Automation Glossary • Half-Duplex vs Full-Duplex

What Is Half-Duplex vs Full-Duplex?

Merobix Engineering • • 5 min read

Half-duplex and full-duplex describe whether a communication link can carry data in both directions at once or only one direction at a time. The distinction is not academic in field automation, because it governs how a two-wire RS-485 bus and a serial radio behave and how much delay creeps into every poll and reply. Understanding it explains why a link that looks fast on paper can feel sluggish once real turnaround delays are counted, and why polling throughput is often the real bottleneck.

Back to Blog

Half-Duplex vs Full-Duplex in one line: Full-duplex links transmit and receive simultaneously on separate paths, while half-duplex links share one path and can only send in one direction at a time, requiring the ends to take turns. Two-wire RS-485 and most serial radios are half-duplex, which adds a turnaround or key-up delay on every reversal that limits polling throughput.

Simplex, Half-Duplex, and Full-Duplex

There are three modes of directional flow. Simplex sends in one direction only, like a broadcast, with no return path at all. Half-duplex allows both directions but not at the same instant, so the two ends alternate who is talking. Full-duplex carries traffic both ways simultaneously, typically because each direction has its own physical path or frequency.

The mode is set by the wiring and the medium, not by the protocol running on top. Two-wire RS-485 shares a single differential pair for both directions, which forces half-duplex operation because only one transmitter may drive the pair at a time. Four-wire RS-485 dedicates one pair to each direction, allowing full-duplex, and RS-232 with separate transmit and receive lines is inherently full-duplex on a short cable.

Radio links follow the same logic at the RF layer. A single-frequency radio channel is half-duplex, since a transmitter and receiver cannot use the same frequency at the same time without jamming each other, whereas a paired-frequency or split-band system can approach full-duplex. Most low-cost telemetry radios in the field are half-duplex, so field comms design almost always has to account for taking turns.

Turnaround, Key-Up, and Polling Throughput

The hidden cost of half-duplex is the time lost every time the link reverses direction. On a two-wire RS-485 bus, a device must finish transmitting, release the line, and let the driver settle before the other end may respond, and the polling master must wait out that turnaround before it listens. Multiply that small gap by hundreds of polls per minute and it becomes a real slice of the cycle time.

Radios make the penalty far larger. A half-duplex serial radio has to key up its transmitter, let the RF settle, and give the far receiver time to lock before any payload gets through, then unkey and reverse the whole process for the reply. This key-up delay happens on every single message exchange and is often much longer than the time spent actually sending data, which is why a radio poll can feel slow even at a healthy baud rate.

In a classic master-slave polling scheme, a full-duplex link lets the master pipeline requests and replies more freely, while a half-duplex link forces a strict request, turnaround, reply, turnaround rhythm. The fix is rarely to raise baud, which does nothing to shrink a fixed key-up delay; instead you reduce the number of reversals by reading more registers per request, polling fewer devices per cycle, or spacing scans to match the link's real turnaround budget.

Duplex Choices in SCADA Field Comms

When a cloud SCADA gateway polls remote sites, the duplex mode of the link between it and the field devices sets a hard ceiling on how often data can refresh. A gateway driving a two-wire RS-485 multidrop of several RTUs, or reaching them over a half-duplex radio, must sequence every exchange and absorb a turnaround on each one. Designing the poll schedule around that reality is what keeps a dashboard feeling live rather than laggy.

This is why site design decisions like two-wire versus four-wire RS-485, or single-frequency versus split-band radio, ripple all the way up to the user experience in the cloud. A half-duplex choice is often the right one for cost and simplicity, but it means the platform and the gateway must be tolerant of the turnaround delays and should not misread a slow-but-healthy link as a fault. Sensible timeouts sized to the link's turnaround are essential.

A cloud platform helps by surfacing per-device scan times and communication health so operators can see when a half-duplex segment is saturated. If a site's update rate drifts because too many devices share one turnaround-limited bus, the data makes the cause obvious and points to the fix, whether that is splitting the bus, trimming the register list, or moving a chatty device to its own link. The duplex mode is invisible on the surface, but its effects show up clearly in the numbers.

Frequently Asked Questions

Is RS-485 half-duplex or full-duplex?

It depends on the wiring. Two-wire RS-485 shares a single differential pair for both directions and is therefore half-duplex, so devices must take turns transmitting. Four-wire RS-485 uses a separate pair for each direction and can run full-duplex. Most industrial Modbus RTU multidrop buses are two-wire half-duplex.

Why is my half-duplex radio link slow even at a high baud rate?

Half-duplex radios add a fixed key-up and receiver-lock delay every time the link reverses direction, and that delay is unrelated to baud rate. Raising baud only shrinks the time spent sending the payload, not the turnaround, so past a point it barely helps. Reducing the number of message reversals, by reading more data per request, usually does more for throughput.

What is turnaround time in a serial link?

Turnaround time is the delay incurred when a half-duplex link switches from one device transmitting to the other, covering line settling on RS-485 or transmitter key-up and receiver lock on a radio. During turnaround no useful data flows, so a large turnaround limits how quickly a polling master can cycle through its devices. It is a key factor in sizing poll schedules and timeouts.

Sources and verification

This page references the protocol specifications published by the organizations below. Editions, product capabilities, and documentation change over time - confirm current requirements and specifications directly with the source.

Last reviewed: July 27, 2026. Merobix is not affiliated with, endorsed by, or sponsored by these organizations; their names are used only to identify the standards and products discussed.

From Definitions to a Live Dashboard

Merobix reads your field devices into a cloud SCADA - the real thing behind these terms, live in days from any browser.

Request a Free Demo +1 (903) 307-7300
More in Automation Glossary
Serial Flow Control  •  Null Modem Cable  •  Power over Ethernet (PoE)  •  Signal-to-Noise Ratio  •  Serial Device Server  •  Beta Ratio  •  All Automation Glossary →
Free SCADA operator training
Merobix University - 70 video lessons & 261 quiz questions, from first login to compliance reporting. No demo call required.
Start free →