Category: Knifemaking

  • Why the Angle Changes Between Heel and Point

    Why the Angle Changes Between Heel and Point

    The progressive grind is my answer to a conflict every hard-use blade contains.

    Everyone wants the sharpest blade possible. Cutting is mostly a question of edge angle: the more acute the edge, the less the material being cut or penetrated resists it.

    That same angle leaves less steel behind the edge. On a blade that will be abused, less steel means an edge that fails sooner.

    Anyone ordering a hard-use blade wants it to hold up, and geometry answers that request with material — you add it, or you leave it where it is. Which is why hard-use blades tend to be thick.

    The two demands pull against each other, and one is normally bought at the expense of the other. The same shape of problem as hardness against toughness.

    The usual settlements

    I am not the first maker to run into this, and there are several established answers.

    The most common is to stop pretending it can be solved and pick a side. On a hard-use blade, toughness wins. The angle opens, the edge survives, the cutting suffers. Most working blades are built this way.

    The second is the compound grind — an acute angle through the slicing section, a more obtuse one toward the narrow end of the blade, which is the part that starts the work and takes the first impact. It works, and it looks striking, which is not nothing. The name collides with compound bevel, which is a second angle sitting behind the edge — a different thing entirely. Here it means two angles along the length of the blade.

     

    Early Bladetricks F1K with a compound grind: an acute slicing section meeting a more obtuse angle toward the point
    An early F1K. Two angles, and a boundary between them.

     

    I used it for years. What bothered me was the step.

    Two angles meeting means a transition, and that transition sits in the middle of an edge that otherwise wants to be one continuous curve. The rest of the knife runs in organic lines. The edge has a seam in it.

    I wanted the same range of angles, arriving gradually.

    How a bevel is actually ground

    Every progressive grind I make happens on a belt grinder.

    The blade blank is held flat against a contact wheel — a rubber-covered wheel that the abrasive belt runs over — and supported on a work rest, a fixed horizontal bar below it. The blank is drawn across, the belt takes steel off, and the bevel appears where the two meet.

    My wheels are ten inches for most work, five or two inches on thinner blades. The machine is a Beaumont KMG-TX.

    Two things stay fixed through the whole pass: the work rest does not move, and the blade’s faces are never tilted out of plane. Whatever changes along the edge, it is not the setup.

    Three measurements decide the angle

    I grind chisel — one side only, the other face left flat. That makes the geometry easier to state than a double bevel, because there is only one bevel to state. Why a point is better finished as a plane than as a cone is a separate argument I have made elsewhere.

    Three measurements fix a chisel bevel completely:

    • the stock thickness — how thick the steel was to begin with
    • the land — the strip of steel left standing at the very edge before sharpening. I leave between 0.3 and 0.7 mm, depending on the knife.
    • the bevel height — how far up the blade the ground surface reaches

    Subtract the land from the stock and you have the steel removed at the edge. Spread that removal over the bevel height and you have a triangle. The angle of that triangle is the edge angle.

    tan θ = (stock − land) ÷ bevel height

    Nothing else enters it. Not the belt, not the pressure, not the hand.

    The wheel adds a little concavity, and less than most people expect. A bevel 26 mm tall, ground on a ten inch wheel, is hollowed by roughly two thirds of a millimetre. The bevel is near enough a plane. The hollow is a detail on top of it. And the hollow follows the bevel down. Where the bevel is 6.5 mm the wheel leaves six hundredths of a millimetre — nothing you could find with a straightedge.

     

    Diagram of a chisel grind hollowed by a 10-inch contact wheel: 0.69 mm of concavity across a 26 mm bevel, falling to almost nothing at the point
    The wheel is 254 mm across. Over 26 mm of bevel it leaves 0.69 mm. Over 6.5 mm it leaves nothing worth the name.

     

    One blade, worked through

    The F1K is one of my recurring models — 310 mm overall, a 168 mm edge, ground from 5 mm stock. It has carried both the compound grind and the progressive one, which makes it the clearest thing I own to compare against itself.

    Half a millimetre of land, a 26 mm bevel, 5 mm of stock. That gives 10° at the edge.

    Not a number I aim for. A number those three measurements produce.

    Where the progression comes from

    A progressive grind comes out of two things. One is drawn, the other is not.

    The one that is drawn is the blade itself. A blade is not a constant height. While it stays taller than the bevel, the bevel height is capped and the angle holds steady. Past the station where the blade becomes shorter than the bevel, the bevel has no choice but to follow the profile down — and the angle opens on its own.

    On the F1K that handover falls about 70 mm from the point.

    Distance from the point Edge angle
    70 mm and back toward the handle 10°
    40 mm 15°
    20 mm 25°
    10 mm 35°

     

    Progressive grind diagram: chisel bevel sections at 10, 15 and 35 degrees along a Bladetricks blade
    The bevel height falls with the blade. The angle opens with it.

    The profile writes that. I don’t.

    The variable that isn’t written down

    The profile is settled at the drawing stage and cannot be argued with afterwards.

    The second one is settled in the first second of the pass.

    The edge does not have to leave the plunge parallel to the work rest. Start it already tipped toward the point and the bevel height begins falling from the first millimetre, so the front of the blade progresses instead of sitting on its floor angle. Tip it further and the progression comes on earlier and harder. Tip it too far and the bevel runs out before the blade does, and the last centimetre ends up thicker than anything that section can use.

    Nothing on the machine holds that angle. The rest is bolted down, the faces stay flat, and then there is a blank, a hand, and a difference of two or three degrees that will not show on any drawing of the finished knife and will be obvious the first time somebody uses it.

    What sets it is the knife. What it is for, where it will be loaded, how it will be carried, who asked for it. Two blanks cut from the same profile, made for two different people, do not get the same tilt.

    I learned it on blades that did not survive the lesson.

    That part is not in the arithmetic, and it is the part I keep.

    What the progressive grind costs and what it buys

    The evidence is in the cross-section, and it is not subtle.

    Measured back from the edge Slicing section Near the point
    3 mm 1.0 mm 2.6 mm
    5 mm 1.35 mm 4.0 mm

    Two and a half times the material at three millimetres. Three times at five. Same blade, same pass, same half-millimetre of land at the apex.

    As pure geometry the gap is wider still — a 10° bevel and a 35° bevel differ by a factor of four. What pulls the finished figure below that is the land itself: the same half millimetre at both ends, which counts for far more on the thin section than on the thick one. Half a millimetre decides more than it has any right to.

     

    Second Bladetricks F1K with a progressive grind, the chisel bevel narrowing steadily from the plunge to the point
    Another F1K, same method. Follow the bevel from the plunge: it never stops narrowing.

     

    So the slicing section is as thin as I am willing to leave a hard-use blade. The section that leads into the work, that meets whatever is there first, that gets levered and twisted and stopped, is carrying three times the steel — exactly where a thin point breaks.

    A fixed angle cannot do both. Set it at 10° and the point is as fragile as the belly. Set it at what the point needs and the blade stops cutting.

    The same method, a different knife

    The Diplomat is a slimmer piece — 180 mm overall, 3 mm stock, a blade 21 mm tall where it leaves the handle, with a 15 mm bevel.

    Different numbers throughout. It still starts at 10°.

    And then the two diverge completely. The F1K holds that angle for the first 90 mm of its edge and spends the whole progression in the last third. The Diplomat has no flat section at all — its profile gives the bevel nowhere to stay constant, so it opens from the plunge onward.

     

    Bladetricks Diplomat, chisel ground on one side, the progressive grind narrowing with the blade toward the point
    The Diplomat. No stretch of this edge holds a constant angle.

     

    Same setup, same progressive grind, two different knives. The difference was decided when the profile was drawn, not at the machine.

    What it gives you

    • The whole slicing section at 10°. Every degree of thickening spent where the blade takes impact.
    • Easier maintenance than a compound grind. There is no boundary to preserve.
    • Easier sharpening. The angle changes, but gradually, and a stone follows it.
    • No steps. A step can catch, and in some situations that matters.
    • Cleaner lines.

    The progressive grind costs more time at the machine than a constant bevel. It is the only way I have found to put both properties on one edge without leaving a seam in it.

  • The File Test Said Something Was Wrong. Sugar, Salt and Copper Told Me What.

    The File Test Said Something Was Wrong. Sugar, Salt and Copper Told Me What.

    After the last hardening session, I ran the usual knifemaker’s check: a fresh file across the edge, straight out of the quench. It’s the oldest and most easily available test there is — hardened steel skates the file, soft steel lets it bite. That day the file bit more than it should have. Not on one blade: across the whole batch. Different steels, each at its proper temperature (my own numbers, cross-checked over the years against Knife Steel Nerds), each with its soak respected. And still, broadly, the hardness was neither what I expected nor what I’d accept.

    When several known steels fail the same way on the same day, the steel is not the suspect. Something upstream of all of them is, and finding it turned into a full kiln calibration, done with what the kitchen had to offer.

    Thinking it through, with a new tool

    I worked the problem with a tool I’ve been learning to use lately: an AI. Double-edged, that one. It reasons fast, it holds more metallurgy than my bookshelf, and it will also state a wrong number with the same confident face as a right one. Working with it is a skill in itself: you learn to interrogate the answers, keep what survives checking against real sources, and discard the rest without sentiment. Same as breaking in any tool, except this one talks back.

    After a long and, frankly, nervous back-and-forth, one explanation kept surviving every attack: an offset between the real temperature inside the chamber and the number on the controller. A thermocouple ages and drifts, and the controller reporting it has no idea.

    Microtemp TFC-900 kiln controller showing process value 1000 degrees Celsius and setpoint 1060
    The number on trial: mid-climb toward a stainless soak.

    That diagnosis fit what I know about the oven itself. Mine is an old pottery kiln, built the old way, with an enormous thermal mass. It takes around twelve hours to climb to the temperatures steels like AEB-L and 14C28N ask for (one of the reasons this workshop runs in batches: when the kiln finally gets there, it earns its keep on a full load, not a single blade). And once there it is precise and stable: it holds a soak flat and does the same thing tomorrow.

    That was exactly the disguise: the kiln reaches “800” today and “800” next week, perfectly repeatable, and repeatably wrong. The instability was never in the machine. It was in the sensor’s idea of the truth, and repeatability had hidden it for years.

    Fine. Hypothesis on the table. Now, how does a man verify it with no electronics background, in a remote spot, without ordering a reference thermocouple and waiting weeks? The way I like doing things anyway: with what’s on hand.

    Melting points don’t drift

    The trick is that pure substances change state at temperatures physics has already fixed. Aluminum melts at 660°C and copper at 1085°C anywhere; they’re actual calibration points of the international temperature scale. Put a known material in the kiln, note what the display reads at the exact moment it transforms, and you have one honest kiln calibration point. Enough points and you have the whole map of the lie.

    The kiln calibration kit came from the kitchen and the scrap bin:

    • Sugar: melts and caramelizes at ~186°C. Checks the tempering range.
    • Aluminum foil: pure aluminum, 660°C. Alloyed aluminum scrap is useless here; alloys melt low and vague. Foil is the pure, cheap form.
    • A magnet on a steel rod: steel stops being magnetic at the Curie point, ~770°C.
    • Coarse salt: 801°C, sitting dead center in the carbon steel hardening range.
    • A ring of copper pipe: 1085°C, covering stainless territory.

    Running the kiln calibration tests

    The method is step, soak, peek: set the kiln below the expected trip, let it stabilize, write down the stable reading before opening the door (once it’s open the display means nothing), quick look, close, raise five degrees, repeat. The melted markers are latches: once tripped, they stay tripped, so a glance answers “did we pass X during that soak” without ambiguity.

    Each marker taught its own lesson. The foil, crumpled loose, melts inside its own oxide skin and keeps its shape: the eye swears it never melted while the metal inside went liquid. The fix: compress the ball hard beforehand, and squeeze the retrieved ball with pliers. Round, smooth beads mean it was liquid. Soft plasticine means hot solid, not melted: try again.

    Two compressed aluminum foil balls from a kiln calibration test, one partially melted, one fully melted showing plier marks and round extruded beads of aluminum
    Top: a foil ball I believed had melted, it hadn’t, not fully. Bottom: one that did, squeezed with pliers; the round, smooth beads extruded through the grip marks are the proof of liquid. The eye lies; the pliers don’t.

    The magnet is a floor, not a target: non-magnetic proves you passed 770, nothing more, and once transformed the steel stays non-magnetic even as it cools, so there’s no latch to re-read later. The salt announced its trip by melting into a clear puddle and eating a hole through its stainless dish: molten salt is not a polite substance. The copper gives a warning first: shiny sweat beads on the black scale about twenty degrees before the ring collapses into a bead.

    Strip of spring steel corroded and pitted by molten salt spilled inside a heat treating kiln
    The salt ate through its foil dish, spilled, and etched this piece of spring steel on its way.

    What the map said

    At sugar temperature, the kiln was honest to within a degree. At aluminum: display 25 degrees high. At the Curie point: 25 again. At salt: around 20. At copper: past 40. The error wasn’t a constant to subtract once, it grew with temperature, so the fix is a kiln calibration curve through the measured points, not a single number scribbled on the controller.

    Melted copper puddle solidified next to a partially melted copper pipe ring on a red refractory dish, from a kiln calibration test
    The top of the map: copper survived ten minutes with the display reading 1125, and collapsed at 1130. True melting point: 1085. That gap, photographed.

    And there was the answer to the file test. Ask that kiln for a carbon steel’s 800°C and the steel soaks at roughly 775, below where its structure finishes transforming. The blade quenches soft, and the display never blinks. Years of the occasional stubborn batch, explained by forty invisible degrees.

    The retest

    Then the part that mattered: I hardened the whole batch again, every steel on its corrected setpoint from the kiln calibration curve. This time the file skated on every edge. From there, tempering: a first cycle shared by all my steels, 150°C for two hours, then a second cycle tailored to each steel and to the work each edge is headed for (a thin slicer and a chopper don’t want the same hardness, and the second temper is where that gets decided).

    Stainless knife blades sealed in tool wrap foil, labeled 14C28N, AEB-L and 420HC, ready for the kiln
    The retest, packed: 14C28N, AEB-L and 420HC in foil, headed back in on corrected setpoints.

    What runs differently now

    The marker test has joined the routine of this workshop: kiln calibration re-run yearly and after any thermocouple or controller change, because drift never announces itself. I’d argue it belongs in the routine of every workshop that hardens steel: it costs a few spoonfuls of sugar and salt, foil, a magnet and a slice of scrap pipe, and it answers a question no display can be trusted to answer about itself. Every setpoint here now comes off the measured curve, never raw from the book.

    Every hardening and temper goes into a pencil log next to the kiln. And the file test keeps its job at the door: it’s the one that raised its hand in the first place.

    Everything the process taught me ended up in writing. With the same AI at the bench, I built a 25-page workshop manual: the kiln calibration method and its measured curve, plus complete hardening and tempering references for every steel I regularly run, each number traced to its source: the steel producers’ own data sheets and Larrin Thomas’s published work. It lives next to the kiln, and it gets a new dated entry every time a number changes and a reason exists.

    The same lessons pushed one project further: I’m now working on the design of a new kiln, purpose-built for blades, with a fairly unconventional layout that should cut the hardening cycle from half a day to a fraction of it. That one deserves its own post when it’s ready.

    I keep trying to learn what actually happens in this workshop. This time it was the oven’s turn. That’s the craft as I understand it: understanding one more piece of the process each year, so that the knives that leave this bench are serious, professional, sensible tools. If you run a kiln, run the markers. If you buy blades, it’s a fair question how the maker knows his oven tells the truth.