Automated fastening: torque, angle and the joints that lie to you

Torque is a proxy, and friction spoils it

Most of the effort applied when tightening is consumed overcoming friction under the head and in the threads, with only a portion converted into clamping force. Anything that changes friction changes the relationship: a different fastener coating, oil on the surface, a burr, a plated component from a new supplier, or a fastener reused after removal. This is why two joints tightened to the same figure can be clamped very differently, and why a torque-only strategy quietly tolerates a supplier change that alters surface condition. Where clamping force genuinely matters, torque alone is weak evidence and should be supplemented.

Watching the signature, not just the endpoint

Modern controlled tools record how torque develops against rotation angle throughout the tightening, and that curve reveals what the endpoint cannot. Cross-threading shows as elevated resistance early. A missing component or a hole that is too large shows as rotation continuing with little torque rise. A joint tightening far too quickly indicates the fastener bottomed out or something is trapped. Setting acceptance windows on the shape of that curve, rather than only on the final value, catches a class of defects that torque checking never sees. It also requires deciding the limits from real joint data rather than from a default.

Feeding fasteners is where the stoppages come from

The driving is reliable; the supply of screws to the driver is not. Fasteners tangle in bowls, jam in feed tubes, arrive at the nose the wrong way round, and vary between batches enough to upset an escapement set up on the previous delivery. Coated or waxed fasteners behave differently from bare ones. Small variations in head form or thread rolling that are irrelevant to function stop a feeder. Specify the fastener with feeding in mind, control the supplier's changes to it, keep the feed path accessible for clearing, and expect this to be the most frequent operator intervention on the station.

The joint itself decides how forgiving the process is

A hard joint reaches its target after very little rotation, so the tool must react quickly and small variations produce large differences. A soft joint with a gasket or compressible material tightens gradually and is far more tolerant, but relaxes afterwards as the material settles. Threads formed directly into plastic or thin sheet have a narrow window between secure and stripped. Understanding which behaviour applies determines tool selection, strategy and limits, and it explains why a setting that works on one joint of a product is entirely wrong for another joint a few centimetres away.

Records are worth having only if they identify the joint

Controlled tools produce a result per tightening, and storing those results is the basis of any later defence that a product was assembled correctly. The value depends on knowing which joint on which unit each record belongs to, which needs the tool to know the sequence and the assembly to carry an identity. Without that association you have a large collection of readings and no ability to answer a question about a specific serial number. Decide the sequence enforcement and the identity linkage at design stage, because retrofitting them into a running station usually means rebuilding the operation.

Frequently asked questions

Is reaching the target torque enough to prove a joint is correct?

No. Torque is consumed largely by friction, so a fastener can reach its target with far less clamping force than intended if the surface condition changed, or with far more if it is dry and galling. Monitoring how torque develops against angle gives a much better picture, since the shape of that relationship exposes cross-threading, missing parts and bottomed fasteners. For joints where failure matters, combine the two and validate the limits against joints you have physically examined.

Why does our screwfeeder jam so often?

Usually fastener variation or geometry the feeder tolerates poorly. Head form, thread rolling, coating thickness and burrs vary between batches and suppliers enough to upset an escapement, and long thin fasteners tangle readily in a bowl. Lock the fastener specification down with the supplier including any change notification requirement, trial with material from more than one batch, keep the feed path accessible so an operator can clear it in seconds, and monitor jam frequency as an indicator that something upstream has changed.

Should every joint on an assembly be monitored and recorded?

Monitor according to consequence. Joints whose failure creates a safety, warranty or functional problem justify full recording with sequence enforcement so a missed fastener stops the unit progressing. Joints holding a cosmetic cover generally do not. Recording everything is technically easy and generates data nobody reviews, while diluting attention to the results that matter. Decide the list with engineering and quality, document why each joint is on or off it, and revisit it when field failures suggest otherwise.