the receiving system cannot mask bits, run the mask in its own logic or pick a gateway that masks on the wire.

Sort each symptom before you blame the network #

Each observation points at a different failure, and mixing them up wastes downtime you do not have:

What you observe What it separates Next check
Upload succeeds, every description blank The controller never held the offline documentation Mine the interface export, archives and prints
A value reads fine, its meaning is unclear Communications work; the semantic map is missing Cross-reference the address and compare it with the screen
Screen value differs from the raw PLC value The interface applies scaling or an expression Open the interface expression and its source address
The screen shows question marks A communications timeout, not missing meaning Diagnose the comms path first
The gateway reads nothing at all DH+ path, node address, baud rate or map is wrong Diagnose the network before the ladder

A screen full of question marks is a communications fault with its own diagnosis; why HMI values show question marks when PLC comms time out works that path first. And no bridge delivers meaning along with the values: the reason Modbus cannot transfer a PLC ladder program over TCP is the same reason a gateway read returns numbers without descriptions — the protocol moves data-table values, never project documentation.

Map only what you can explain #

The decision for every address is the same test: it enters the production map when you can say what sets it, what it means to the receiving system, and what clears it; until then it stays out. Polling registers you cannot explain produces confident nonsense, and with the tools slated for replacement in a couple of years, a partial correct map beats a complete wrong one every time. Verify each mapped alarm against one normal occurrence — the bit's timing should match the operator's report and the interface's alarm history — and record the meaning in the project and on the drawings. Where the receiving system is a PC that logs, alarms or re-serves the mapped data beside the controller, that machine belongs in the cabinet as an industrial mini PC, specified for the enclosure rather than parked on the shop floor in an office case.

Questions asked

Do I need RSLogix 500 just to read alarms through the gateway?

Only for the recovery work. Once an address is identified, documented and mapped, the gateway reads it on its own and the programming software can come off the bench. You need the license for the upload, the cross-reference work and the documentation edits, so buy it or take a temporary license from the distributor for that phase, then put the bench back to its normal duties.

Can another device share the gateway's DH+ node address?

No. DH+ is a token-passing network, and every station needs its own address within the 0-63 range; a duplicate makes stations drop off intermittently and is miserable to diagnose. Browse the live network before you pick, take an unused number for the gateway, and record the assignment on the drawings so the next device added to the line does not collide with it.

Which SLC data files usually hold alarms?

Alarm bits most often sit in the B3 binary file, with alarm codes and counts in N7 integer words, but treat that as a search pattern rather than a rule. The default SLC 500 layout is O0 outputs, I1 inputs, S2 status, B3 binary, T4 timers, C5 counters, R6 control, N7 integers and F8 floats. The cross-reference report tells you where a given address is actually used.

How many passes through undocumented logic does this take?

Expect several. Someone who has read a lot of code usually circles the program three or more times: a first pass to label I/O from the drawings, a second to trace alarms from the interface addresses, and later passes that reinterpret earlier finds as shared latches and suppression conditions turn up. Each pass is cheap once the upload and the tag export sit in one project.

Keep reading

  • Troubleshooting

    Why Modbus cannot transfer a PLC ladder program over TCP

    No Modbus address reaches ladder memory: the application protocol defines four data tables — coils, discrete inputs, input registers, holding registers — and program storage is not in any of them. Swapping RTU for TCP changes transport, not protocol. On DirectLogic PLCs the program moves only over K-Sequence or DirectNET through the vendor's programming tools; Modbus stays for mapped user data.

    PLC programming

  • Troubleshooting

    PLC-5 not visible in RSLinx after processor replacement

    A replacement PLC-5 stays invisible in RSLinx because the 1785-ENET sidecar answers the network on its own while the processor waits for an application that configures the Ethernet channel 3A. Leave the non-isolated serial port disconnected until the cable is checked, bring the processor up over serial DF1 or DH+, download the saved program, and the Ethernet path returns.

    PLC programming

  • How-to

    CompactLogix MSG to SLC 5/04 through a DH+ gateway

    The CIP path in the CompactLogix MSG ends at the 1756-DHRIO module in the remote chassis. On the MSG configuration tab, change the communication method from CIP to DH+, then enter Channel A and the SLC 5/04 node address in the DH+ fields; the SLC is never part of the path itself.

    Industrial networks