An RS-485 segment carries one polling master. Set the HMI panel as that master with a unique station address per servo drive, matching baud and serial format, and terminate the pair with 120 ohms at each end. The PLC gets its data on a supported listening path or its own port and network; a second talker on the same pair reproduces the errors whatever tag you change.
No, not as two talkers on one pair. An RS-485 segment carries exactly one device that initiates exchanges. Give that role to the HMI panel, set each of the four servo drives to its own station number over a shared baud rate and serial format, and the drives answer one at a time. The PLC gets its drive data through a receive-only listening role where its documentation supports one, or from a separate link of its own. When a second controller transmits on the same pair, its frames land inside the panel's exchanges, both ends discard the corrupted frames, and no tag edit clears the fault codes, because the collision happens on the wire before any address is compared. This page settles who transmits, sets the addresses, wires and terminates the segment, and brings the bus back one drive at a time.
One pair, one master
RS-485, standardized as TIA/EIA-485-A, defines the electrical side of the bus: a differential pair, many receivers on one segment, and drivers that take turns transmitting. The standard says nothing about who takes which turn. Master/slave protocols supply that rule: one master issues each request, the addressed slave answers, and every other device stays silent until polled.
An HMI panel is almost always a master on a serial link. It polls, the drives respond; that is what its drive drivers are built to do. A PLC serial port that runs its own polling schedule against the same drives is a second master. Two uncoordinated transmitters on one pair garble each other's frames: the panel's driver logs failed exchanges, and the drives discard anything that fails its checksum. The fault survives every tag edit because addressing never gets a chance to matter.
The general case of two talkers on one pair is worked through in two Modbus RTU masters on one RS-485 bus: why and the fix. One serial port carrying two talkers fails the same way whatever the devices are; why two HMI panels fail on one PLC serial port walks that variant.
What each symptom points to
Match what the panel and the drives show to one cause, then test that cause directly:
- Fault codes such as E008 and E009 that no tag change clears. These are communication faults on the panel's side of the link. With two talkers on the pair, frames corrupt before any address is checked, so every address fails the same way. Test: take the PLC off the drive pair. Codes that stop at that moment point at a second master.
- Faults begin the moment the PLC joins the pair. Panel and drives were talking; the PLC's cable goes on and the errors return. The PLC port is transmitting its own polling traffic. The only quiet exception is a port the program never writes to; one that retries after a failed exchange is another talker.
- One drive never answers while its neighbors do. That drive's station number or serial format disagrees with the panel's driver entry for it. Re-read the drive's serial communication parameters.
- Errors climb when the last drive drops off the chain or when the run gets long. Missing or misplaced termination at the segment's ends. The wiring section below sets it.
- Errors track vibration or worsen near a drive's motor cabling. Noise picked up on the pair is its own diagnosis; fix a PLC communication timeout on an HMI servo link covers routing and shielding.
Decide who masters the bus
Pick one transmitter for the pair before editing anything else. In a panel-plus-drives setup the panel is the usual master: its drivers speak the drives' protocol, the drives are slaves by design, and the panel is the screen people watch. The PLC keeps that job only when it genuinely sequences the axes; then the roles flip and the panel reads through the PLC.
| Arrangement | Who transmits on the drive pair | Where the second device gets drive data |
|---|---|---|
| Panel masters the drives (usual choice) | The panel only | The PLC holds its port in receive-only mode where its manual documents that role; otherwise it takes no drive data from this pair |
| PLC masters the drives | The PLC only | The panel polls the PLC over its own link and displays values held in PLC memory |
| Both must command the drives | Neither, on one pair | Move one device to its own serial port or network; one set of slaves answers one master |
Watching values that live on a link you do not control is a standard pattern; it is the same one used to connect SCADA to a PLC when an HMI panel holds the port. A receive-only role exists only where the PLC's own documentation gives it, and wiring a receiver onto the pair does not create the role, so read the serial-port documentation before counting on it.
Set station numbers and serial format
Each drive carries three settings in its serial communication parameter group, and the panel's driver setup repeats the same three per device: the station number, the baud rate and the format (data bits, parity, stop bits). Modbus-style station numbers run from 1 to 247, with 0 reserved for broadcast, so four drives take 1, 2, 3 and 4. The only rule is that no two devices share a number, and the drive count sets nothing: you choose the numbers and enter them in both places. Baud and format must match end to end; the example below uses one common working point. Terminal labels differ between devices (A/B on some, D+/D- on others), so match the names in each manual and never infer polarity from a wire color.
Example: four drives at stations 1 to 4, 19200 baud, 8E1
Scenario: one HMI panel polls four servo drives at stations 1, 2, 3 and 4 over a single pair at 19200 baud, 8 data bits, even parity, 1 stop bit. That format carries 11 bits per byte. Each exchange reads four holding registers, which takes an 8-byte request and a 13-byte reply: 21 bytes, so 21 × 11 = 231 bits, and 231 ÷ 19200 baud = 12.0 ms of wire time. Modbus RTU adds a silent gap of 3.5 characters between frames: 3.5 × 11 = 38.5 bits, about 2.0 ms, before and after each exchange. Allow 5 ms for the drive to process the request, and one exchange takes about 12.0 + 2.0 + 2.0 + 5.0 = 21 ms.
Polling all four drives in turn takes 4 × 21 = 84 ms, so each drive gets a fresh read about 12 times a second. The ownership of the wire is the point: the master alone uses the 12 ms of frame time per exchange, the quiet windows run about 9 ms, and nothing coordinates a second device with those windows. A second master fires when its own schedule says, its frames land in the panel's exchanges at random, and both sides discard the wreckage. That is the E008 and E009 pattern rebuilt from timing numbers.
Need faster refresh? At 38400 baud the frame time halves to 6.0 ms and a full sweep drops to 4 × (6.0 + 1.0 + 1.0 + 5.0) = 52 ms, about 19 refreshes per drive per second. One master, four station numbers, and the refresh rate is yours to set with the baud.
Terminate and wire the segment
Termination is a 120 ohm resistor across the pair at the two extreme ends of the cable: the panel's connector at one end and the last drive at the other. The value comes from the twisted pair's characteristic impedance, about 120 ohms on typical bus cable, so the resistor absorbs the signal instead of letting it reflect back and smear the edges. Resistors at every device do the opposite: they overload the drivers and flatten the signal, so terminate at the ends only.
The device budget is generous here. TIA/EIA-485-A budgets 32 unit loads per segment, counting every receiver, and a passive listener on the pair takes a load too. Modern transceivers often draw a fraction of a load, which is why some datasheets allow more than 32 devices. Four drives, one master and one listener sit well inside any of those budgets, and the standard's guideline of about 1200 m at 100 kbps covers any drive daisy chain on one floor at drive-bus speeds.
Make these wiring changes with the drives and the panel powered down. Check that each drive's communication-loss protection is enabled and timed: when the pair breaks or the frame train stops, the drive should trip on its own supervision timer instead of holding its last command indefinitely.
Bring the bus back one drive at a time
- Panel plus drive 1. Take the PLC off the pair. Connect the panel and drive 1 only, set station 1 in the drive's parameters and station 1 in the panel's driver, with matching baud and format.
- Poll one live value. Read a register that moves with the axis, such as position, and check the number tracks the axis instead of freezing.
- Add drives 2, 3 and 4 one at a time. Wire each onto the pair at its station number and poll it before adding the next. A drive that fails alone while its neighbors talk has its own number or format to fix.
- Reconnect the PLC for the role you chose. For a documented receive-only listener, watch the panel's fault display while it sits on the pair, at idle and while polling. Faults that return at that moment mean the PLC is transmitting, and it moves to its own port or network.
- Check all four stations together. Run the panel's polling loop, read a moving value on every drive, and check each number appears under the drive that produced it.
End state: four stations answering, values landing under the right drives, and one device transmitting on the pair.
The decision
Recommendation: make the HMI panel the one master of the RS-485 segment. Set the four servo drives to their own station numbers at one shared baud and serial format, terminate the pair with 120 ohms at the two far ends, and bring the bus up one drive at a time until all four answer. Keep the PLC on the pair only where its documentation supports a receive-only role; if it must ever transmit, give the PLC the master's job and let the panel read through it, or move the two devices onto separate links. Stop editing tags while two talkers share the pair: the fault codes regenerate whatever the address table says.
Questions
Can the PLC listen on the same pair while the HMI panel is the master?
Only where the PLC's serial port documentation supports a receive-only role, and only while its program never writes to that port. A port that retransmits after a failed exchange is still a second master. A listener also draws at least one unit load, so count it in the segment's device budget. Where the documentation offers no receive-only mode, put the PLC on its own port or network.
What do fault codes E008 and E009 mean here?
They are the panel's communication-fault codes, and each panel manual lists its own definitions. In the two-master pattern they share one cause: frames corrupted on the wire before addressing is ever compared. Clear the master conflict first and read the codes again. A code that survives a single-master bus points to the next cause through its definition in the panel manual.
How many servo drives can one HMI panel poll on RS-485?
Electrically, more than a line of drives needs: TIA/EIA-485-A budgets 32 unit loads per segment and modern transceivers often draw a fraction of one. The working ceilings sit elsewhere, in the panel driver's device list and in the refresh rate you need. At the example's 21 ms per exchange, every added drive adds about 21 ms to the sweep, so size the bus around how fresh each axis's numbers must be.
Can both devices share the pair if they take turns on a schedule?
Not under a standard master/slave drive protocol. Modbus RTU and its relatives define timing only between one master and its slaves; two masters have no shared schedule to keep, each polls when its own program says, and the gaps close as scans stretch or a drive answers slowly. Overlapping frames are discarded on both sides. Use one master on the pair, or separate links.
Planning information, not legal or engineering advice. Rules differ by place: check codes and permits with your local authority and your licensed trades.