13.2 Buses, Topologies, and Digital Protocols

Why buses sit in 4.A with analog

Specification 4.A is not only 4–20 mA. Once the process variable is a number on a wire, the exam asks how that number is shared: analog overlay, fieldbus segment, or industrial Ethernet. You are not being tested on OSI-layer trivia. You are being tested on physical layer, power, topology, and when a protocol is the wrong tool.

Protocol examples (what you actually specify them for)

HART (Highway Addressable Remote Transducer) is a slow digital frequency-shift-keyed channel (Bell 202, 1200 baud, 1200/2200 Hz) riding on a 4–20 mA analog. The analog current remains the real-time process variable for the regulatory loop. The digital channel carries tagging, range, diagnostics, and secondary variables. Point-to-point is the normal control case. Multidrop HART parks transmitters at a fixed 4 mA and talks digitally only — do not use that mode for a modulating process variable into an analog input. HART needs enough loop resistance (commonly about 230–600 Ω total, often the 250 Ω analog-input burden) for the FSK waveform to develop.

FOUNDATION Fieldbus H1 puts many devices on one pair at 31.25 kbps (IEC 61158-2). Devices are bus-powered. Function blocks can run in the field (control in the field). There is no 4–20 mA process variable unless you add a converter.

PROFIBUS PA uses the same IEC 61158-2 physical layer as FF H1 (31.25 kbps, trunk and spur, bus power, intrinsically safe-capable segments). PROFIBUS DP is a different physical layer: RS-485, much faster, typically a programmable logic controller to remote I/O and drives — not the two-wire process spur.

Modbus RTU is a master–slave register protocol on serial (RS-232 / RS-485). Modbus TCP carries the same application data on Ethernet (commonly TCP port 502). Easy to map; no publisher/subscriber control-in-the-field story.

EtherNet/IP carries CIP (Common Industrial Protocol) on Ethernet: plant backbone, variable-frequency drives, remote I/O, and some transmitters with Ethernet. Physical plant wiring is usually a star through switches; Device Level Ring (DLR) is the common availability topology.

DeviceNet is CIP on Controller Area Network (CAN), with 24 VDC power and data in one trunk. Baud versus length is hard: typical 125 kbps / 500 m, 250 kbps / 250 m, 500 kbps / 100 m. Drop length is limited. Think device-level I/O on a machine or skid, not a 2000 m tank farm.

Topologies

• Point-to-point. Classic 4–20 mA and HART analog mode. One pair, one process variable, simple troubleshooting. A cut pair loses one loop, not a segment.
• Bus (trunk and spur). FF H1, PROFIBUS PA, RS-485 daisy-chain. One trunk, many nodes, terminators at both trunk ends. A trunk fault can take many devices.
• Star. Ethernet switches, some analog marshalling. A switch failure isolates one arm unless you dual-home.
• Ring. Each switch or device has two paths. Used for availability (DLR, media redundancy).
• Redundant ring. Dual rings or equivalent so that a single cable cut must not lose the segment. Higher cost; specify only when the availability target requires it.

Do not put RS-485 terminators on a 4–20 mA pair, and do not treat a HART analog home run as a fieldbus trunk just because the transmitter is “smart.”

Limitations: segment current, spur length, terminators

On FF H1 / PA, segment current — not a brochure “32 devices” — usually sets how many transmitters you can hang. A power conditioner might supply a few hundred milliamperes; each device draws on the order of 10–20 mA. Add inrush and intrinsically safe remaining-current margins. If the stem gives conditioner current and per-device draw, divide and leave margin; do not assume the theoretical node count.

Spur length shrinks as you add spurs. A single spur might be allowed up to about 120 m on type A cable in a lightly loaded segment; a forest of spurs forces much shorter stubs. The exam item will give lengths if it wants a numeric pass/fail. Home-run each transmitter to the trunk with a short spur; do not daisy-chain spurs into a second undocumented trunk.

Terminators (typically 100 Ω at each end of an FF H1 / PA / RS-485 trunk) prevent reflections. One missing terminator → reflections and CRC errors. Three terminators (someone left a “spare” enabled on a device) → over-damping and still errors. Analog 4–20 mA loops do not get RS-485 terminators.

Worked decision: HART overlay versus a fieldbus segment

Plant A — overlay is enough. A crude unit already has 40 home-run 4–20 mA pairs into HART-capable distributed control system analog inputs. You need to pull serial numbers, rerange from the shop, and read diagnostic status. Regulatory control must keep using the analog process variable with no new segment engineering.

Decision: HART overlay. The copper and analog I/O already exist. Digital traffic is slow (on the order of one to two updates per second on a busy analog loop), which is acceptable for configuration and secondary variables, not for replacing the analog PID with a HART-only process variable.

Plant B — a fieldbus segment is the point. A new 12-transmitter tank farm is 400 m from the rack room. You want one (or two redundant) pairs, bus-powered devices, and the option to run a PID in a valve positioner.

Decision: FOUNDATION Fieldbus H1 or PROFIBUS PA, with type A cable, two trunk terminators, a current budget, and a spur schedule. HART overlay would still require 12 analog pairs or a HART multiplexer architecture you do not have yet. DeviceNet’s 500 m / 125 kbps trunk sounds close on distance, but it is the wrong physical layer and power model for that tank farm.

If the stem says “keep the existing analog input cards and home runs,” HART is the answer. If it says “one pair, many devices, bus-powered, control in the field,” fieldbus is the answer. If it says “PLC rack I/O at megabits per second,” that is DP or Ethernet, not PA.

[!TIP]

Exam habit

Name the physical layer before the logo. PA is not DP. HART is not a replacement for 4–20 mA in point-to-point control. Terminators belong on trunks, not on every spur.