Why a Drilled Hole Has Two Depths

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

You cannot, and the obstacle is not the drill. It is that a twist bit does not make a flat bottomed hole, so a hole drilled to "exactly 20 mm" has two depths at once and the question has not said which one it wants. They are further apart than anyone expects. On a 3/8 inch bit with an ordinary grind they differ by 2.86 mm, which is 14 percent of the figure you set, and the difference changes when you change bits.

Once that is on the table the problem becomes tractable, because the remaining error sources are small and one of them is handled by a technique that involves no purchase at all.

The cone is the whole problem, and it is arithmetic

A twist drill ends in a conical point. The tip reaches the bottom of the hole first, and the hole only reaches full diameter where the cone meets the flutes. The distance between those two planes is half the diameter divided by the tangent of half the included point angle.

The ordinary grind on a general purpose high speed steel bit is an included angle of 118 degrees. Half of that is 59 degrees, whose tangent is 1.664, so the taper is the radius divided by 1.664, which is 0.300 of the diameter. On a 3/8 inch bit, which is 9.525 mm, that is 2.86 mm of cone.

So plunge until the tip sits 20.00 mm down and you have made a hole that is 20.00 mm deep and also 17.14 mm deep. Which number you need depends entirely on what goes in it. A dowel, a shelf pin, a threaded insert, a bolt shank, a cabinet hinge boss: every one of those is a cylinder with a flat or near flat end, and every one of them bottoms on the shoulder at 17.14 mm. The cone below it holds nothing and supports nothing. The only job for which the tip figure is the right figure is not drilling through, and that is a job you measure differently anyway.

Our own catalogue publishes four different point angles

This is where it stops being a technicality, because the multiplier is not a constant. It is a property of the grind on the bit in the chuck, and we sell bits with four different ones.

A 10 piece 3/8 inch SDS Plus carbide rotary hammer bit set publishes tips of YG8 tungsten carbide "ground to a 135-degree angle". Half of 135 is 67.5 degrees, tangent 2.414, so the taper is 0.207 of the diameter, which on its own 3/8 inch bits is 1.97 mm. Compare that with the 2.86 mm above. Two bits of identical diameter, set to identical depth, leave sockets that differ by 0.89 mm in the only dimension that matters for holding anything. Nothing about the operator changed.

At the other extreme a 7 piece M2 high speed steel 60 degree centre drill set is ground at 60 degrees, whose half angle has a tangent of 0.577, so the taper is 0.866 of the diameter. On its largest size that is a cone 4.33 mm long on a 5 mm pilot, and the cone is most of the hole. That is not a defect, it is the entire function: a centre drill exists to make a short conical seat that a lathe centre or a following bit can sit in without wandering, and its record is clear that it is a "specialist starter, not general-purpose". If you reach for one to establish a depth you will be out by most of the depth.

And the countersinks sit in between at 82 degrees, giving 0.575 of the diameter, because 82 degrees is the included angle of a flat head screw rather than a drilling angle at all. Our page on countersinking screws for a flush, clean finish states that plainly, and is right to: "Most wood screws use an 82 degree countersink. Confirm your screws match before drilling." That is the correct instruction and the correct reason, and it is worth noticing that the one angle in this whole set which is actually standardised is the one that is not about drilling.

A stop collar fixes the right problem and then leans on the wrong surface

The standard answer to this question is a stop collar, and it is a good answer, because of what it changes rather than what it blocks. A collar converts a dynamic judgement into a static measurement. Instead of watching a spinning bit disappear into a dusty hole and deciding when to stop, you set the bit's protrusion below the collar face once, on a bench, with a rule or a caliper against a flat reference. That is a far better class of measurement.

A four blade countersink set with adjustable depth collars is the plain version, and its record describes the mechanism in exactly the right terms: "Once set, the collar bottoms out on the surface at the same point every hole." Read that sentence as a specification rather than a selling point and it names its own limitation. The collar does not reference the bottom of the hole, or the chuck, or anything rigid. It references the workpiece surface, and that surface is usually wood.

Put a number on it. These sets run a 1/4 inch hex shank, 6.35 mm, inside a collar of roughly 12 mm, which leaves an annular bearing face of 81.4 square millimetres. A hundred and fifty newtons of thrust, which is about what a hand leaning on a drill delivers, is 1.84 megapascals on that face. Softwood's compressive strength across the grain, at the limit where it stops springing back, is around 3 to 5 megapascals. So you are at a third to a half of the yield of your own reference plane, which means it deflects while you are drilling and the hole ends up deeper than the setting, by an amount set by how hard you pushed. Lean twice as hard and the reference surface is at its limit and starts denting for good.

Our page on why drill bits go blunt in metal already identified the fix for part of this and it deserves the credit rather than a repeat: a collar that rides on a bearing does not spin against the work, so it does not scorch or burnish its way downward while it is loaded. That is a real distinction between collar designs and that page makes it with the right product.

The other half of the collar problem is the clamp, and all of our own records own up to it. The four blade set lists "Collar screws can loosen under vibration" among its drawbacks. An eleven piece adjustable countersink set with replaceable pilots lists "Depth collar screw can slip if loose" and its performance text repeats that "the depth collar set screw needs the Allen wrench snug or it can slip under load". That is not a quality complaint, it is a description of what the joint is. A single hex screw pressing an aluminium collar onto a hardened steel shank is a friction grip, and friction grips creep under repeated axial impulses. Which is worth connecting to the same record's selling point: the "universal 1/4-inch hex shank fits impact drivers". An impact driver delivers nothing but repeated impulses. The feature that makes the set convenient is the feature that unsets the collar.

The same page of ours on countersinking for a clean, flush finish gets the handling right and is worth following: "Run slow, near 200 RPM, especially in wood." That figure comes straight from the four blade set's own record, and slow is also the right speed for depth, because a collar arriving slowly at a yielding surface settles instead of punching into it.

Two record defects found while doing this

The carbide set above lists its suitable materials as "block, concrete, cement, porcelain tile, ceramic pots, stone, brick and wood". The first seven are what a YG8 carbide tip on an SDS Plus shank is for. Wood is not. The shank only fits a rotary hammer, the hammer action splits timber rather than cutting it, and a blunt 135 degree carbide tip tears fibres instead of severing them. Its cons are honest about the thing that actually constrains it, which is that all ten bits are the same diameter.

The same record publishes bits "each about 6 1/3 inches long" and no flute length anywhere, which is the figure that decides how deep a hole you can drill. A good part of an SDS Plus bit's length is shank, and past the end of the flutes the chips have nowhere to go, so the bit stops advancing whatever your collar says. Overall length is not drilling depth and no listing in this category separates them.

The centre drill set has a stranger one. It publishes its sizes as "1.0 to 5.0" with no unit in any prose field, in a category where the convention is millimetres of pilot diameter but where nothing says so. And it publishes a workpiece hardness range "from 170HB up to 200HB" in the same sentence that lists aluminium and brass as suitable materials. Aluminium is roughly 15 to 30 on that scale and brass is roughly 50 to 100, so both sit far below the stated floor. A minimum hardness is not a sensible limit for a cutting tool in the first place, since softer stock is easier to cut. Its cons correctly keep only the ceiling: "Hardness rated only to 200HB".

The honest limit, and the technique that beats all of it

The limit is specific and slightly absurd. Everything above turns on the point angle, and the point angle of an ordinary twist bit is the one figure no bit listing we carry publishes. Two of our sets publish theirs, at 135 degrees on carbide and 60 degrees on centre drills, and both of those are specialist. For the general purpose bits that do almost all the drilling anyone does, the 118 degrees used throughout this page is the industry default rather than a number read off a record, and the only place you will find your own bit's angle stated is on a sharpening jig. Our drill bit sharpener coverage is where the figure turns up at all, as a machine setting: one unit is fixed at 118 degrees, another adjusts from 95 to 135. Which is a good reminder that a resharpened bit may not have the angle it left the factory with.

So measure it rather than assuming it. Lay the bit on paper, trace the point, and any protractor settles it in ten seconds. Or skip the angle entirely with the substitution trick: drill a test hole in scrap, push a flat ended offcut of the dowel or bolt in, and mark where it stops. That mark is the shoulder depth, measured directly, and it does not care what the grind is.

And then the technique that actually solves the question, which involves no hardware. Clamp a block of scrap on top of the work and drill through the block into it. A block does four things at once. It turns the depth into a protrusion below a known thickness, so you set it with a caliper on a bench rather than by eye on a spinning bit. It gives you a rigid reference plane instead of a compressible one, because the block is sacrificial and you can clamp it hard. It guides the bit square for the first part of the plunge, which is where a hand drill wanders. And it holds the surface fibres down at the rim so the hole starts clean. Measure the block with the same caliper you set the bit with and the only remaining error is the cone, which you now know the size of.

One last thing that is not about depth but comes up in the same breath. If the drill is tilted, what you lose is position rather than depth. Ten degrees off vertical shortens a 20 mm plunge to 19.70 mm, which is 0.3 mm and nobody cares. The same ten degrees through an 18 mm shelf board walks the hole's centre 3.17 mm sideways by the time it exits, which matters enormously. Depth and squareness are two problems and the scrap block happens to help with both.