Z-Axis Force Control on a Removal Spindle: Choosing Between Rigid Pressing and Compliant Float

At a micro-screw removal station, the first question the Z axis must answer is not whether it can press down, but how many newtons it applies. An M0.8 head is about 1.5 mm across with under two square millimetres of usable contact area, and a bit needs 5 to 15 N of axial preload to engage reliably. Below that band the bit slips in the slot and cams out; above it the head takes a dent, thin-wall linkage arms or PCBs distort locally, and in severe cases micro-cracking inside the board only shows up at test. On a rigid hard-stop Z axis, the applied force is a function of workpiece height tolerance rather than process requirement. That is why force control is not optional in micro-screw work.

The goal is controllable force, not force that is as high as possible

Reading force control as pressing harder is the usual mistake. The real targets are three: no impact peak at contact, a stable plateau during loading, and no rebound when reaction torque arrives at breakaway. These map to soft landing, constant-force holding and dynamic response. Fail any one and the floor symptom is either slippage or denting, two faults with opposite root causes.

The impact peak comes from velocity. Descending rigidly, the axis strikes the head at feed speed and the kinetic energy converts into an impact force that can be several times the setpoint. This is the leading cause of dents on ultra-thin countersunk heads, and it is exactly where people conclude the force was set too high. The correct fix is to approach at contact speed, detect contact, then switch to working force.

Three implementation routes: mechanical float, pneumatic constant pressure, servo force control

• Mechanical float: a spring plus linear guide forming a compliant unit, 0.5 to 2 mm of travel, 0.5 to 2 N/mm stiffness. Simple, air-independent, lowest cost. The force varies linearly with compression, so workpiece height variation turns directly into force variation, and stiffness cannot be changed on the fly.
• Pneumatic constant pressure: a precision regulator with a low-friction cylinder, supplied at 0.05 to 0.2 MPa, holding force within about five percent. Force decouples from travel, so height tolerance no longer affects pressure. Response is in the tens of milliseconds, changeover needs a new regulator setting, and seal ageing produces slow drift.
• Servo force control: a servo or voice-coil actuator with a force sensor or current-loop estimate, 0.1 N resolution, 20 to 50 Hz bandwidth, programmable force-position profiles. The full force curve can be logged straight into traceability data. The cost is higher and calibration plus thermal compensation are required.

The boundaries are clear. For one variant at high volume with stable workpiece height, mechanical float is enough. For mixed variants with height variation, pneumatic constant pressure gives the best return. Servo force control only pays when per-screw force curves must be recorded or the process window is extremely narrow.

Setting the parameters: force first, then travel, then bandwidth

Target force follows head diameter and slot type. Cross or slotted M0.8 commonly sits at 5 to 10 N, M1.2 around 10 to 15 N, M2.0 around 15 to 25 N. Threadlocked fasteners may be run twenty to thirty percent higher during breakaway to suppress cam-out, but never beyond the denting threshold of that head, which has to be measured rather than copied.

Travel should cover workpiece height tolerance plus fixture positioning error, typically plus or minus 1 mm. Too little and the mechanism bottoms out, degrading force control into rigidity. Too much and the compliant unit loses meaning while the bit swings more under side load.

Stiffness sets how compliant the axis feels. Lower stiffness in a mechanical float is softer, but too low lets the bit rock under reaction torque and wears the slot faster. Pneumatic and servo equivalents are set by control parameters, and tuning them too soft introduces following lag.

Bandwidth sets response. Too low and the force spikes when the head lifts or the thread seizes. Too high and micro-roughness on the workpiece is treated as a disturbance, filling the force curve with meaningless ripple. Three curves settle acceptance: whether the contact segment rises smoothly without an impact peak, whether the plateau holds within plus or minus ten percent, and whether the force drop at breakaway stays acceptable. On a desktop small screw removal equipment platform, a common arrangement pairs pneumatic constant pressure with mechanical float for coarse compliance and a servo for fine positioning, splitting cost and performance across two stages.

Coupling with the nozzle, the bit and the vision system

The first coupling is deciding what floats. Floating the bit gives low mass and fast response for high-frequency disturbance; floating the workpiece gives high inertia and stability for heavy loads. Reaction torque is small on micro screws, so the bit usually floats. The second is interference between vacuum pick-up and force control: the nozzle must press against the surface to seal, and excessive lip compression shortens nozzle life sharply. The practical answer is two-stage force: 3 N to establish the seal, then working force for removal. The third is bit sticking, where a picked screw clings to the bit; the force curve shows no unloading after breakaway, which is the cheapest possible sticking detector.

The fourth coupling is the one most often missed. Vision supplies X-Y and slot angle while force control supplies Z, and the two accuracies must match. If vision residual is 0.1 mm while the floating joint allows only plus or minus 0.5 mm laterally, the bit gets wedged into the slot by the side component of force: it appears to work at first and wears visibly later. This is why an automatic vision screw removal machine configuration normally specifies plus or minus 1 mm of lateral float. Absorbing vision residual mechanically is cheaper than chasing ever higher camera accuracy.

A failure-mode list and how to accept the station

Six failure modes cover the floor, and each points at a parameter rather than different hardware:

• Head dents or craters: target force too high, or no soft landing at contact;
• Bit cam-out: target force too low, or the bit edge is worn;
• Proud screws and missed screws: the force-to-position switching threshold is wrong so contact is never detected;
• Lateral bit wear: stiffness too low, or insufficient float travel;
• Torque spike at breakaway: bandwidth too low to respond to a seized thread;
• Workpiece distortion: fixture support is inadequate, and no amount of force control compensates for weak support.

Acceptance should not rely on a single number. Run thirty consecutive screws and record contact peak, plateau mean and standard deviation, and unload time after breakaway; only units inside all three bands pass. Keep three samples for destructive inspection of head denting and slot deformation. Force control is not a universal patch: without fixture rigidity, bit condition and a torque ceiling, even good force control merely turns a visible fault into an invisible one. In an M0.8-M2.0 screw removal solution the Z-axis force control is normally delivered as one parameter set together with fixture rigidity and torque monitoring rather than as a stand-alone option, for precisely this reason.