Two limits, not one
Every bolted joint has a torque window bounded by two failures. At the bottom is insufficient preload: the joint is not clamped hard enough, and vibration or service load works it loose. At the top is over-torque: the thread strips, the head shears, or the material under the head yields.
The size of the window depends on the joint, not on the screw alone. It is set by the strength of the mating thread, the bearing area under the head, and the friction coefficient of the interface. Change any one of those and the window moves.
Why it narrows as the screw gets smaller
Torque converts to preload through a relationship that scales with diameter, but the variability does not scale down with it. A driver that holds a torque repeatability of a few percent at M6 is still a few percent at M1 — and a few percent of a very small number is a very small absolute margin.
At the same time the strength of the mating thread falls faster than the screw strength, because the engaged thread area scales with diameter while the material it engages may be plastic or a thin light-alloy wall. So the upper limit of the window comes down at the same time as the noise floor is unchanged.
The three variables that set the window
- Mating thread strength — a brass insert is not the same joint as a tapped aluminium hole or a moulded plastic boss, even with the same screw.
- Bearing area under the head — a small head concentrates load, and in soft material that becomes the limiting factor before the thread does.
- Friction coefficient — this is what converts torque into preload, so a change of finish changes the preload achieved at the same torque setting.
What to do about it
The practical answer is to stop treating the torque value as a number to be looked up and start treating it as a property of the joint to be measured. The joint that matters is the one on your drawing: your insert, your material, your finish, your driver.
Two specifications make the window more controllable. First, control the friction coefficient by specifying the finish rather than leaving it to the plating supplier. Second, validate the drive against the production driver at production rate, because a recess that engages a hand driver can cam out on a pneumatic tool.
- Determine the upper limit from the strength of the mating thread, not from the screw
- Set the target so the upper tolerance of the driver stays below that limit
- Specify the finish because it sets friction, and friction sets the preload you actually get
- Validate on the production driver and the production joint, then record the setting