No. A 100 ft two-wire RS-485 run sits far inside the no-termination zone at every normal HMI baud rate: 475 ft at 115,200 baud, 1,900 ft at 38,400. Leave a working link alone, keep the Terminate switch off unless the converter sits at a cable end that needs it, and read the switch table before touching the other DIP switches.
No, you do not need one. A 100 ft two-wire shielded RS-485 run, at the baud rates a PLC-to-HMI link actually uses, sits far inside the zone where termination buys you anything. The cable tables used across RS-485 wiring guides put termination past 475 ft at 115,200 baud and past 1,900 ft at 38,400 baud. If the link communicates now, leave the wiring and the DIP switches exactly as they are. Termination solves one problem: reflections off the ends of an electrically long cable. This page says when a run crosses that line, where the two resistors go when it does, and which switches to leave alone.
What the termination resistor does
RS-485 pushes a differential signal down one twisted pair. Give the pair enough length and it stops being two wires and behaves as a transmission line with a characteristic impedance, 120 Ω for the usual 22 AWG data twisted pair. A wave reaching an unterminated end meets an impedance mismatch, and part of the energy reflects back and distorts the following edges. A resistor equal to the characteristic impedance, wired across the pair at that end, absorbs the wave instead of reflecting it. That is the entire job.
Termination is a fix for one geometry: the two physical ends of one continuous run. The value on the resistor is a separate question that only long links ask, and RS-422 termination resistor value for a long HMI cable answers it. Termination also costs current: cable guides for terminated networks quote steady currents of 60 mA or more through the pair, which is why every termination beyond the two ends makes the drivers work harder for nothing.
Where the length thresholds sit
The wiring tables published with RS-485 cable guidance give the no-termination zone directly:
| Data rate | Termination needed beyond | Maximum run with termination |
|---|---|---|
| 115,200 baud | 475 ft | 3,250 ft |
| 57,600 baud | 950 ft | 4,000 ft |
| 38,400 baud | 1,900 ft | 4,000 ft |
The numbers scale the way the physics scales: raise the speed and the safe unterminated length falls. The 4,000 ft ceiling at the slower rates comes from accepting about 66 percent signal loss at the far receiver. Below about 33 ft (10 m), device wiring guides accept plain unshielded twisted pair around 22 AWG; past that, shielded pair is the right cable, which is what a 100 ft run should already be. For links that genuinely run long, connecting a remote HMI screen 1,500 feet from the PLC goes into the converter and wiring choices that distance forces.
Example: 100 feet at 19,200 baud
Take the run in question and assume a common PLC-to-HMI rate of 19,200 baud. That rate is an assumption for the arithmetic, not a reading from your equipment.
- Twisted pair carries a signal at roughly two-thirds the speed of light, about 1.5 ns per foot.
- One-way delay: 100 ft × 1.5 ns/ft = 150 ns.
- Round trip: 2 × 150 ns = 300 ns, call it 0.3 µs.
- One bit at 19,200 baud: 1 ÷ 19,200 s = 52 µs.
- The full round-trip ringing occupies 300 ns of a 52,000 ns bit, about 0.6 percent.
- A UART samples near the middle of the bit, roughly 26 µs after the edge. Even ten round trips of ringing, 3 µs, finish 23 µs before the sample point.
The reflection is over long before the receiver looks at the bit. Even at 115,200 baud, where one bit lasts 8.7 µs, the 0.3 µs round trip is 3.5 percent of a bit, and the tables still put a 100 ft run inside the safe zone. Decision for this run: the termination switch stays off.
Where the resistors go when you do need them
Two resistors, one at each physical end of the continuous pair, and none anywhere else. The isolating converter carries one internally behind a DIP switch marked Terminate; on this type of converter the resistor is in circuit when that switch is down. The communication module on the HMI side has its own built-in one, and wiring to the manual's diagram normally leaves it in circuit. A device spliced into the middle of the run never terminates. When you enable one end, enable the other the same day: the unterminated end still reflects, so do both or neither.
The DIP switches to leave alone
Suppose the panel came up working with two switches marked 1/2 DPX on and everything else off. Keep that state. The documentation for this class of converter names one switch as the termination control, Terminate, down for resistor in. What the 1/2 DPX pair controls is defined in the converter manual's switch table, and the label alone will not tell you. Read that table, match each position to your wiring, and change a switch only when its documented function applies to your installation. One change at a time, then retest.
Shut both ends of the link down before you open terminals or probe continuity. A serial pair inside a plant panel is not the place to work live.
When to flip the switch anyway
Two symptoms point at reflections: a link that runs clean at a low rate and corrupts at a high one, and dropouts that track a cable length after a panel move. Work the checks in order before touching any switch: continuity with power off, terminal assignment against the manual's diagram, the shield, then protocol, rate, character format and node addressing at both ends. A swapped pair, an open conductor or mismatched settings stays broken with any resistor in place. If the checks pass and dropouts remain, put the two end resistors in, one change at a time, and retest at operating speed. For a link that fails outright, RS-485 converter wiring troubleshooting when the link fails walks the wiring, and RS-232 to RS-422 cable distance for an HMI panel link has the distance numbers for the other common serial link.
Recommendation: leave the working 100 ft link alone: Terminate switch off, the two 1/2 DPX switches as found, no external resistors. Flip Terminate down only if the converter sits at a physical end of the pair and the link drops data at speed with wiring and settings verified first, and put the matching resistor in at the far end the same day. Treat both ends as routine once a run passes 475 ft at 115,200 baud or 1,900 ft at 38,400 baud.
Questions
Should I add an external 120 Ω resistor at the converter for margin?
No. If the built-in resistor is already switched in, an external 120 Ω across the same pair runs in parallel with it: 120 × 120 ÷ (120 + 120) = 60 Ω. That end then loads the drivers at half the cable's impedance, on top of the 60 mA or more a properly terminated network already draws. One resistor per end, the built-in one, is the whole design.
Where do the resistors go when several devices share one RS-485 pair?
One at each physical end of the continuous run, none anywhere else. A device spliced into the middle never terminates. If you extend the bus later and a former end device lands mid-run, switch its resistor out the same day; leaving it in loads every driver on the pair. The new far end takes it instead.
How far can a terminated RS-485 link go?
About 4,000 ft, 1,200 m, at the slower rates, a figure derived from accepting roughly 66 percent signal loss at the far receiver; at 115,200 baud the tables give 3,250 ft. Past those distances the fix is a repeater or a fiber link, never a bigger resistor. At 100 ft you are nowhere near any of this.
Planning information, not legal or engineering advice. Rules differ by place: check codes and permits with your local authority and your licensed trades.