Extracting Stripped and Sheared Micro Screws: Four Process Routes and Where Each Stops Working

The conclusion first: a stripped head and a sheared shank are different failures and do not take the same route; classifying the failure matters more than choosing the tool.

On a micro-screw removal floor, "it will not come out" resolves into two states. In the first the head is still present but the cross or hex recess is rounded, so the bit spins. In the second the head has sheared off and only a stub of thread remains in the hole, or the break sits at the transition between thread and shank. The first is a broken torque path; the second has no geometry to grip at all. Treating both the same way is the main reason rework scrap rates stay high.

Classify the failure first: four questions choose the route

The conclusion first: four questions asked before touching the part remove more than half the risk of choosing the wrong tool.

The first question is how much torque-transmitting geometry remains on the head. If the recess depth is under thirty percent of nominal, or a hex has gone circular, trying another bit of the same size only enlarges the damage. The second is whether the screw is already loose: some apparent cam-outs are screws that have backed out and are simply trapped, which a light lateral shake with the part held down will reveal, and those need no extractor at all. The third is hole depth and access, since EDM extraction of a deep stub requires reaching the break, and once depth-to-diameter passes about three to one, electrode stiffness becomes the limit. The fourth is the parent material: aluminium, magnesium, titanium and engineering plastics tolerate cutting heat very differently, aluminium and magnesium gall badly under high-speed drilling, and plastics soften or melt from local heating.

Route one: reverse-flute taps and extractors, for a damaged head over an intact thread

The conclusion first: the cheapest route and the one most dependent on technique, viable above M1.2 in steel or aluminium.

A reverse-flute tap re-cuts a left-hand engagement inside the damaged bore, while an extractor drives an eccentric edge into the remaining wall. Both assume enough wall thickness to bite. At M0.8 and below there is under a tenth of a millimetre of wall, the edge has almost nothing to grip, and forcing it typically splits the remnant down the middle, converting a repairable part into scrap.

The real risk on this route is axial load. A reverse tap needs downward pressure to start cutting, and the parent material of a micro screw is often a thin wall itself, so a little excess load distorts the part; once the threaded boss deforms, a new screw will not seat even if the old one came out. The workable practice is to clamp the part in a fixture with backing so the load path runs through the fixture rather than the wall.

Route two: micro EDM for sheared stubs, where hardness is high, the hole is deep and cutting force is forbidden

The conclusion first: EDM is the only route that applies essentially no mechanical force, paid for in equipment and cycle time, and it suits high-value parts and deep breaks.

Micro EDM uses a tungsten or copper-tungsten electrode a few tenths of a millimetre across to erode the stub face layer by layer, first sinking a flat into the bore and then cutting a slot with a square electrode that can be gripped or driven in reverse. No cutting force is generated, which makes it the friendliest route for thin walls, distortion-prone parts and already-assembled hardware; hardness is no obstacle either, so heat-treated 12.9 screws and threadlocked joints are both workable.

The governing parameters are electrode wear and flushing. Smaller electrodes wear faster, deep holes need a succession of electrodes, and the coaxiality of each succession decides whether the tool actually reaches the stub centre; insufficient flushing piles erosion debris into a secondary short. In practice a small screw removal equipment platform with coordinate positioning and Z-axis force control is often repurposed as the EDM positioning base, precisely because it can bring the electrode directly over the hole and hold position at micron level without relying on operator feel.

Routes three and four: laser assist and design concession, for the smallest sizes

The conclusion first: below M0.8 the optimal answer is often not extraction at all but accepting the stub and changing the design.

Laser assist uses short pulses to sink a conical crater in the stub face, giving a reverse tool something to bite, or to vaporise the remnant thread outright before re-tapping or switching to a self-tapping screw. It is non-contact and force-free, but the heat-affected zone alters local hardness, markedly so in aluminium, and reach into a deep hole is limited by diameter and aspect ratio. Pairing laser alignment with an automatic vision screw removal machine is a common combination, because the ablation point must land on the stub centre and a two-tenths offset can damage the parent thread.

Design concession moves the problem upstream instead: replace the vulnerable joint with a stud-and-nut arrangement that can be renewed, or convert a screwed joint into a snap feature with a single screw, deleting the extraction step from the process. On watches, earbuds and AI glasses this is already the mainstream choice. Admitting that some joints should not be screwed is itself a sign of design maturity.

Move extraction upstream into the removal station: prevention is always cheaper than extraction

The conclusion first: one extraction costs roughly twenty to fifty times a normal removal, so process attention belongs on preventing the break.

The gap comes from three parts: the labour of extraction, the material loss of a scrapped part, and the opportunity cost of a stopped line. The causes of stripping and shearing are highly concentrated: no torque ceiling, worn bits left in service, excessive axial load, wrong release order, and threadlocker or corrosion that was never identified. All five can be constrained at the removal station.

The method is to make the torque-angle curve the arbiter: stop when breakaway exceeds the model ceiling rather than adding force; classify a mismatch between angle growth and torque as bit slip and change the bit immediately instead of retrying; verify after removal that no debris remains in the hole. In an M0.8-M2.0 screw removal solution these constraints are delivered with the extraction routes as one rework specification: the first part ensures the great majority of screws come out normally, the second covers only the residual small-probability events.

One last point about people. Extraction depends heavily on feel, feel does not transfer through documentation, and that is why staff turnover at these stations shows up directly as scrap-rate variation. Writing the decision rules into the machine and the extraction routes into the work instruction is the only way to keep rework quality independent of who is on shift. Experience is not replaced; it moves from the hands into the parameters.