CANopen Displacement Sensor: EMCY, Node Start, PDO and SYNC Communication Examples

CANopen magnetostrictive displacement sensors report position over the bus, but the first day on site usually fails on one thing: which telegram does what. This article walks through the four frames you actually use on a Germanjet Series 19 CANopen displacement sensor — EMCY, node start, PDO and SYNC — and the order to send them in so the master and the sensor agree.

The frame sequences below come from the Germanjet Series 19 CANopen model (194). Within the same brand and series they transfer directly; across brands, trust the vendor's EDS and manual.

The four telegrams and their jobs

• EMCY (Emergency) — the sensor pushes an error frame when something is wrong inside (over-voltage, memory error, position fault). It is not a request; you read it to diagnose.
• NMT node start — the master puts the node into Operational state. Nothing is sent or received until the node is started.
• PDO (Process Data Object) — the actual position value. TPDO carries the position from sensor to master; RPDO writes to the sensor.
• SYNC — a bus-wide heartbeat. When SYNC is enabled, the sensor publishes its TPDO on every SYNC, which synchronises several sensors on the same bus.

Default COB-IDs and the frames you actually see

On power-up the sensor immediately sends one EMCY frame with COB-ID 0xFF — the "I am here" frame, which also reveals its default node address 0x7F, 127 in decimal. The master then sends the node start frame: COB-ID 0x000, two data bytes 01 00, where 01 means start and 00 means all nodes. Sending 01 7F starts node 0x7F alone.

Once started, the sensor answers with PDO1, COB-ID 0x1FF, seven data bytes: four bytes of position, two bytes of velocity and one reserved byte (00). Both values are sent LSB first — receiving 03 83 00 00 means the position is 0x00008303, or 33539 counts; at the 0.005 mm resolution that is 33539 × 0.005 = 167.695 mm.

This is where synchronous and asynchronous operation part company. In the synchronous connection the sensor answers every SYNC frame, COB-ID 0x080 (no data), with one PDO. The master sends node start once at initialisation and is driven by SYNC afterwards, so you decide how many frames are read per cycle and the rhythm is fully under control. In the asynchronous connection the sensor enters Operational state after node start and then publishes its PDO on its own timer — no SYNC at all. How fast it publishes is set by the CAN event timer, which is covered in a separate article.

With two magnet rings the principle is the same, just one frame more. PDO1 stays at 0x1FF for the first ring and PDO2 uses 0x2FF for the second; each frame keeps the 4-byte position + 2-byte velocity + 1 reserved byte layout. Under SYNC one SYNC brings out both frames; asynchronously each ring's PDO follows its own event timer. Two positions from one sensor, carried in two separate frames.

What the EMCY frame actually reports

The EMCY sent at power-up is empty; it only says "I am here". Real alarms follow whenever the internal error status changes — on error entry and again on error exit — always 8 bytes with a fixed layout.

• Bytes 0–1: standard error code.
• Byte 2: error register, the same value as object 0x1001 — 0x00 clear, 0x11 communication, 0x05 main voltage, 0x81 transducer, combined faults bit-wise.
• Bytes 3–7: manufacturer-specific.

Code	Meaning	Usual cause on site
00 00	No error	Normal frame after reset
50 00	Device hardware (magnet not detected)	Magnet dropped off, outside active stroke, or never fitted
81 00	Communication error	Bus noise, missing termination
31 00	Main voltage error	Supply outside the permitted window
81 30	Life guard error	Node guarding enabled, master not polling
63 00	Data set error	Parameter area failed its check, needs reconfiguration
81 20	Node went CAN passive	Transmit error counter high, poor bus quality

The two seen most often are 50 00 and 81 30. The first is rarely an electronics failure — it is the magnet position. The second almost always means guarding was enabled and the master never sent the remote frame, so the device concludes the link is dead and stops sending PDOs; on site it looks like "the reading disappeared after a few seconds".

Magnets, COB-IDs and the error LEDs

Two magnets simply means one more frame. The device supports up to four magnets, one TPDO each, with the COB-ID stepping in 0x100.

Magnet	COB-ID at default node 0x7F	General rule (node = N)	Length
Magnet 1 (TPDO1)	0x1FF	0x180 + N	7 bytes
Magnet 2 (TPDO2)	0x2FF	0x280 + N	7 bytes
Magnet 3 (TPDO3)	0x3FF	0x380 + N	7 bytes
Magnet 4 (TPDO4)	0x4FF	0x480 + N	7 bytes

Every frame has the same layout: four bytes of position, two of velocity, one of limit status (bits 0–3 map to limit switches 0–3). In synchronous mode one SYNC brings all of them out in magnet order; in asynchronous mode each frame runs on its own event timer. Multi-magnet and the 8-byte limit covers the trade-offs when allocating PDOs.

Position is a 32-bit signed value. With the zero point mid-stroke, or the magnet travelling negative, the high bits are all ones and an unsigned read returns four billion-odd — that is a sign handling bug, not a failed sensor. Do not hard-code the step either: 0x6501 gives the step actually in use, in units of 0.001 µm.

The two LEDs on the electronics head answer "how far did it get" without any tool:

Green	Red	Meaning
Off	Off	No supply
Off	On	Start-up failure
On	Off	Normal operation (magnet within active stroke, supply healthy)
On	On	Magnet not detected or lost
Flashing	Flashing	Supply near its limits; still running but needs correcting

The red LED is paired with a 50 00 EMCY on the bus; when both agree, the problem is at the magnet, and there is no point digging through telegrams.

Four details that are easy to miss on site

• The default node address is 0x7F. If several sensors share one bus, give each of them its own address first — do not leave them all answering on 0x7F.
• Node start is sent once, at initialisation. In synchronous mode the rest is driven by SYNC only; repeating node start every cycle is harmless but pointless.
• Confirm the resolution before converting counts to millimetres. The 0.005 mm used above is an example — the order code decides it, and the same raw value means a different position at a different resolution.
• The PDO already carries velocity as well as position. If the controller needs speed feedback it is in the same frame; no extra channel is required.

When one frame does not add up, work back from the COB-ID of the EMCY and PDO frames to the node address — that check usually finds it straight away.

Start-up order that works

1. Power the sensor and let it boot (roughly 0.5 s).
2. Send NMT start to the node — after this the sensor is online and answers SDO.
3. Read the object dictionary over SDO to confirm node ID and baud match the master.
4. Enable the TPDO you need (position, and optionally velocity) and map it to the correct COB-ID.
5. If you use SYNC, enable the SYNC consumer and check the period on the master side.
6. Watch the EMCY telegram for the first few minutes — most wiring mistakes surface there before the PLC logic sees them.

Full parameter changes — baud rate, node address and the CAN event timer — are covered step by step in changing CANopen parameters over SDO. If several sensors must share one bus node, see seven electronic rulers on one CANbus node.

Common first-run mistakes

• Node never started → sensor is silent even though the wiring is correct. Always send node start.
• COB-ID mismatch → position never arrives. Check the PDO mapping and the node ID on both sides.
• SYNC period shorter than the sensor cycle → duplicate or skipped frames. Keep the SYNC period longer than one measurement cycle.
• EMCY flooding → usually supply or shield, not the sensor. Fix the 24 V and the screen before touching the mapping.

Output choice is a selection-stage decision; see how to fix the output at selection if you are still deciding between CANopen, SSI and 4-20 mA.