Table of Contents

Known Limits of the Load Leaves

What Reading the Load Leaves does not cover, case by case: the readings the engine is known to get wrong, the assumptions the simulated spindle speed rests on, and the steps, inputs, programs and versions on which a leaf reads otherwise than it appears to. How Machining State reads a cut's load is on that page.

  • The engine's known misreads colour a step as they read it. The load leaves take the forces and ratios at their word. Cases have shown these:
    • A rest-machining cutter meeting a corner head-on — a slot's end, an inside corner — or running its whole edge length along a thin layer read its forces and removal 2 to 7 times high: a Stress overload or a spindle overload there, at a corner or along a thin wall, is suspect.
    • A part set up turned 120° about a diagonal — a side plate standing in a vise, the tool axis along the part's X — read no force at all on its drilling and circle milling: those cuts are Normal cutting, green, however hard they are, and a cut found no contact for is an Air cut.
    • Two readings the tree does not use, but a copy may: a ball-nose cutter going up or down a slope read its material removal rate (Mrr_mm3ds) 8 to 19 times high, and a narrow slot its cutting depth (CuttingDepth_mm) at twice each layer's; the forces were right in both.
  • The spindle speed under load is a simulation on assumptions, checked against no machine. The two curves are taken as hard limits of the drive, though they are thermal ratings a drive can exceed for a moment, so the simulated spindle as a rule starts to slow at a lower load than the machine's and slows faster. The rotor inertia is derived from the capability's Full-Power Run-Up Duration, on its Thermal / Energy node, shown while Show Physics Options is on (Spindle Capability), read as 1 s where it is left empty: whether that makes the simulated spindle slow too early or too late depends on the machine. The simulation is checked for durations from 0.01 s up; the field takes shorter ones too, and with one the speed can lose accuracy where it settles on the torque curve, and a step's speed can fail to compute, which the play reports once (Spindle-SpeedUnderLoad--Faulted). Besides:
    • The simulated speed is not fed back into the cut, in Feedback Mode as well: a step's force, its spindle ratios, its stress ratio and its breakage probability stay the values at the programmed speed, so the further the spindle has slowed, the further the real cut is from the simulated one.
    • A step past its end point with neither spindle ratio over 100 % takes no spindle leaf, even one that has slowed all the way to the floor: the two sustained leaves need the step to be in a run, and the ratios are read at the programmed speed, where the thicker chip of a slowed spindle does not show. Spindle Speed Deviation still says how much it has slowed; to colour such steps, copy the tree and add a leaf on SpindleSlowdownEndpointRatio ≥ 1 (Copying one into the project's trees).
    • The simulation reads the two curves on the programmed speed's side of the gear shift speed, while the denominators of the two spindle ratios read the whole curves. The two differ at some speeds in a capability whose curves do not carry a point on each side of the gear shift speed next to it, whose power curve does not start at 0 rpm, or whose gear shift box is ticked with the speed left at 0: a steady cut can then slow with both ratios under 100 % — no spindle leaf takes it, and Spindle Speed Deviation shows it — or hold its speed with the torque ratio over 100 %, a Spindle torque overload showing a deviation of 0 %. Where the shortest-duration torque curve has no point at all on the programmed speed's side, the speed under load cannot be computed on the steps at that speed: the speed, the deviation and the endpoint ratio have no value there, a step over a limit stays yellow, and the play says so once, with the reason (Spindle-SpeedUnderLoad--Faulted). The shipped capability files have no such case; how to place the points is under Spindle Capability.
    • The first step of a torque run that starts anew is yellow, whatever the spindle's speed by then, unless it falls inside a power run past its first step with the spindle at its end point: such a step is a Spindle power sustained overload.
    • A deviation of 99 % is the simulation's floor, not a speed a machine settles at. A spindle that reached the floor is back at the programmed speed in one step once the load allows, sooner than one that stayed above the floor, which recovers step by step.
    • While the speed recovers, the shown speed is about a step late: it is the step's mean.
    • The simulation goes step by step. Under a Machining Motion Resolution other than Feed Per Cycle — Scaled Feed Per Cycle, Feed Per Tooth, Fixed Pace (Machining Motion Resolution) — a step is not one revolution, and the one step of margin, the one step by which the chip follows the speed, the time the power curve's mean is taken over and the mean the speed is shown as are all that step's length. The longer the step, the later the thicker chip adds to the load, the later a step can turn orange, and the less the lowest speed inside the step shows.
    • Every program is read as a feed per minute. Under a feed per revolution the feed per tooth does not grow as the spindle slows, and the steps turn orange too early.
    • A change of the programmed speed is taken as reached at once: the run-up and the run-down are not simulated.
    • A CSV program whose speed column is a measured speed starts again from that speed wherever two steps in a row differ by 0.6 rpm or more, which leaves the slowing of one step alone; where they differ by less, the slowing the measurement already holds is counted again.
    • The dip inside a tooth is read from the torque at the angles one revolution is divided into, 36 by default, each standing for its share of the turn: the peak comes out lower than a finer division gives, and the dip with it. A step on which a flute's edge stays in contact for less than four of those divisions of the turn — a light cut, an entry or an exit — is divided into 360 when that count is below 360 and EnableNarrowArcFineDivision is on, as by default, so its short torque peak is read near where it is. A step whose torque by angle cannot be read — a recorded play that did not keep it — is read as a steady cut, with no dip inside a tooth and no note.
    • The friction and the windage of the turning spindle are not part of the load.
    • The rotor inertia is the rotor's alone: the tool and the holder in the spindle are not added to it, so with a heavy tool the simulated spindle loses speed faster than the machine's.
    • A spindle with a gear shift is simulated with one rotor inertia, derived from its whole shortest-duration torque curve, both ranges together, while the inertia that counts at the spindle differs by range: a motor's counts by the square of each range's own ratio. With a duration timed as the field asks, converted where the drive accelerated with more than the shortest-duration curve (Full-power run-up duration), that one inertia is not below the high range's: below the gear shift speed the curve holds the low range's torque, more than the high range gives there. So in the high range the simulated spindle loses speed no faster than the machine's. In the low range it loses speed faster than the machine's where the run-up was timed from the low range straight into the high one without stopping, the shift itself taking little time; timed in the high range alone, as a machine that selects its range at standstill runs up, or with a stop to shift, which way the low range errs depends on the machine.
    • Where the rotor inertia cannot be derived — the shortest-duration torque curve reaches no speed above zero, holds a torque that is not above zero, or holds a speed or a torque that is not a finite number (only the capability's XML, a script or a REST client sets one), or the Full-Power Run-Up Duration on Thermal / Energy is not a finite number above zero — 0 typed into the field, or a negative or non-finite value set in the capability's XML, by a script or by a REST client; a NaN shows as an empty field, but is not read as 1 s — the speed and the deviation have no value, and an interrupted cut whose peak alone exceeds the torque curve cannot be judged. The endpoint ratio then reads infinity on a step whose mean cutting power exceeds what the spindle gives at the programmed speed, the lower of what the two curves give there, so such a step is a sustained overload once past the first step of its run, its lane without a number, the deviation having none, as Spindle torque overload's lane is on such a step; any other step that cuts reads the end point itself. The play says so once (Spindle-Inertia--Unreadable), and once more after the spindle capability is changed or corrected. The endpoint ratio of such a step is read against the spindle capability as it is now, not as it was played: once the curve or the duration is corrected, the step reads NaN and falls from its sustained overload to a yellow leaf until a recompute gives it a speed (Recomputing a Finished Strip). A torque curve without points is not this case: it leaves no spindle limits to read at any speed, which the play reports as Spindle-Limits--Unreadable, and the play is one without a spindle capability to read against (Without a spindle capability).
  • A step on a very short move can read a spindle ratio over its limit. The first step after a zigzag ramp turns back in a roughing program, a move of 0.011 mm, read a power ratio of 107 % beside a torque ratio of 97 %: a lone Spindle power overload among normal cuts. Read that step's Move Length, its engagement and the steps beside it before taking it for an overload; HiNC filters no such step out.
  • A cut the physics does not compute while its switch is on has no ratio, so it is Load not evaluated, and no note says why — the physics line reads the switch alone, and the tools line names only tools whose cuts the physics ran on. That happens without a physics licence, on a tool the physics does not run on, such as one that is no milling cutter, and while the spindle stands. Play-Physics--None warns when no cut of a play got a physics result, as when the spindle never turned.
  • A thermal stress the physics did not evaluate reads 0 %, and no note says so: the thermal stress ratio is 0 % on a step played with the physics off, for a tool that is not a milling cutter, for a cutter whose Flute Profile does not agree with itself, for a cutter without a flute material, and on a step whose thermal model has not run since the tool change, its inputs missing or the model unavailable. Such a cut never falls into Thermal stress overload and is judged by the other leaves alone. A solid-end cutter without a flute material has no stress ratio either and is Load not evaluated; an insert-end cutter without one reads its stress ratio off its body — AlloySteel42CrMo unless the cutter names another material —, so its cuts can be Normal cutting whatever heat their edge took. Set the flute material on the cutter's Material node (Cutter).
  • A spindle capability nobody chose. A project that never set a spindle capability has none, and its plays are read without the five spindle leaves (Without a spindle capability). So does a project file saved by an earlier version of HiNC that holds, with no capability file, exactly the values of the TMV-720A-STD-8000RPM spindle under that name: such a file cannot tell values pasted in on purpose from the ones nobody chose. Load your machine's capability, or for a TMV-720A the shipped TMV-720A-STD-8000RPM.default.SpindleCapability, which the project then keeps as a file of its own; a capability installed on the project, pasted or applied in XML Mode too, is saved so that it is read (Spindle Capability).
  • An old copy of a cutter material. A project keeps its own copy of each cutter material it loaded; when a later HiNC ships the material with other strength values, the copy keeps the old ones and the stress ratio follows them. Reload the shipped material on the cutter's Material node to take the current values (Cutter), then recompute or replay: until then the steps keep the stress ratio they were played with, after an edit made in place as well.
  • A tool change builds a step of its own when it is the strip's first step: the step lasts the tool-change time, reads the modal rapid state and falls into Air cut or Rapid, or is a cut when the tool touches. No field tells it from the moves around it. An M06 in the middle of a program builds none, with the deflection transformation on as well (EnableDeflectionTransformation, off by default, on with Feedback Mode): the transformation shifts each step's sweep by the tip deflection the previous step's largest force bends the tool to, so the stock and the geometry difference carry the deflected tool, while the toolpath and the step's feed and feed per tooth stay as programmed. The first step after a change to another tool is not shifted: the new tool starts undeflected, with none of the previous tool's deflection.
  • A CSV program has no rapid, so its steps are Air cut or cuts, and a review of a recorded play judges no rapid cut and neither spindle sustained overload: their notes say so, and no more.
  • Collision detection switched on without a collision licence detects nothing, and the note Collision detection was on without a collision licence says so — a play with no machining chain to detect against takes the same note. A rapid plunge before the spindle has turned is then a cut, neither a collision nor a rapid cut (Reading Rapid cut).
  • A step whose property read fails falls into no leaf, and the session says so once for the tree (StepColorTree-Evaluate--Faulted).
  • An earlier HiNC opening a project saved with a built-in tree as its color index does not know the key: its play ends with the warning Play-InspectingKey--Unresolved.
  • The three overload-revolution properties of the Science view, Spindle Torque Overload Revolutions, Spindle Power Overload Revolutions and Stress Overload Revolutions, start from a colour range of 0 to 10 revolutions, the two step ordinals, Spindle Torque Overload Step Ordinal and Spindle Power Overload Step Ordinal, from 0 to 10 steps, and Spindle Speed Deviation from 0 % to 20 %, where every other number starts from 0 to 1, which Colors shows as 0 % to 100 % for a ratio — Spindle Speed Under Load, in rpm, among them, until its range is set; a project that saved a range for one keeps it, 0 to 1 included: set it on Colors.

See Also