What Is Modbus Serial Response Delay and Turnaround Time?

The 3.5-Character Silent Interval

Modbus RTU has no start or stop marker inside its frames the way some protocols do; instead it marks the boundary between one message and the next with silence. The rule is that a gap of at least three and a half character times of quiet on the line signals the end of a frame, and any new activity after that silence is treated as the start of a new frame. A character time is simply how long it takes to send one character at the current baud rate, so the required silent interval scales with baud: faster links have shorter character times and therefore shorter required gaps, and slower links need longer ones.

This silent interval, often called the inter-frame gap, is what lets a receiver know where one message stops and the next begins on a stream that would otherwise be an unbroken run of bytes. Within a single frame, the bytes must arrive close together, and a gap of more than about one and a half character times inside a frame is itself treated as an error, so RTU timing is fairly strict about both keeping a frame's bytes together and separating one frame from the next by enough silence. Devices rely on these timings to frame messages correctly.

When the silent interval is not respected, framing breaks. If a master fires a new request too soon after the previous reply, before the required quiet has elapsed, a slave can run the two together or misinterpret where the frame starts, producing corruption or a missed message rather than a clean exchange. This is a different failure from a total timeout: the device is present and fast, but the master crowded the frames. Honouring at least the 3.5-character silence between transactions, especially at low baud rates where that gap is longer in real time, is the first piece of getting RTU timing right.

Direction Turnaround on a Shared Line

The second timing concern is turnaround on a half-duplex link. Two-wire RS-485 uses a single pair shared by every device, so only one device can drive the line at a time, and a device must switch its transceiver from transmit mode to receive mode, or the reverse, around each exchange. That switch is not instantaneous. After the master finishes transmitting a request, its driver has to disable its transmitter and enable its receiver before it can hear the reply, and the slave has to enable its transmitter before it starts answering. This transmit-to-receive changeover is the turnaround time.

If the turnaround is mishandled, the reply can be lost or clipped. Should the master still be effectively holding the line, or not yet listening, when the slave begins to answer, the beginning of the reply can be missed or the two can contend on the shared pair. Some hardware and drivers need a deliberate short delay after transmitting before they start receiving, and some need a delay before releasing the line, precisely to give the transceivers time to change direction cleanly. On a bus with slow or cheap converters, an inadequate turnaround delay shows up as intermittent missed replies that improve when a small delay is added.

Radios and modems stretch this further, because the link itself adds latency and its own keying delays. A licensed radio or a serial-to-cellular hop may need time to key up its transmitter, carry the request, and key down before the reply travels back, all on top of the RS-485 direction change at each end. The practical consequence is that the wait the master must allow between finishing its request and expecting the reply to arrive is much longer over a radio than over a direct wire, and the turnaround delays have to be set generously enough to cover the radio's keying as well as the transceiver changeover, or the start of replies gets lost.

Tuning the Delay Versus the Total Timeout in SCADA

It is important to separate these frame-timing delays from the overall request timeout, because they solve different problems. The total response timeout is how long the master waits before deciding a device is not responding at all and giving up, and it is covered by the general Modbus timeout topic. The delays discussed here are smaller and structural: the silent interval between frames and the turnaround before the reply can be received. A link can have a perfectly generous total timeout and still fail because the frame gaps are too short or the turnaround delay is too small, so tuning one does not substitute for tuning the other.

Tuning these delays means matching them to the physical link. On a fast, direct RS-485 bus the built-in timings are usually adequate and little adjustment is needed. On a low-baud link the character time is long, so the required silent interval is longer in real time and the master should space its transactions accordingly. On a radio or cellular path, add turnaround and inter-frame delay generously to cover keying and latency, and reduce how aggressively the master polls, since crowding a slow link with rapid back-to-back requests is exactly what breaks framing. Many drivers expose settings such as a delay between polls, an inter-frame delay, or a transmit-to-receive turnaround delay precisely so you can adapt to a slow bus.

For SCADA and remote monitoring this frame-level tuning is what makes a marginal serial link dependable rather than erratic. A cloud SCADA platform such as Merobix reaching a remote RTU over a radio or cellular serial link relies on these delays being set to match the path, because a configuration copied from a fast bench bus will crowd the frames and lose replies once it runs over a slow real-world link. Getting the silent interval and turnaround right, distinct from and in addition to a sensible total timeout, is what turns a serial site that communicates intermittently into one that polls cleanly, which is the difference between data you can trust and a site that keeps flagging bad quality for no visible reason.

Frequently Asked Questions