RPM to kHz: Spindle Speed and Tooth-Passing Frequency

A Spindle Speed Is Also a Cutting Frequency

When a milling cut starts to howl and the finished wall comes off the machine covered in fine ripples, the fix is almost never more rigidity and almost always a different spindle speed. Getting there means leaving the world of revolutions per minute and thinking of the cut as a periodic excitation: so many impacts per second, hammering a tool and a workpiece that have natural frequencies of their own. Kilohertz is the axis that conversation happens on, because the interesting numbers — tooth impacts, structural modes, the tone a microphone picks up — sit in that range.

Conversion factor: divide by 60 000 — 1 RPM = 0.000016666667 kHz, since there are 60 seconds in a minute and 1 000 hertz in a kilohertz. A spindle at 12 000 RPM turns at 0.2 kHz; put a four-flute cutter in it and the tooth-passing frequency is four times that, 0.8 kHz, because TPF = RPM × teeth ÷ 60.

What Actually Excites the Tool

Flute Count Multiplies the Excitation

The shaft turns once, but a four-flute end mill delivers four cutting impacts in that time. Swapping a two-flute roughing cutter for a four-flute finisher at the same programmed speed doubles the drive frequency without changing a single line of the program.

The Regenerative Loop Behind Chatter

Each tooth cuts a surface the previous tooth left behind. When the tool's vibration and the wave already on that surface fall out of phase, chip thickness varies more each pass and the oscillation feeds itself. Speed decides that phase relationship, which is why it is the lever that works.

Why the Kilohertz Scale Fits This Job

Slender end mills and shrink-fit holders commonly show dominant modes somewhere between 500 Hz and 5 kHz. Written in hertz the numbers are long; written in kilohertz a lobe chart, a modal test result and an audio spectrum all share one readable axis.

Spindle Rate Alone Is Rarely the Answer

The bare conversion gives revolutions per second, useful for runout and balance work. For anything to do with cutting stability it is only the starting point: the flute count has to be applied before the number can be compared with a measured tone.

Working a Chatter Problem at the Machine

A practical order of operations when a cut is unstable and the speed override is the only thing you are willing to touch.

Enter the programmed spindle speed

Put the S value from the program into the RPM field. It starts at 1 and recalculates on each keystroke, so trying 9500, 9800 and 10200 in succession costs nothing. Spaces in a figure such as 12 000 are ignored, and a comma is accepted where a decimal point is meant.

Multiply by the number of effective teeth

The kilohertz reading is the shaft rate; multiply it by the flute count for tooth-passing frequency. Count only the teeth actually engaged — on a staggered or serrated cutter, or one with a chipped edge, the effective number is not the one on the catalogue page.

Compare it with the dominant mode

Take the tool-and-holder natural frequency, from a tap test or from the tone the cut is producing, and divide it by the flute count and then by successive whole numbers. Each result is a candidate speed sitting near the top of a stability lobe, where deeper cuts survive.

Go back the other way from a measured tone

With a frequency already in hand from a microphone or accelerometer, the swap arrows reverse the pair, or you can type straight into the second field — both are editable. The copy button lifts the bare digits so a candidate speed can be dropped into a cutting-data sheet without stripping the unit out first.

Tooth-Passing Frequency Across the Speed Range

The columns below show why the flute count matters more than the speed dial does. A two-flute cutter has to be spun to 24 000 RPM before it excites anything at 0.8 kHz; a four-flute cutter reaches the same point at 12 000 RPM, and that difference alone can move a stable cut into an unstable one.

Programmed speed	Shaft rate	Two-flute tooth passing	Four-flute tooth passing
3 000 RPM	0.05 kHz	0.1 kHz	0.2 kHz
6 000 RPM	0.1 kHz	0.2 kHz	0.4 kHz
8 000 RPM	0.1333 kHz	0.2667 kHz	0.5333 kHz
10 000 RPM	0.1667 kHz	0.3333 kHz	0.6667 kHz
12 000 RPM	0.2 kHz	0.4 kHz	0.8 kHz
18 000 RPM	0.3 kHz	0.6 kHz	1.2 kHz
24 000 RPM	0.4 kHz	0.8 kHz	1.6 kHz

Notice how far below the audible tone these figures stay. A cut squealing at 1.5 kHz is not being driven at 1.5 kHz by anything in the table — the tool is ringing near its own mode and the tooth impacts are merely topping it up on every pass. That gap between excitation and response is the whole reason a lobe chart exists.

Step Through the Candidate Lobe Speeds

A lobe calculation throws out a handful of speeds to try; entering each one in turn shows immediately which of them land near a mode and which sit clear of it.

Measured Tone Back to a Spindle Value

Reversing the direction converts a peak captured in the enclosure into the shaft speed that would match it, which quickly rules a spindle harmonic in or out.

Digits Ready for a Cutting-Data Sheet

Copying returns the value with no unit and no thousands spacing, so a chosen speed drops into a tooling spreadsheet as a number rather than as text to be cleaned up.

Kilohertz or Plain Hertz, Whichever Suits

Either dropdown can be searched and changed, so a tap-test result quoted in hertz and a lobe chart drawn in kilohertz can be reconciled on the same page.

Chatter and Spindle Frequency Questions

Why is tooth-passing frequency, not spindle rate, the number that matters?

Because the structure is disturbed once per tooth, not once per revolution. Tooth-passing frequency is RPM multiplied by the tooth count and divided by 60, so a three-flute cutter at 10 000 RPM excites at 500 Hz while a six-flute cutter at the same speed excites at 1 kHz. Everything in stability analysis is referenced to that rate: the lobes are spaced by it, the phase between one tooth's wave and the next depends on it, and the sidebands in a measured spectrum are separated by it. Shaft rate on its own only tells you about runout, balance and once-per-revolution marks.

How does a stability lobe diagram point at a usable speed?

A lobe chart plots depth of cut against spindle speed, and its scalloped upper boundary marks where a cut stops being stable. The peaks of those scallops sit where the tooth-passing frequency divides evenly into the dominant natural frequency, because each tooth then meets the surface wave in phase and chip thickness stops growing. For a mode at 1 200 Hz with a four-flute cutter, the candidates are 60 × 1200 ÷ 4 = 18 000 RPM, then 9 000, 6 000 and 4 500 RPM as the divisor increases. The lobes get narrow and crowded at low speed, which is why the technique earns its keep on high-speed machining centres rather than on a heavy roughing mill.

Why does the noise not match the spindle frequency at all?

Because what you hear is the tool assembly ringing at close to its own natural frequency, not the rate at which it is being hit. Chatter is self-excited: the impacts supply energy, the structure decides the pitch. That is why the tone stays stubbornly around the same value while the speed is adjusted, and why it usually lands well above every entry in the table on this page. Forced vibration behaves differently — it sits exactly on tooth-passing frequency or a whole multiple of it and rises and falls with the speed — so the two are straightforward to tell apart with an override dial and an attentive ear.

What sample rate does a chatter microphone or accelerometer need?

Cover the range where the modes live, which for typical end mills and holders means up to about 5 kHz, and allow the usual analyser margin of 2.56 times the highest frequency of interest. That works out at roughly 12.8 kHz of sampling, so an ordinary 44.1 or 48 kHz audio input has ample headroom for the job. Mounting is the tighter constraint: a magnet-mounted accelerometer starts losing fidelity somewhere around 2 kHz, while a stud-mounted one keeps its response well past 10 kHz. A microphone in the enclosure sidesteps mounting entirely but picks up coolant, chip evacuation and the spindle drive along with the cut.

Can the marks on the finished surface be read back as a frequency?

They can, and it is a good trick when no instrumentation is available. Count the ripples left around one revolution of the cutter and multiply by the shaft rate in hertz. Twelve thousand RPM is 200 revolutions per second, so a wall carrying seven or eight waves per revolution was vibrating near 1.5 kHz. On a 12 mm cutter those 7.5 waves work out at about 5 mm apart along a circumference of roughly 37.7 mm. The fractional part of that wave count is the useful bit: a whole number means each tooth is tracking the previous wave exactly and the cut is stable, while something near a half is the worst possible phase and the classic fingerprint of regenerative chatter.

RPM to Kilohertz