How Is the CNC Dynamic Card Implemented in Bus Mode?
A step-by-step answer for controls and maintenance engineers. We cover card addressing, cycle timing, cabling, grounding and the first-move checks that catch a bad bus before the first part is cut.
In this article
- 1
- 2
- 3
- 4
- 5
- 6
- 7
Key takeaways
What the dynamic card does in CNC dynamic card bus mode
A dynamic card is the interface board that sits between the CNC controller and the machine hardware: sensors, relays, drives and safety contacts. In point-to-point wiring every one of those signals needs its own conductor back to the cabinet. On a machine with 200 I/O points that is 200 wires, and every added option means another terminal block.
Bus mode changes the topology. One shared channel carries frames for every node. The dynamic card becomes a bus node with its own address, and the controller polls it the same way it polls a servo drive or a remote I/O slice. Fieldbus, CANopen, EtherCAT and Modbus RTU all work this way; the difference is frame format and cycle discipline, not the wiring idea.
The practical gain is not just fewer cables. A bus node reports diagnostics the controller can read: short circuit on an output group, open circuit on a sensor input, undervoltage on the card supply. With point-to-point wiring those faults reach you as a stopped machine and a multimeter.
The trade is that the bus is now shared. One damaged node can disturb traffic for every other node on the segment. That is why the setup steps below put addressing, termination and grounding before anything else.
Cycle time budget: where the milliseconds go
Bus cycle time is not the same as servo update rate, and mixing the two up is the most common reason a retrofit feels sluggish. The controller sends one process-data frame per cycle. Inside that frame each node gets its slice of inputs and outputs. If the frame is 2 ms long, a proximity sensor that trips at the wrong moment is only seen on the next frame.
Work the budget in three parts. Transmission time is frame length divided by bit rate: a 64-byte frame at 10 Mbit/s is roughly 51 µs plus overhead, while the same frame at 500 kbit/s costs about 1 ms. Card scan time is the local loop on the dynamic card, usually 100–500 µs. Servo or drive update is set by the drive, not the card.
For most machine-tool loops, keep the total bus latency under 4 ms. A tool-change sequence or a pallet clamp tolerates 10 ms without anyone noticing. A high-speed probing move on a machining center does not: the probe touch must land in the same cycle as the axis latch, or the measured point drifts.
If latency is too high, raise the bit rate before you shorten the frame. Doubling the rate from 500 kbit/s to 1 Mbit/s usually halves transmission time and keeps the I/O map intact. Reducing the frame means re-mapping nodes, which costs commissioning time and invites mistakes.
Cable, termination and grounding rules that hold up
Use shielded twisted pair for the bus, with the pair carrying the differential signal and the shield tied to PE at one point only, normally the cabinet entry. Tying the shield at both ends builds a loop that carries the drive return current straight through your data pair. The symptom is not a dead bus; it is random node drops during spindle acceleration.
Terminate the segment with 120 Ω resistors at both physical ends, not at the controller and not at the middle node. A star topology with three branches and three terminations will work on the bench and fail on the machine. Keep drop cables under 0.3 m if the standard allows drops at all; most fieldbuses prefer daisy-chain.
Separate the bus cable from motor and spindle power by at least 100 mm, and cross power cables at 90° where they must cross. On a retrofit, run the bus in its own trunking rather than reusing the old I/O duct. In our own builds we keep the bus trunk on the opposite side of the cabinet from the servo drives and their braking resistors.
Node power deserves its own check. Most dynamic cards accept 24 V DC ±10 %. Measure at the card terminal, not at the power supply, with all outputs energised. A 0.5 V drop along the trunk turns into a 10 % margin loss and shows up as cards that reset when a solenoid fires.
First-move checks before you cut metal
After the bus scans clean, do not jump straight to a program. Command each output group individually from the controller and watch the physical contactor. Then trip each input by hand and confirm the bit toggles in the I/O monitor within one cycle. This is boring work and it finds swapped addresses in ten minutes instead of two hours.
Check the safety path separately. The dynamic card may carry the E-stop readback or the guard interlock, but the safety function must not depend on bus traffic alone. A hardwired series loop into the drive enable is still the normal arrangement, with the bus reporting status rather than enforcing the stop.
Log the bus error counters before and after a full warm-up. A node that accumulates CRC or lost-frame counts at idle will fail under load. If counters climb while the machine sits still, the problem is usually termination or a shield loop, not firmware.
Finally, run the heaviest spindle and axis move you can while watching cycle time on the controller. Latency that holds at 2 ms on a stationary machine can double once the drives are working. That number is your baseline for the next service visit.
Six steps to implement the dynamic card in bus mode
Work in this order. Each step has a pass condition; do not move on until it is met.
- 11. Map every signal and assign addressesList inputs, outputs and analogue points per card. Assign a unique node address to each card, normally 1–63 on a rotary or DIP switch. Rule: no two cards share an address, and leave the highest addresses free for future nodes.
- 22. Set baud rate and card parameters offlineSet bit rate, frame length and watchdog time with the card unpowered or on a bench supply. Start at the lowest rate your cycle budget allows, commonly 500 kbit/s to 1 Mbit/s, and set the watchdog to 3× the bus cycle.
- 33. Wire the trunk, then terminateDaisy-chain the shielded twisted pair, keep drops under 0.3 m, fit 120 Ω at both physical ends only. Keep 100 mm clearance from spindle and servo power, and cross at 90° when unavoidable.
- 44. Power up and scan the busApply 24 V DC ±10 % to each node, measure at the card terminal. Run a bus scan. Pass condition: every configured address appears, no duplicate and no unknown node.
- 55. Force I/O and time the cycleToggle each output group and read back each input. Measure the total cycle with the controller's scope or trace tool. Pass condition: under 4 ms for motion loops, correct bit within one cycle.
- 66. Run the load test and record countersRun the heaviest spindle and axis move available. Recheck cycle time and bus error counters. Pass condition: no lost frames, latency within budget. Save the log as the machine baseline.
Bus mode symptoms, causes and fixes
Three columns: what you see, what usually causes it, what to change.
| Symptom | Likely cause | Action |
|---|---|---|
| No node found on scan | Address 0 or duplicate address | Set unique address 1–63, rescan |
| Node drops during spindle ramp | Shield tied at both ends | Ground shield at cabinet entry only |
| Random CRC errors at idle | Missing or doubled termination | Fit 120 Ω at both physical ends |
| Card resets when solenoid fires | 24 V sag below tolerance | Measure at card, upsize trunk, separate supply |
| Motion feels late or jerky | Bus cycle above 4 ms | Raise bit rate before shortening frame |
| One bad node stops the segment | Node holding the bus dominant | Isolate the node, check its transceiver |
Questions engineers ask before a retrofit
Can I keep the existing point-to-point I/O and add a bus card?
Yes, and many retrofits do exactly that. Run the new dynamic card on the bus while the old terminal blocks stay on hardwired I/O. The controller sees both. The risk is documentation: two I/O philosophies in one cabinet confuse the next technician, so label the bus trunk clearly and keep one master signal list.
Move signals to the bus in groups that share a function, such as all coolant and chip-conveyor I/O. That way a group can be rolled back together if something misbehaves.
How many nodes can one segment carry?
It depends on the fieldbus and the frame length, not on the dynamic card itself. Common limits are 32 or 63 nodes per segment, and the practical ceiling is usually cycle time rather than node count. Every added node gets a slice of the same frame.
If you need more points, add a second segment on another controller port rather than stretching one trunk across the whole machine.
Does bus mode change the achievable machining tolerance?
No. Tolerance comes from the machine geometry, spindle, thermal behaviour and the servo loop. Bus latency affects when a signal is seen, not how accurately an axis holds position on a commanded path.
Where it does matter is probing and in-process gauging, because those depend on latching an axis position at the instant an input changes. For that case, keep the bus cycle well under the probing feed's response time.
What watchdog setting should I use?
Set the node watchdog to about three times the nominal bus cycle. At a 2 ms cycle that is 6 ms. Short enough to stop the machine quickly, long enough to ride out a single lost frame without a spurious trip.
If you see watchdog trips with clean error counters, the cycle time is probably being exceeded under load. Log it before you lengthen the watchdog.
Is the bus cable the same as the encoder cable?
No. Encoder cables are specified for continuous flexing and carry a different signal set. Bus cable is specified for impedance and shield coverage. Using encoder cable for a bus trunk often works on the bench and then fails once the drag chain moves.
Buy cable rated for the fieldbus you are running, and check the impedance and capacitance per metre against the standard.
Where does the dynamic card fit if the drive is already on the bus?
It becomes another node on the same segment, with its own address and its own process-data slice. Order of nodes in the frame is set in the controller configuration, so document that order.
Keep the drive and the card in the same cycle if the card reads a signal that the drive acts on. Splitting them across two cycles adds latency you cannot remove later.
Send us the I/O list and we will quote the card bracket
Upload your drawing and signal schedule. We reply with a quotation and a free DFM analysis within 12 hours, and parts ship in 3–5 days.
12-hour quote100% inspectionFrom one prototype to 10,000+ parts
More from GreatLight
We publish setup notes, tooling trials and inspection data from the factory floor.
