Why a Drill Clutch Number Is Not a Torque

This page contains affiliate links. If you buy through one we may earn a commission, at no cost to you.

Useful, and not for the reason printed on the ring. The numbers are ordinal. They index a spring preload, they have no units, and no maker publishes what any of them means in newton metres. What a clutch does well is repeat the same depth on the same joint, and what it cannot do is hit a figure.

There are two separate reasons, and both are arithmetic. The first is that a ladder of settings spaced evenly across a range is nearly useless at the bottom of that range, where almost all the work is. The second is that whether the clutch governs at all depends on a property of the joint nobody can read off a tool, and on a small driver the motor's own ceiling is often doing the limiting instead.

Seven settings is not seven useful settings

We do not carry drills, but we carry powered drivers that publish exactly what a drill clutch does not: numbered levels with units on them. That makes the ladder arithmetic checkable.

A Fanttik L2 Pro cordless electric screwdriver set publishes "7 torque settings from 0.5 to 6 N.m". Space those evenly, which is what a mechanical ring does, and the step is 0.9167 newton metres: 0.5, 1.417, 2.333, 3.250, 4.167, 5.083, 6.0.

Look at the two ends of that ladder. Clicking from the first setting to the second multiplies the torque by 2.833. Clicking from the sixth to the seventh multiplies it by 1.180. The control is almost 2.5 times coarser, proportionally, at the bottom than at the top, and the bottom is where a small screw lives.

Space the same seven settings geometrically instead, each 1.5131 times the last, and you get 0.5, 0.757, 1.145, 1.733, 2.622, 3.967, 6.0. Every click is the same 51 percent change, which is what a human expects from a dial. Nobody builds that, because the mechanism is a spring and a ramp and its natural behaviour is linear in preload.

The consequence is specific. A small wood screw seating in softwood wants somewhere in the region of one and a half to three newton metres. On the linear ladder, that entire window sits between settings one and two. The control has one usable position in exactly the range where you want three, and six positions above it that are all too much.

A Fittwt mini electric screwdriver set is the best behaved tool in this catalogue for precisely this reason, and not because it is more precise. It publishes "Six torque levels from 0.2 to 0.7 newton metres", which is a span of only 3.5 to 1. Evenly spaced that is a 0.1 newton metre step, so its first click multiplies by 1.5 and its last by 1.167. A narrow range makes an even ladder behave itself. A wide range does not, and the wide ranges are the ones that get advertised.

The part that decides whether the clutch matters at all

Now the dynamic half, which is the actual answer to the real world question.

A clutch is a ball or a cam sitting in a detent against a spring. To release, it has to climb out of that detent, which takes a finite amount of output rotation: a few degrees. During those few degrees the screw keeps turning and the joint keeps tightening.

So the overshoot is the product of two things: how far the detent has to travel, and how fast the torque is rising as it does. That second quantity is the joint rate, in newton metres per degree, and it is a property of what you are screwing into rather than of the tool.

A long screw pulling into timber is a soft joint. Call it 0.2 newton metres per degree. Over 5 degrees of detent travel that adds 1.0 newton metre, so a setting of 2.333 delivers 3.33: an overshoot of 43 percent. Annoying, survivable, repeatable.

A machine screw bottoming on metal is a hard joint. Call it 3.0 newton metres per degree. The same 5 degrees would add 15 newton metres, turning a 2.333 setting into 17.3: an overshoot of 644 percent. The ratio of the two overshoots is 15, which is exactly the ratio of the joint rates, because the detent travel is the same in both cases.

My first conclusion from that was wrong, and the ceiling is why

The obvious reading is that a clutch is dangerous on hard joints. Then I put the 15 newton metres back against the tool that was supposed to deliver it, and it cannot: the Fanttik publishes a 6 newton metre ceiling and its own drawbacks say so plainly, "Torque ceiling of 6 N.m limits heavy jobs".

Redo it. Starting at a 2.333 setting on a 3.0 newton metre per degree joint, the torque reaches 6 newton metres after only (6 minus 2.333) divided by 3, which is 1.22 degrees. That is less than a quarter of the detent travel. The motor stalls before the clutch finishes releasing, so on that joint the clutch setting does nothing at all and the thing limiting the torque is the tool running out of grunt.

Which gives a threshold, and it is the one genuinely useful expression on this page. The joint rate at which the ceiling arrives before the clutch is the ceiling minus the setting, divided by the detent travel. For this tool at this setting that is 0.73 newton metres per degree. On anything softer than that, the clutch governs. On anything harder, the motor does, and the ring is decoration.

And the scaling runs against intuition. Put the same setting on a drill with a 20 newton metre ceiling and the threshold becomes 3.53 newton metres per degree. On a 40 newton metre drill it becomes 7.53, ten times the small driver's. A more powerful tool does not control torque better. It removes the stall that was accidentally doing the controlling, so it has ten times more room to overshoot before anything stops it. The clutch on a weak tool is backed up by the tool's weakness. The clutch on a strong tool is on its own.

Why the mechanism has to exist, even though it is imprecise

A fair question is why not simply listen and stop. The rotation speeds our own listings publish answer it.

The Fanttik and the Fittwt both publish 250 revolutions a minute, which is 1,500 degrees a second. A SATA 32-piece cordless electric screwdriver publishes 320 revolutions a minute and 1,920 degrees a second. An alert human reaction is about 200 milliseconds, which at those speeds is 300 and 384 degrees of output rotation. On a screw with a 1.25 mm thread pitch that is 1.04 and 1.33 millimetres of extra travel after you decide to stop.

So on a powered tool there is no driving by ear. Something mechanical has to make the decision, which is why the clutch is there and why our own guide to driving screws cleanly gives it the correct job description: "use the clutch setting to control depth, so you do not overdrive". That is exactly right, and it is worth noticing how carefully it is phrased. Depth, on a repeated joint. Not torque, and not a number.

The spec that cannot be solved

Two of our powered tools publish a setting count and one endpoint, which is not enough to reconstruct anything.

The SATA publishes "three torque settings" and "up to 5 N.m electric", with no bottom of the range anywhere in prose. Three settings topping out at 5 could be 1, 3 and 5, in which case the first click triples the torque. Or it could be 0.5, 2.75 and 5, in which case the first click multiplies by 5.5. Both satisfy everything published. There is no way to choose between them, and they imply completely different tools.

The same record also publishes "Up to 5 N.m electric plus 8 N.m manual torque", which is worth reading twice. The tool's largest number belongs to its manual mode, and the manual mode has no torque control of any kind: it is you turning a handle. So the headline capability is the uncontrolled capability. An iFu 42-in-1 cordless precision electric screwdriver kit does the same thing in the other direction, publishing "0.2 N.m of electric torque for [delicate screws]" with a firmer manual mode above it, and its own drawbacks are honest that the "Electric torque is a light 0.2 N.m, not for tough fasteners". In both cases the controlled figure and the maximum figure are different numbers for different modes, and only one of them is the one the clutch applies to.

Nobody publishes a tolerance on a powered setting. The hand tools do

This is the finding that puts the whole question in proportion. Across every powered driver in this catalogue there is not one published accuracy figure for a torque setting. Counts, spans and ceilings, but no tolerance.

Both manual torque tools publish one. A YIYEIE 90-piece master torque screwdriver set publishes "a 10 to 70 in-lb range in 0.5 in-lb steps" with "A click [that] signals when you hit the preset value", and its own drawbacks add the right caveat: "Accuracy depends on the tool's calibration". A DSBYKJ 17-piece torque wrench set publishes a "5 to 30 Nm range, plus or minus 3 percent accuracy, a stated 10,000-cycle test" and a 72-tooth ratchet, and is straight about its limits in its own drawbacks: "Low 5 to 30 Nm range, not for car lug nuts".

Put the resolution side by side. The YIYEIE's 10 to 70 inch pounds is 1.130 to 7.909 newton metres, and its half inch pound increment is 0.0565 newton metres. That is 121 settings over a span that overlaps the Fanttik's 0.5 to 6 almost entirely. Its step is 5.0 percent of its lowest setting. The Fanttik's first step is 183 percent of its lowest setting. Sixteen times finer, over the same ground, in a tool with no motor in it.

Which is not an argument for using a hand driver for everything. It is an argument about what each tool is for. The powered clutch exists because a human cannot react inside 300 degrees of rotation. The hand tool exists because a figure that matters needs a tolerance, and a tolerance needs a mechanism that moves slowly enough to have one. Our own page on what a torque wrench does and why the range matters covers the wrench side properly, and our page on why a lubricated bolt snaps at the right torque works out what happens downstream of the number, which is a separate mechanism again and not repeated here.

One more thing the catalogue does not reconcile

The torque tools on this site span 0.2 to 30 newton metres across six products, in two unit systems, and not one listing converts. The YIYEIE is in inch pounds and everything else is in newton metres, which means its 10 to 70 looks like a bigger number than the Fanttik's 0.5 to 6 and is actually a near identical range, 1.13 to 7.91. A reader comparing the two from the listings alone would conclude the hand tool is roughly ten times stronger. It is not. It is the same strength with sixteen times the resolution and a published tolerance.

The honest limit, and what to do with a clutch ring

Three of the inputs here are representative rather than published, and they are the ones that set the absolute numbers. The 5 degrees of detent travel is typical for a ball clutch and not stated by anyone. The 0.2 and 3.0 newton metres per degree joint rates are engineering figures for soft and hard joints, not measurements of any particular screw. And the even spacing of the seven settings is an assumption, because the listing gives endpoints and a count and nothing between them.

The threshold expression is exact, and the conclusion that it scales with the tool's ceiling survives any numbers you substitute. So does the ladder arithmetic, which depends only on the endpoints and the count.

Which leaves one practical move, and it is the opposite of what the ring invites. Do not look for the setting that corresponds to your screw. There is no such correspondence published and the bottom of the ladder is too coarse to find it anyway. Drive one screw into an offcut of the actual material at a low setting, click up until it seats flush, and then leave it there for the whole batch. That uses the only thing a clutch is genuinely good at, which is repeating itself, and it sidesteps both of the problems on this page: you never need to know the newton metres, and you have calibrated against the real joint rate rather than guessing it.