Honeywell 621-9940C​ Serial I/O Module with Fault-State, Auto-Test and Normal Mode DIP Selection

Description

Application Scenarios:

A bulk-chemicals terminal loads road tankers across eight loading bays spread over roughly 400 metres of pipe rack. Each bay has its own local I/O — valve position feedback, permissive contacts, flow-meter pulse inputs, earthing interlocks and batch totaliser signals — and the plant’s IPC 620 processor sits in a control room at one end of the site. Running every one of those field cables back to the processor rack would mean hundreds of long parallel runs through a classified area, an installation cost the project could not justify and a maintenance burden the plant did not want.The architecture the plant built instead puts a small remote I/O card file at each loading bay, and each of those card files is governed by a 621-9940C. The module sits in slot N of its rack, gathers the state of every I/O module beside it, and reports it up a single serial channel to the 621-9939 SLM in the processor card file. Because the 621-9940C​ lets each slot in the rack be configured independently for 0, 8, 16 or 32 points, the plant can mix a 32-point digital input card, a 16-point output card and an 8-point analog card in the same rack without wasting addressing space — and because its starting address is set in increments of eight anywhere from 0 to 2,032, every bay gets a unique, non-overlapping block in the processor’s I/O map. When a new bay was added two years later, the commissioning work was essentially: set SW1 on the new 621-9940C​ to a free address, set SW3 through SW5 for the slot densities, terminate the channel properly, and the processor saw the new rack.That pattern is the reason the 621-9940C​ exists and the reason it still matters. It appears wherever I/O has to be distributed away from the processor — conveyor lines in mining and cement, remote wellhead and pump stations in oil and gas, distributed water and wastewater lift stations, long packaging and material-handling lines, and multi-bay batch plants. In every one of these, the pain point it resolves is the same: the cost, complexity and electrical noise exposure of pulling every field signal back to a central rack. One serial channel replaces hundreds of home-run cables, and the 621-9940C​ is the card that makes that channel work.

Parameter:

Technical Principles and Innovative Values:

• Innovation Point 1 — Address allocation that the processor discovers automatically. The 621-9940C​ does not need to be declared to the controller in a configuration file. At start-up the SLM queries every SIOM on the channel, reads its starting address and the number of allocated addresses, and builds the access sequence from what it finds. That self-describing behaviour means a replacement module can be configured in minutes with nothing but a set of DIP switches — no programming terminal, no software licence, no download.
• Innovation Point 2 — Mixed I/O density within a single rack. Each slot in a card file governed by the 621-9940C​ can be independently set to 0, 8, 16 or 32 points via SW3, SW4 and SW5. This is a genuinely unusual capability: most remote I/O systems force a uniform slot allocation, so a rack holding one 32-point card and three 8-point cards either wastes addressing space or wastes physical slots. Mixing densities lets a rack be built to match the actual signal count, which lowers both hardware cost and addressing consumption.
• Innovation Point 3 — Deterministic fault behaviour configured in hardware. SW2 on the 621-9940C​ sets what the rack’s outputs do when a fault occurs. This matters more than it sounds: for a remote loading bay, the difference between “outputs hold last state” and “outputs drop to zero” is the difference between a controlled pause and an unintended valve closure on a live transfer. Having that decision encoded in hardware switches rather than buried in logic means it survives a program download and cannot be changed by accident.
• Innovation Point 4 — A distributed architecture built on one cable. The entire economic case for the 621-9940C​ rests on this: a remote card file communicates with the processor over a single specified serial cable rather than through hundreds of individual field runs. The reduction is not just copper and tray — it is terminations, marshalling, drawing time, commissioning labour and, critically, the amount of long cable acting as an antenna for electrical noise in a plant environment.
• Innovation Point 5 — Built-in test mode. The 621-9940C​ can be switched between normal and auto-test operation from SW2, which allows a technician to exercise the card file’s own diagnostics without a programming device. On a platform of this vintage, that is a meaningful advantage — it turns “is the rack healthy?” into a question answerable at the rack, in minutes, with the tool you already have in your pocket.
• Innovation Point 6 — Documented physical layer discipline. Honeywell specifies the serial cable type (Belden 9729 or equivalent) and the termination value (200 Ω at ½ watt) rather than leaving it to the installer. In practice this is why well-built IPC 620 serial channels run for decades: the physical layer is specified, the terminations are known, and the failure modes are documented. When a channel does misbehave, the first three things to check are written down.
• Innovation Point 7 — Clear suffix semantics that prevent specification errors. Honeywell’s suffix convention is explicit: “R” means the module can be removed and inserted under power, “C” means CE Mark compliance. Understanding that the 621-9940C​ carries the CE suffix — and that the 621-9938RC carries both — prevents the single most common ordering mistake on this platform, which is specifying a plain 621-9940 into a rack that requires the CE-compliant build.