How to Select an RS-485 or RS-422 Transceiver for Industrial and Long-Line Designs

Why this interface refuses to go away #

RS-485 is the quiet workhorse of industrial communication. Where a UART-to-USB cable fails after a few meters and Ethernet feels heavy for a chain of sensors, a differential transceiver keeps a signal clean across hundreds of meters of twisted pair in electrically hostile environments. If your design talks to motor drives, energy meters, building controllers, or a row of field instruments, the question is rarely *whether* to use RS-485 but *which* transceiver to put on the board.

RS-422 and RS-485 share the same differential signaling physics but differ in topology. Understanding that difference is the first selection decision, not a detail you resolve later.

RS-485 vs RS-422: the topology that decides everything #

Both standards use a balanced differential pair (the A/B or Y/Z lines) referenced to a local ground. The distinction is in how many drivers may sit on the bus:

• RS-422 allows one driver and up to ten receivers. It is effectively point-to-point or one-transmitter-to-many-receivers. Full-duplex wiring (separate pairs for each direction) is common.
• RS-485 allows multiple drivers on the same pair, so it supports true multipoint networks — many nodes that can both transmit and receive on a shared bus. Half-duplex (one pair, direction switched) is the typical form.

If your network has more than two devices sharing a cable and any of them must talk back, you need RS-485. If you only ever have one sender feeding several listeners, RS-422 is sufficient and slightly simpler. Most modern "RS-485" transceivers are built to the TIA-485-A specification and also interoperate with RS-422 receivers on the same pair.

Data rate versus cable length #

The single most misunderstood constraint is the trade-off between speed and distance. A differential transceiver does not have a fixed maximum rate; the limit is set by the cable's capacitance and the driver's slew rate. Practical implementations reach tens of megabits per second over a few meters, while near the maximum bus length (commonly cited around 1200 m) the usable rate falls to roughly 100 kbps or less.

This is why a part's headline "up to 20 Mbps" is only meaningful together with your cable plan. Choose a high-slew-rate driver for short, fast links and a controlled-slew-rate (limited dV/dt) driver for long, noisy runs where ringing would otherwise break the signal.

Half-duplex vs full-duplex wiring #

Half-duplex RS-485 uses one pair and a direction-control signal (DE/RE) from the MCU. It is cheaper in cable and connectors and fits the vast majority of sensor and metering networks. The software must never drive the bus while another node is transmitting, which means a clean media-access scheme — often a simple master polls slaves, or a token passes between nodes.

Full-duplex uses two pairs and lets a node send and receive at the same time. You reach for it when protocol latency matters or when you are really building an RS-422-style link. For a new multipoint design, half-duplex is the default unless a specific timing need forces the second pair.

Fail-safe biasing: what the bus does when nothing is driving it #

An undriven RS-485 pair floats to an undefined state. Early transceivers needed external bias resistors (a pull-up on B, pull-down on A) plus a terminating resistor so an idle bus rested at a known logic level. Modern transceivers integrate *fail-safe* receive thresholds, so an open, shorted, or idle bus reliably reads a logic high instead of glitching.

Specify an integrated-fail-safe part unless you have a concrete reason not to — it removes a handful of passive components and a common source of field returns where the bus "works until it goes quiet." If you must mix old and new nodes, plan the biasing for the worst-case node.

Termination and stub length #

The bus characteristic impedance is typically about 120 Ω, so a termination resistor of that value belongs at each *electrical end* of the run. Terminating every node loads the bus and reduces the number of devices you can hang on it; terminating none invites reflections at high rates.

Stubs — the short cables branching from the main run to each node — should be kept as short as the enclosure allows. A long stub is a stub antenna that picks up noise and reflects signals back into the bus. For a clean layout, think of the bus as a backbone with nodes tapping off as close as possible, not as a chain of equal segments.

Isolation: breaking ground loops in real installations #

On a bench, RS-485 often runs without isolation. In the field, nodes may sit hundreds of meters apart with ground potentials that differ by volts — exactly the condition that destroys transceivers and corrupts data. Galvanic isolation (a digital isolator plus an isolated power path) breaks that loop and is the difference between a link that survives a motor-start surge and one that resets randomly.

For a fixed installation with disparate earth grounds, treat isolation as a requirement, not an option. The trade-off is board area, cost, and the need for an isolated supply (often a small isolated DC-DC module or a transformer-based regulator). Many qualified industrial designs pair the transceiver with a dedicated isolator rather than relying on the transceiver's own I/O tolerance.

EMC, ESD and transient protection #

RS-485 nodes in industrial and outdoor settings face ESD, fast transients, and surge. A robust design adds:

• ESD clamps on the bus pins (often integrated, but external TVS diodes help at the connector).
• A common-mode choke to suppress radiated and conducted noise.
• Proper grounding of the cable shield at one end (or both with care) to avoid creating another ground loop.

These belong in the schematic from the start. Retrofitting protection after an EMC test failure is expensive and rarely as clean as a planned layout.

Node count and unit load #

The original specification assumed a maximum of 32 "unit loads" on a bus. A standard transceiver presents one unit load, so 32 nodes was the classic limit. Modern "fractional unit load" receivers (1/8 unit load is common) let you place many more nodes — often over 200 — on the same cable. If your network is large, confirm the unit-load rating of every transceiver you mix, because the weakest node sets the bus capacity.

A worked example: a daisy-chained sensor network #

Consider a building-automation link with 24 temperature and energy nodes spread along 250 m of conduit, polled by a master every second. A sensible choice:

• Half-duplex RS-485 on a single shielded twisted pair, master polling each node in turn.
• A controlled-slew-rate driver rated for a few megabits (far above the needed rate, leaving margin against cable loss).
• Integrated fail-safe to keep the idle bus stable between polls.
• 120 Ω termination at the two physical ends, short stubs into each node.
• Isolation at the master (which sits in a different cabinet with a separate earth) and TVS at each field node.

A part such as the Texas Instruments SN65HVD11DR illustrates the profile well: a 3.3 V RS-485 transceiver specified up to 10 Mbps with fail-safe and a unit-load rating that supports many nodes — a concrete starting point for evaluation, not an automatic approval.

Sourcing and qualification notes #

As with any line item, freeze the exact orderable code, package, and temperature grade, and verify date code and traceability before release. For long-life industrial products, confirm the transceiver's lifecycle status and identify a second source or pin-compatible alternative before the design is locked — a single-source interface part is a future shortage risk exactly when the product is hardest to re-spin.

Cross-check the selected device against the full interface catalogue and, where the design also carries CAN, keep the two transceiver choices consistent in supply and qualification terms.

Conclusion #

Selecting an RS-485 or RS-422 transceiver is a sequence of small, linked decisions: multipoint or point-to-point first, then half- or full-duplex, then a driver slew rate matched to your cable length and data rate, then fail-safe and termination sized to the bus, and finally isolation and EMC protection scaled to the installation. Start from the network topology and the longest run, not from a headline data-rate number, and the rest of the choices follow. Get those right and RS-485 will keep your nodes talking reliably for the life of the product.

Sources:

• Texas Instruments SN65HVD11 / SN65HVD3082 RS-485 transceiver datasheets and application notes (ti.com)
• TIA-485-A and TIA-422-B interface standards (tiaonline.org)