Native Ethernet on the HMI and PLC with a wireless bridge is the link to build when both devices can take Ethernet. A serial-to-Ethernet converter pair in tunnel mode is the fallback for fixed serial ports. Check any copper run against the 4,000 ft RS-485 planning limit at 100 kbps, prove the link on a short bench cable, then span the yard.
Build the wireless hop with Ethernet at both ends when the HMI panel and the PLC can both take Ethernet. Fall back to a serial-to-Ethernet converter pair in tunnel mode only when a serial port is fixed at one end. That decision rarely changes once the numbers are on the table. The case used here is the classic yard install: the panel sits in the house, the PLC sits 140 ft away at a turbine with a 105 ft mast, and you would prefer not to run a data cable past it. Below is the side-by-side, the copper distance math behind the radio choice, and the bench order that keeps commissioning day short.
What each box actually carries
A serial-to-Ethernet converter, usually sold as a device server, terminates a serial port on one side and an IP network on the other, and moves bytes between them inside TCP or UDP packets. It carries bytes. It does not translate protocols. A wireless Ethernet bridge joins two wired Ethernet segments across a radio link and passes whole frames, so each side sees the other as ordinary LAN.
That difference sets the design:
- With native Ethernet at both ends, a bridge gives the panel and the controller a normal LAN drop each. The panel runs its own Ethernet driver straight to the PLC, and no protocol conversion exists anywhere in the path.
- With a serial-only end, a pair of device servers in tunnel mode can carry the serial traffic, but each serial port must be set to match the panel driver exactly, and both servers must pass the data transparently.
No converter makes two incompatible protocols talk. If the panel driver and the PLC port do not speak the same protocol, fix the driver selection first; checking the panel driver list is the step that catches a mismatch before any cable is cut.
Side-by-side comparison
| Wireless bridge, Ethernet at both ends | Serial-to-Ethernet tunnel pair | Direct RS-485 copper | |
|---|---|---|---|
| What it carries | Ethernet frames over a radio hop | Serial bytes inside TCP or UDP packets | The serial signal itself on twisted pair |
| Hardware | Ethernet on the panel and on the PLC port or module, plus the bridge pair | Two device servers, one per serial port | One twisted-pair cable, no converters |
| What must work | Both endpoints in one IP subnet, bridge associated to the network, panel Ethernet driver selected | Both serial ports matched to the driver, transparent tunnel mode on both units, IP path up | Interface type and serial settings matched at both ends |
| Typical failure | Endpoints on different subnets, or the bridge never associates | Bytes pass on the bench but the driver fails through the tunnel mode | Wiring, termination or grounding fault on the run |
| Reach | Wherever the radios associate | Wherever the IP network reaches | 4,000 ft at 100 kbps on 22 AWG twisted pair |
| Pick it when | Both devices already speak Ethernet | One end is fixed to serial and copper is excluded | Copper is allowed; simplest and cheapest |
Where each path wins
The bridge wins when both ends already speak Ethernet. One radio hop, one subnet, and the panel's native driver, with no converter to blame when something misbehaves. The work is addressing and association: both endpoints in the same subnet, the bridge configured to join the house network, then a ping across before the panel project points at the controller's address. At a 140 ft hop across the yard, mount radios rated for weather exposure at each building and keep the Ethernet runs indoors; indoor consumer gear through two walls at that range drops out. A first attempt that times out has its own checklist in fixing an HMI PLC Ethernet timeout on first setup.
The tunnel pair wins when a serial port is fixed and copper is excluded. Converters already on the shelf are the cheapest route, and the serial wiring stays short at each end. The cost is configuration: two serial ports matched to the driver, the documented tunnel mode on both units, and a path kept dedicated while testing. A pair that passes bytes back to back on the bench is not proof the panel driver talks through that mode, so the final bench step is the real driver through the real pair.
Copper wins whenever it is allowed. RS-485 needs no converters, no addressing plan and no radio, and its reach at panel polling speeds is far beyond a yard run; the example below puts numbers on that. When a copper link misbehaves later, the faults are physical and findable: the RS-485 converter wiring checks cover the usual suspects.
One safety note on the radio hop: going wireless keeps the data cable out of the yard, and the power conductors and grounding still cross it. Bond and surge-protect what remains, per the equipment installation instructions and NFPA 70 (the National Electrical Code) where the job is in the US. If a copper run ever does cross the yard, put it in grounded metal conduit and support it properly at bends; a conduit sweep needs support at both ends of the bend.
Example: the copper option for a 140 ft yard run
Inputs. The run is 140 ft, house wall to PLC cabinet at the turbine, the distance given for the job. The planning reach is 4,000 ft (1,200 m) at 100 kbps, the figure published in RS-485 design guides for 22 AWG 120 ohm twisted pair. The port speed is 19.2 kbps, an illustrative setting for a panel driver; the driver's manual prints the real value.
The rule. RS-485 (the TIA-485 standard) carries a balanced differential signal over twisted pair, and it fits the run when the length stays inside the reach published for the bit rate in use. The 4,000 ft figure belongs to about 100 kbps; reach shrinks as the rate rises, and RS-485 signaling is specified up to 10 Mbit/s at short range.
Arithmetic.
- Margin: 4,000 ft minus 140 ft = 3,860 ft of spare reach.
- Share of the limit: 140 / 4,000 = 0.035, so the run uses 3.5% of the published reach.
- Rate headroom: 19.2 kbps / 100 kbps = 0.192. The port runs at 19.2% of the rate the 4,000 ft figure assumes, and slower rates allow more cable, so 4,000 ft is a floor for this run.
Result. The 140 ft run sits at 3.5% of the planning reach with the port speed well inside the envelope. Distance is a non-issue in this yard.
Decision. Under 4,000 ft and at or below 100 kbps, direct copper meets the published envelope. The wireless choice in this yard is about keeping conductors away from the mast, and about nothing else. Accept copper and direct RS-485 is the simplest correct link; exclude copper and the choice between the two wireless paths is made on hardware capability alone. Where distance is the open question, connecting a remote HMI screen 1,500 feet from the PLC works the longer case, and serial distance limits by interface lists the reach of each interface.
Bench-test the path before spanning the yard
- Record both ends before touching hardware: the driver selected in the panel software and its port settings, meaning interface, baud rate, data bits, parity, stop bits and node address.
- Wire the panel to the PLC over a 6 ft bench cable, using the documented wiring for the interface.
- Load a screen that reads one PLC value, then confirm the displayed value tracks a known change.
- Poll for several minutes and check the panel's communication diagnostics and the PLC port indicators; both stay clear.
- Add the network hardware on the bench. For a tunnel pair, connect both servers to a small switch, set both serial ports to the recorded values, and repeat steps 3 and 4 with the real driver talking through the pair.
- Move the bridges or radios into the path, keep both endpoints in one IP subnet, and run the same read test across the yard before anything is mounted.
Stay read-only through all of it: the link passes reads before any panel write reaches the process, and one change at a time keeps the cause findable. If the house end also needs a notice board playing its own content between shifts, that is a separate screen on its own link; the shop-floor displays we sell cover that job, and it stays out of the control path.
The decision, case by case
- Both ends offer Ethernet: build the bridge. One subnet, the panel's native Ethernet driver, the bench read test, then the yard. That is the answer for the yard case in this article, where both endpoints can take Ethernet.
- One end is locked to serial: build the tunnel pair. Match both serial ports to the driver, run the documented tunnel mode, keep the tunnel dedicated to the HMI path while testing, and test PLC programming through it only after the HMI link is stable.
- Copper is acceptable: run direct RS-485. The example shows why: 140 ft sits at 3.5% of the published reach at panel speeds, and every later fault is physical and findable.
- The serial end will not change and no radio is wanted: replace the weak endpoint, an Ethernet module in the PLC or a panel with built-in Ethernet. That removes the tunnel entirely and is the permanent fix once the budget allows.
Recommendation: when the panel and the PLC both offer Ethernet, build the wireless bridge path: one subnet, the panel's native driver, and the bench read test before the radios go in. Choose the serial-to-Ethernet tunnel pair only when a serial port is fixed at one end or the converters are already on the shelf, and bench the pair back to back with the real driver before mounting anything. When copper is allowed, direct RS-485 is the simplest correct link at any yard-scale distance.
Questions
Can the same tunnel carry PLC programming traffic later?
Treat it as a separate test, never part of the first commissioning. Programming tools are built for a direct port and often time out through a device server tunnel, and the PLC port may accept only one protocol at a time. Get the HMI link stable, then try the programming connection through the tunnel, and keep a direct serial cable as the fallback for uploads.
Does the 4,000 ft RS-485 figure apply at any data rate?
It is the low-speed case. The 4,000 ft (1,200 m) planning figure assumes roughly 100 kbps on 22 AWG twisted pair. RS-485 signaling itself can reach 10 Mbit/s, and usable cable length falls as the bit rate rises. HMI polling traffic runs far below 100 kbps, so for panel links the 4,000 ft number is the planning value to use.
Does going wireless remove the lightning risk at the turbine?
It removes the data cable, and nothing else. The mast still needs its own lightning protection designed to the applicable code, and the power conductors and grounding between the buildings remain exposed. Surge-protect and bond what still crosses the yard, and inspect that protection after any strike in the area.
Planning information, not legal or engineering advice. Rules differ by place: check codes and permits with your local authority and your licensed trades.