From the machine you own to a depth of cut you can defend
Four steps. The first three take about twenty seconds. The fourth is the one that turns an estimate into a measurement.
Tell it what you own
Pick your machine and your spindle from the list — 34 machines and 32 spindles, including every trim router people actually bolt on. Each machine carries its architecture, its published feed ceiling where one exists, and an honest estimate where it does not.
Pick the material and the cutter
21 materials and 10 cutter geometries, or choose one of 47 named products from seven suppliers and get that maker’s own chipload figures instead of generic ones. Set your stickout, because deflection scales with its cube.
Read what is binding
Feedbench scores nine constraints at once and sorts them worst first. You get a feed, a speed and a depth — and next to them, the name of the thing that stopped the numbers going higher, what it ran out of, and what to do about it on this architecture.
Cut the test grid and find the truth
Generate a G-code program that cuts feeds against depths in scrap. Run it. The cell where the finish falls apart is your machine’s real limit, and it beats anybody’s model, including this one.
“What is the worst number here” and “what set my depth” are not the same question
Feedbench shrinks your depth of cut until every depth-sensitive limit is satisfied. That means those limits all finish sitting at about 100% — by construction, not by coincidence.
Meanwhile a limit that depth cannot move — your spindle will not spin slower, your machine will not feed faster — sits wherever it sits, often far higher. Sort the list worst-first and that one always wins, and the constraint that actually chose your depth never appears at the top.
So Feedbench answers both. The headline names the hardest limit in your setup. A second panel names the one that set the depth of cut, and says why it is not the headline.
Shapeoko 3, Makita trim router, 12.7 mm upcut slotting hardwood. The headline is the machine feed rate at 104% — it wants 2,600 mm/min and the machine tops out at 2,500. But feed rate is fixed. What actually drove the depth down to 2.49 mm was spindle power, sitting at exactly 100% with 288 W needed against 288 W available.
Act on the headline and you go shopping for a faster machine. Act on the real answer and you take a lighter cut, or fit a stronger spindle.
Some answers are worse than no answer
Feedbench will tell you that a combination is not viable rather than hand you a number that technically satisfies the arithmetic. A 12.7 mm cutter slotting brass on a light hobby machine has no safe starting depth, and the model says so instead of recommending a whisker-thin pass that would be nonsense in practice.
It also refuses to model ferrous metals. Steel with a carbide cutter wants somewhere between 1,500 and 3,000 RPM on a quarter-inch tool. The lowest spindle floor anywhere in the dataset is 3,500. There is no honest answer to give, so none is given.
And where a value is inferred rather than published — a feed ceiling the vendor never stated, a trim router’s output power, a loop stiffness nobody measures — the interface says so on the figure itself, every time.