Servo Drives, Encoders, and Homing Faults on Stone CNC Machines

A stone CNC machine that will not home is a machine that is not making money, and the fault message on the screen is rarely helpful about why. Servo alarms, following error trips, encoder faults and homing failures all announce themselves in similar ways: production stops, the operator reboots, sometimes it clears and sometimes it does not. In the absence of a systematic approach, shops end up replacing components on suspicion, which is expensive and frequently leaves the actual cause in place.

The underlying system is more approachable than the alarm codes suggest. A servo axis is a closed loop: the controller commands a position, the drive powers the motor, the encoder reports where the axis actually went, and the controller compares the two and corrects. Almost every fault in this category is a break somewhere in that loop, and the loop has a small number of components. Understanding where each one sits, and what its failure looks like, turns a mysterious alarm into a short, ordered investigation.

How the Closed Loop Works

The controller issues a position command, typically many times per second. The servo drive converts that command into current supplied to the motor, and the motor turns. Attached to the motor, or in some designs to the axis itself, an encoder measures actual rotation or position and reports it back. The drive and controller compare commanded position against actual position, and the difference between them is called following error. A small following error is normal and is how the system regulates itself.

The system alarms when following error exceeds a configured threshold, because that indicates the axis is not going where it was told. Crucially, this can happen for entirely mechanical reasons as well as electrical ones. If the axis is physically obstructed, if a way is dry and binding, if a cutting load is beyond what the drive can deliver, or if something has jammed, the motor cannot achieve the commanded position and the loop reports the discrepancy. A following error alarm is therefore a symptom rather than a diagnosis.

Encoders come in incremental and absolute varieties, and the distinction matters for homing. An incremental encoder reports change in position but has no knowledge of where it is when powered on, so the machine must find a reference position at startup. An absolute encoder retains position knowledge, usually with a backup battery, and can report where the axis is immediately at power up. Machines with absolute encoders that suddenly need homing every morning are usually telling you the backup battery has failed.

Homing is the routine that establishes the relationship between the machine's coordinate system and physical reality. It typically involves driving the axis toward a limit or home switch, detecting the transition, then backing off and searching for a precise reference mark on the encoder. Because it involves both a switch and an encoder signal, homing faults can originate in either, which is why they are among the more commonly misdiagnosed problems.

The final element is the power and signal environment. Servo systems generate substantial electrical noise, encoders transmit low level signals, and a stone shop is full of water, vibration and heavy inductive loads switching on and off. Cable shielding, grounding integrity and connector condition are not peripheral details in this system; they are frequently the actual cause when faults appear intermittently and defy component replacement.

Diagnosing the Common Fault Families

Following Error and Servo Alarms

Start by determining whether the axis can move freely by hand with power removed and the drive disconnected according to the machine's safe procedure. An axis that is stiff, notchy or immovable has a mechanical problem, and no amount of drive tuning will fix it. This single check separates the mechanical from the electrical branch of the investigation and takes only a few minutes.

If the axis moves freely, the next questions concern load and tuning. A following error that appears only during heavy cuts points toward the drive reaching its limit or toward tuning that is too soft to hold position under load. A following error that appears during rapid moves points more toward acceleration settings or a motor that is not delivering expected torque. A following error at any speed, including slow jogging, points toward feedback rather than power.

Tuning should be treated with caution rather than as a first response. Adjusting servo gains can make an alarm stop appearing without fixing anything, because a looser tolerance simply tolerates a larger error. If the underlying problem is a binding axis or a failing bearing, raising the following error threshold converts an alarm into accelerating mechanical damage. Any tuning change should be recorded, along with the reason, so that the next person to look at the machine understands what was altered and why.

Encoder Faults

Encoder faults often present as erratic behaviour rather than a clean failure. Position that drifts, an axis that jumps unexpectedly, or intermittent alarms that clear on restart all suggest a feedback signal that is being corrupted rather than lost. In a stone shop the leading causes are moisture ingress into a connector, a damaged cable in a drag chain that has been flexing for years, or shield continuity that has been broken during a previous repair.

Inspecting the encoder cable along its full run, particularly where it enters cable carriers and where it has been secured with ties, frequently reveals the problem. Cable that has been chafing against a fixed edge will fail eventually, and the failure is usually intermittent for a long period before it becomes permanent, which is exactly the pattern that makes these faults so frustrating to chase.

Homing Failures Specifically

Homing switches live in a hostile position by design, since they must be located at the extreme of travel where slurry and water accumulate. Contamination on a proximity sensor face, water in the switch body, or a build-up of dried slurry on the actuating flag all prevent reliable detection. Cleaning and inspecting these switches should be part of routine maintenance rather than something done only when homing fails.

Mechanical alignment of the switch and its actuating flag matters as much as the switch itself. A flag that has been knocked during a crash, or a switch bracket that has loosened under vibration, changes the trigger point. The machine may still home, but to a slightly different position than before, which shows up as parts that are consistently offset by a small amount. That symptom is often misread as a fixture or program problem.

The two-stage nature of most homing routines is worth understanding when diagnosing. The switch provides a coarse reference and the encoder marker provides the fine one. If the switch trigger point drifts to sit very close to the encoder marker, the machine can pick a different marker on different attempts, producing a home position that shifts by a fixed increment. This intermittent, quantised error is a classic signature and points directly at switch position rather than at anything failing.

Approach speed and direction parameters affect repeatability. Homing too fast reduces the precision with which the transition is detected, and parameters that have been altered, perhaps to speed up a slow startup routine, can undermine accuracy in ways that are not obvious. Recording the original homing parameters when the machine is commissioned gives a reference point that makes later comparison trivial.

After any crash, homing should be verified rather than assumed. An impact can move a switch bracket, bend a flag, or shift the relationship between the encoder and the mechanics. Running a home cycle and then verifying a known reference position on the machine confirms that the coordinate system still corresponds to physical reality, which is the assumption every subsequent program depends on.

Prevention, Environment and Sensible Spares

Environmental protection is the highest value preventive measure for this whole class of fault. Enclosures that are properly sealed, cable glands that are intact, drag chains that are not worn through, and connectors that are oriented so water drains away rather than pooling all prevent the moisture ingress that causes most intermittent electrical faults in stone shops. Inspecting these during routine maintenance costs very little.

Grounding and shielding deserve periodic verification, especially after any electrical work. A shield that was disconnected during a repair and never reconnected may cause no symptoms for months and then produce intermittent encoder faults when a new load is added to the shop's supply. Documenting the correct grounding arrangement, and checking it after any intervention, prevents a category of fault that is otherwise almost impossible to trace.

Heat is an underestimated factor. Servo drives and control cabinets are sensitive to ambient temperature, and cabinet cooling systems in stone shops clog with dust. A cabinet filter that has not been changed in two years raises internal temperatures, which shortens component life and can cause thermal alarms that appear to be axis faults. Filter replacement is trivial maintenance with a real reliability payoff.

Absolute encoder batteries should be replaced on schedule rather than on failure, because failure means losing position reference and, on some machines, a more involved recovery procedure. Noting the battery type and the replacement date in the machine log converts an unpredictable disruption into a five minute planned task. This is one of the most commonly neglected items on stone CNC machines.

Spares strategy should follow lead time rather than probability. Encoder cables, limit and home switches, connectors and cabinet filters are inexpensive and fail relatively often, so they belong on the shelf. Drives and motors are expensive and fail rarely, so what matters is knowing the exact part numbers, the supplier and the realistic lead time in advance. That information in the machine file turns a crisis into a scheduled repair.

Operator reporting closes the loop on all of this. Faults that clear on restart are frequently never reported, so the pattern that would reveal the cause never assembles. A simple log at the machine where operators note the date, the alarm and what they were doing at the time costs nothing and repeatedly turns an unsolvable intermittent fault into an obvious one. The information exists either way; the only question is whether anyone wrote it down.

The overall discipline is to work the loop in order rather than by intuition: mechanical freedom first, then load and tuning, then feedback signal integrity, then power and environment. Most faults in this category are found in the first two steps, and most of the expensive mistakes come from skipping to component replacement before the cheap checks have been done. A written diagnostic sequence posted at the machine is worth more than it sounds.

Machines that run reliably still need tooling that matches the material, and the CNC bits, wheels and abrasives that keep production moving are stocked at Dynamic Stone Tools. Shops planning preventive maintenance alongside a tooling refresh can browse machinery accessories and consumables at dynamicstonetools.com, where the catalogue is organised by fabrication stage so it is easy to see what each process needs.