What Is RS-485?

Back to Blog

Differential Signalling and Multidrop

RS-485's durability comes from differential (balanced) signalling. Data is sent as the voltage difference between two wires (A and B) rather than relative to ground. Because electrical noise picked up along a cable run affects both wires almost equally, the receiver rejects the common-mode noise and reads only the difference - which is why RS-485 works reliably over long runs in electrically noisy plant environments where a single-ended signal would fail.

It is also a multidrop bus: many devices share one pair of wires, typically in a daisy-chained line topology with termination resistors at each end to prevent reflections, and bias resistors to hold the idle state defined. A classic RS-485 segment supports up to 32 standard unit loads and around 1,200 metres, with the trade-off that speed decreases as distance increases. Only one device transmits at a time, so the higher-layer protocol handles addressing and turn-taking.

RS-485, Serial Protocols, and Modern SCADA

By contrast, RS-232 is a point-to-point, single-ended standard limited to short distances and one device per port - fine for a local console but unsuited to field networks. RS-485's multidrop, long-distance, noise-tolerant design is why it became the workhorse for field wiring, carrying Modbus RTU meters, PROFIBUS DP, DNP3 serial, and BACnet MS/TP across pump houses, tank farms, and skids.

In cloud and remote SCADA, serial devices reach the network through a serial-to-Ethernet gateway or an RTU/edge device with RS-485 ports. The gateway terminates the serial link, speaks the field protocol (often Modbus RTU) to the daisy-chained instruments, and forwards data upward over TCP/IP. A cloud SCADA like Merobix reads those serial devices over Modbus - polled locally by an edge gateway and pushed outbound - so legacy RS-485 meters and controllers gain cloud dashboards and historization without being replaced.

Wiring Details That Bite: Polarity, Common, and Shield

The first field lesson of RS-485 is that the A and B labels are not trustworthy: the standard's naming and common vendor practice disagree, so one manufacturer's A is another's B. If a freshly wired device refuses to answer, swapping the pair is a zero-cost first test. The second lesson is that the differential pair is not the whole circuit - a signal common still matters. Two panels grounded far apart can sit at different potentials, and without a common reference or galvanic isolation that difference rides on the receivers as common-mode voltage; isolated converters and isolated ports are the standard cure on long inter-building runs.

Shielding follows the same discipline as any instrument cable: keep the pair twisted right up to the terminals, keep it out of trays shared with power conductors, and land the shield per the site standard - the arguments for landing it at one end versus both are laid out in shield grounding at one end vs both ends. Carry the shield's continuity through every junction box; a shield that stops at the first splice protects only the first span.

Termination and Bias, Practically

Termination goes at the two physical ends of the bus and nowhere else, with resistors matched to the cable's characteristic impedance per the cable datasheet. Devices scattered mid-bus with their internal termination switched on load the line until the drivers cannot swing it - after polarity, this is the most common commissioning find, so walk the chain and check every device's termination setting. Biasing is the quieter cousin: something must hold the line in a defined idle state between transmissions, or receivers chatter on noise and frame errors appear from nowhere. Bias at one point on the segment, usually the master or the converter, and disable the bias that every other device helpfully enabled at the factory.

Planning a Segment: A Symbolic Example

Planning a segment is arithmetic you can do on a napkin. Each transceiver presents some fraction of a unit load, stated on its datasheet; the sum along the segment must stay within the 32-unit-load budget the standard's driver model assumes, and fractional-load transceivers are how modern buses fit more physical devices inside it. Suppose a segment must serve N meters, each presenting a load of L: if N times L stays under the budget and the cable run suits the chosen data rate, one daisy-chained segment with short stubs does the job. If either total busts its limit, a repeater starts a fresh electrical segment with its own termination and bias, and the count starts over. What the arithmetic never allows is a star: every branch point is a reflection generator, so keep the topology a line.

When the Bus Misbehaves

Faults on a working bus sort into recognizable signatures. Clean comms to nearby devices and errors to distant ones points at termination, bias, or cumulative cable trouble. Errors that appear only when a motor starts or a drive runs point at coupling - revisit shielding, separation, and routing. A bus that dies entirely after a lightning-season storm usually means failed transceivers, and argues for isolation and surge protection at the exposed ends. On Modbus segments the first visible symptom is often a rising count of Modbus RTU CRC errors, which is the protocol seeing the corrupted frames the wiring gave it.

Converters and gateways add one more failure family: turnaround timing. An auto-direction converter that releases the line late clips the start of the slave's reply, producing maddening intermittent timeouts that worsen at higher baud rates - the fixes are covered in turnaround problems with converters. When a segment resists all of the above, split it: disconnect half the chain and test, then halve again. Binary search finds the offending device or span faster than any amount of staring at configuration screens.

Frequently Asked Questions