Lost Steps: Preparation, Placement and Process Control
A machine tool cutting an irregular arc, a depth of cut that varies from pass to pass, and edges that look rough rather than machined: the symptoms point at the motion control system, and the first response is usually to replace the driver or the encoder cable. When neither changes anything, the fault is often not in the drive at all but in the electrical environment around it, and the lost steps are a symptom of interference rather than of a failing component.
That distinction matters because the two explanations lead to different work. A genuine drive fault is repaired by replacing hardware. Interference induced into the pulse train is repaired by changing the coupling path, and it will recur on the next machine unless the layout and the filtering are addressed. The case described here is representative of a large class of small machine tools where several compact drives share a control cabinet.
Symptom and First Assumptions
The reported behaviour was a stepper axis that followed an irregular path, a second axis whose depth varied, and a finished surface that was visibly rough. The driver was replaced, then the encoder cable, with no change. Measurements at the drive showed nothing abnormal in the supply, and the pulse train looked correct when observed on an oscilloscope with the machine idle.
The investigation moved to the cabinet and found several compact variable frequency drives mounted side by side on the roof of the enclosure, driving the spindle motor and the extraction fan. Two coupling paths were present at the same time: a conducted path through the supply wiring, and a radiated path through the air inside the cabinet. The pulse train cable for the stepper axis ran in the same duct as the spindle drive output for more than two metres, which provided a direct capacitive coupling path into the signal that the controller counts.
<img src="https://www.gopcba.com/wp-content/uploads/2026/01/impendance.png" alt="EMC filter mounted at the input of a compact variable frequency drive” />
Finding the Coupling Paths
The mechanism behind the coupling starts inside the drive. Each switching transition of the output stage produces a voltage change of several kilovolts per microsecond, and the pulse width modulated output travelling along the motor cable is reflected at the motor end, producing an overvoltage at the terminals. The common mode current that results flows through the parasitic capacitance between the motor winding and its frame to earth, and then back through the earth network to the drive.
That return current is the problem for the control system. It shares the earth conductors with the control electronics, and the voltage it develops across the impedance of those conductors appears in series with every signal referenced to them. Radiated coupling adds a second contribution: the motor cable and the drive enclosure emit a field that the pulse train cable picks up along its length, and several drives mounted side by side increase the field without changing any single drive’s performance.
Countermeasures That Worked
The first measure was an EMC filter at the input of each drive. Its current rating was chosen from the drive input current with about thirty percent margin, and it was mounted on the incoming side of the drive so that the noise generated inside the drive is returned to its source rather than conducted along the supply. The filter housing was bonded flat to the mounting plate, with no rubber or plastic washer under it, because the enclosure is part of the filter’s return path and an insulating washer defeats it.
The earth connection of the filter was made with a short, heavy conductor run directly to the main earth bar rather than through a terminal block, and its length was kept under about thirty centimetres. The input and output wiring of the filter were then separated into different ducts and were not bundled together, since a filter whose input and output wires run parallel re-couples the noise it has just removed.
Cable and Cabinet Practice
The feed to the motor was replaced with a shielded cable, with the shield bonded around its full circumference at the drive end and the motor frame earthed to the same reference. That addresses the radiated field along the motor cable, which is the dominant antenna in most installations. The switching frequency of the drives was then reduced, because a lower carrier frequency reduces both the switching losses and the emissions, at the cost of slightly more audible motor noise.
The cable routing inside the cabinet was corrected at the same time. Power and signal conductors were separated into different ducts with a wide physical interval between them, the pulse and encoder cables were replaced with twisted pair constructions carrying an overall shield earthed at one end, and the pulse train interference that had coupled along two metres of shared duct disappeared once the paths were separated. The general principles behind this arrangement are those described in EMI suppression principles, and the current path they depend on is discussed in ground current and harmonic distortion.
Verifying the Result
Verification used a measurement that reflects the symptom rather than the electrical quantity. The deviation of the axis from a commanded arc was measured before and after, and it fell from about three tenths of a millimetre to under five hundredths. The depth variation on the second axis disappeared, and the extraction fan could start and stop without affecting the finish, which is the practical test that the coupling path has been removed rather than merely reduced.
A functional check of this kind is more meaningful than a spectrum measurement in this application, because the requirement is machining accuracy rather than compliance with a limit. The same diagnosis applies wherever a switching drive shares an enclosure with a control system, and the shield termination practice that makes the difference is described in PCB cable assembly. Where the control board itself is in the same enclosure, the conventions in mixed signal design guidelines apply to its own grounding.
Process Control and Verification
On a design of this kind, lost steps is the item that decides how the rest of the board is arranged. Reviewing the design before the data is released is cheaper than correcting it after the panel is in the tank, because every step downstream inherits the decision made at the front end. Documenting the assumption is part of the design work, and a short note on the drawing prevents a question that would otherwise arrive a day later and cost a day of schedule.
Process Control and Verification
On a design of this kind, lost steps is the item that decides how the rest of the board is arranged. Reviewing the design before the data is released is cheaper than correcting it after the panel is in the tank, because every step downstream inherits the decision made at the front end. Documenting the assumption is part of the design work, and a short note on the drawing prevents a question that would otherwise arrive a day later and cost a day of schedule.
Applying the Same Diagnosis Elsewhere
The pattern repeats across machine types. A CNC router that loses position, a laser cutter whose optical head drifts, a packaging line whose communication link drops when a conveyor starts: each is a control system sharing an enclosure with switching power electronics, and each is diagnosed the same way. Establish whether the symptom correlates with the operation of a specific drive, then examine the routing of the signal cable relative to the power cable before replacing any hardware.
The measurements that confirm the diagnosis are modest. A current probe on the earth conductor shows the common mode current that the drives are returning, and a voltage measurement between the control board reference and the drive earth reveals the potential difference that the signal is being referenced against. Where both are present and the symptom correlates with drive operation, the cause is established, and the countermeasures are those described in the sections above rather than a component substitution.
FAQ
Why did replacing the driver not fix the lost steps? Because the driver was not the cause. The pulse train reaching it had already been corrupted by interference coupled from the spindle drive, so the driver was counting pulses that were never generated.
Which measure contributed most? Separating the power and signal routing produced the largest change, and the input filter removed the conducted path. In installations of this kind, the routing is usually the dominant factor.
Is a lower switching frequency always acceptable? Where motor noise and torque ripple permit, yes. The trade is a small increase in acoustic noise and ripple against a reduction in losses and emissions.
