The CNC Gear Keeping Machine in Kinco Servo Systems
A CNC gear keeping machine holds angular position while a gear or a gear-driven part is cut, ground, or checked. This page explains how the Kinco servo loop does that job, what it can and cannot hold, and how to tell whether your part belongs on this setup.
In this article
- 1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
Key takeaways
What a CNC gear keeping machine actually holds
A CNC gear keeping machine is a rotary workholding and indexing setup that keeps a gear blank at a commanded angular position while metal is removed. The cutting tool does the cutting. The keeping function belongs to the servo axis and its feedback device. On a Kinco servo drive, the position command comes from the CNC, the encoder reports actual angle, and the drive corrects the error thousands of times per second.
That split matters when you troubleshoot. If a tooth flank comes out with a lead error, the fixture or the machine geometry is usually at fault. If the error repeats once per revolution, look at the encoder coupling or the resolver mounting. If it wanders slowly, look at thermal drift in the drive and the ball screw, not at the cutter.
The device is not a gear hobbing machine in the usual sense. It does not generate a tooth profile from a hob and a synchronized feed. It holds a position so a separate operation can happen at that position: milling a keyway between teeth, drilling an oil hole on a pitch line, grinding a flank on a known index, or probing a spline to map its error.
Think of it as a high-resolution dividing head with a closed loop. The Kinco drive closes the loop on the rotary axis. The CNC tells it where to sit. Everything the operator cares about, from index error to repeatability, comes out of that closed loop.
- 1Position commandSent from the CNC to the Kinco drive over the fieldbus or pulse train.
- 2FeedbackEncoder or resolver on the rotary axis, not on the motor shaft alone.
- 3CorrectionDrive adjusts current to the motor until the error falls inside the window.
How the Kinco servo loop keeps angular position
A Kinco servo drive runs three loops stacked inside each other. The current loop runs fastest, in the tens of kilohertz. The velocity loop sits above it. The position loop sits on top and compares commanded angle to measured angle. The output of the position loop becomes the velocity command, and the velocity loop turns that into a torque command.
The stiffness of the position loop decides how hard the axis fights a disturbance. A milling cut on a gear flank is a periodic disturbance, once per tooth. If the loop is too soft, the blank rotates slightly under the cut and the flank goes wavy. If the loop is too stiff for the mechanics, the axis buzzes and the encoder sees vibration instead of position.
Tuning is a mechanical problem before it is an electronic one. A hollow fixture with a long overhang will not accept high gain no matter what the drive allows. A short, preloaded, large-diameter fixture will. We check the first resonance of the rotary assembly before touching the gain parameters, because that number sets the ceiling.
The drive also limits how fast the axis can follow a step. On a gear keeping machine, most moves are index moves: rotate 30 degrees, stop, cut, rotate again. The settling time after each index adds up. On a 40-tooth gear with a cut at every tooth, a 50 ms settle per index costs two seconds per part. That is real money on a 10,000-piece run.
- 1Current loopFastest loop; sets torque response and motor heating.
- 2Velocity loopDamps the axis and shapes how it approaches the target.
- 3Position loopHolds the commanded angle against cutting forces.
Where the setup holds well, and where it does not
The setup is comfortable with spur and helical gears in the module 0.5 mm to 6 mm range, with outside diameters up to a few hundred millimeters, held in a chuck or a mandrel. Small module work is the sweet spot because the cutting forces are light and the index accuracy dominates the result. That is where a well-tuned rotary axis earns its keep.
It struggles when the part is large and the tooth load is heavy. A module 10 gear in 4140 steel puts a bending moment on the fixture that a Ø400 mm rotary table can feel. The loop will hold position, but the fixture deflects, and deflection is not something the encoder sees. At that point you are better off with a dedicated gear machine that supports the blank on both sides.
It also struggles when the blank moves for reasons the loop cannot correct. Thermal growth is the common one. A gear cut dry will grow 20–40 μm on a 200 mm diameter as it heats, and the encoder reads the fixture, not the part. If the tolerance is ±0.005 mm, that drift eats the budget by itself. Coolant, a warm-up cycle, and letting the part stabilize before the final pass are the usual fixes.
Thin-walled ring gears are another bad fit. Clamping force distorts the ring, the loop holds the fixture perfectly, and the part springs back when unclamped. If you must hold a thin ring, use a low-pressure expanding mandrel and take the finishing pass with the lightest clamp you can get away with.
- 1Good fitModule 0.5–6 mm, light cuts, index-dominant accuracy.
- 2Marginal fitHeavy module, long overhang, dry cutting on tight tolerance.
- 3Bad fitThin rings, hot blanks, parts that need two-sided support.
Rigidity, feedback, and thermal drift in practice
Three things decide the accuracy you get: fixture rigidity, feedback placement, and thermal stability. The first two are mechanical choices you make before the first cut. The third is a process choice you manage during the run.
Feedback placement is the one people get wrong most often. If the encoder sits on the motor shaft and the reduction is a worm gearbox, the backlash in that gearbox never appears in the feedback. The drive thinks it is at position while the table lags. Put the encoder on the table, or use a dual-feedback setup, and the loop sees the real angle. On our rotary tables the feedback is on the table side for this reason.
Rigidity is about the shortest load path from the cutter to the machine base. A mandrel that clamps the bore close to the cut is stiffer than one that clamps at the far end. A fixture with a full-face seat is stiffer than three jack screws. None of this shows up in the drive parameters, and all of it shows up in the flank.
Thermal drift is manageable but not free. A 30-minute warm-up cycle before the first part, a stable coolant temperature, and a measuring pass on a known feature every hour will keep a tight-tolerance run inside ±0.005 mm. Skip those and the first part of the morning will not match the last part of the afternoon.
- 1Feedback on the tableCatches gearbox backlash the motor encoder cannot see.
- 2Short load pathClamp close to the cut; avoid long unsupported overhangs.
- 3Warm-up and re-checkRun 30 minutes, probe a known feature, then cut.
How the keeping function fits a real machining sequence
On a real part, the keeping function is one step in a longer sequence. Say a splined shaft needs six oil holes drilled between the spline teeth, each on a pitch line. The spline is cut first, then the shaft goes into the rotary fixture. The drive indexes to tooth one, the drill comes in, the drive indexes to tooth two, and so on. The whole operation takes one setup instead of six.
The value here is position repeatability, not position accuracy. If the index is off by 0.02 degrees but repeats within 0.002 degrees, the holes will be evenly spaced relative to each other and the spline, which is what the drawing actually calls for. Repeatability comes from the loop and the encoder. Accuracy comes from the initial calibration.
This is also why probing matters. If the drive indexes to a nominal position and the spline is 0.05 degrees off from nominal, every hole inherits that offset. A probe touch on the first tooth, followed by an offset shift in the program, removes the error before the first hole is drilled. On a 10,000-piece run that probe cycle costs a few seconds per part and saves a scrap bin.
The same logic applies to grinding a flank on a known index, or milling a keyway between teeth. The keeping function is a positioning service. It is only as good as the calibration that tells it where tooth one really is.
- 1Index and cutRotate to a commanded angle, stop, run the operation.
- 2Probe and offsetFind the real datum, shift the program, then run.
- 3Repeat, do not re-clampKeep the part in one fixture for the whole sequence.
Material and tolerance limits worth knowing
Aluminium and brass are the easy case. Light cutting forces, low thermal growth, and good chip evacuation mean the loop has little to fight. A 6061 gear blank in a well-tuned fixture will hold its index all day.
Stainless 303 and 304 are workable but demand more from the fixture. The material work-hardens, so a dwell at the index point will rub instead of cut and push the axis. Keep the feed engaged through the cut and do not let the tool sit at the commanded angle with the spindle turning.
Hardened steels above 45 HRC shift the problem to the cutting tool, not the loop. The loop still holds position, but the tool wears and the flank finish drifts. On a tight-tolerance run, plan a tool change interval based on flank measurement, not on time.
Plastics and POM are the opposite case. Cutting forces are tiny, so the loop is barely tested, but thermal growth is large and clamping marks are permanent. Use low clamp pressure, sharp tooling, and expect to hold ±0.02 mm on a plastic gear rather than ±0.005 mm.
- 1Easy6061, 2024, brass C36000 — light loads, stable.
- 2Workable303, 304, 17-4PH — plan for work hardening.
- 3HardAbove 45 HRC — tool wear drives the result.
How to judge a quote for gear keeping work
Ask three questions when you review a quote for a CNC gear keeping machine job. First, where is the feedback? If the answer is motor encoder only and there is a gearbox between the motor and the table, the quoted accuracy is not the accuracy you will get. Second, how is the part clamped? A quote that does not mention the fixture is a quote that has not thought about distortion.
Third, how is the first article checked? On gear work, the check is usually a lead and profile measurement on a gear tester, or a runout check on a CMM with a rotary axis. If the quote does not mention either, ask. A shop that cannot measure the index cannot hold it.
For the parts that fit, the setup is efficient. One fixture, one program, six or forty positions, no re-clamping between operations. The time saving is real and it shows up in the unit price on a run of a few hundred pieces or more.
For the parts that do not fit, pushing them onto this setup is a slow way to make scrap. A dedicated gear machine or a two-sided supported fixture will beat it on both accuracy and cycle time.
- 1Ask about feedbackTable-side encoder or motor-side only?
- 2Ask about clampingWhat fixture, what pressure, what distortion?
- 3Ask about first articleWhich measurement, on which machine?
When a CNC gear keeping machine fits, and when another process does
Match the part to the process before you quote the job.
| Part condition | Keeping machine fit | Better alternative | Reason |
|---|---|---|---|
| Module 0.5–3 mm spur gear, light cuts | Strong fit | None needed | Light load, index accuracy dominates |
| Module 4–6 mm helical, single setup | Good fit | None needed | Loop holds under moderate load |
| Module 8 mm and above in steel | Marginal | Dedicated gear hobbing machine | Fixture deflection exceeds loop authority |
| Thin-wall ring gear, tight bore | Poor fit | Expanding mandrel plus light finishing pass | Clamp distortion, not loop error |
| Hot blank, dry cut, ±0.005 mm | Poor fit | Wet cutting plus thermal stabilization | Encoder reads fixture, not part |
| Oil holes on a pitch line | Strong fit | None needed | Repeatable index, one setup |
| Flank grinding on a known index | Good fit | None needed | Needs calibration probe first |
| Long splined shaft, 4,000 mm | Conditional | Supported fixture, slow index | Overhang limits gain and speed |
The verdict
If your part is a small or medium module gear that needs repeated angular positioning in one setup, a CNC gear keeping machine on a Kinco servo axis is the right call. If it is a heavy module gear, a thin ring, or a hot blank held to ±0.005 mm, use a dedicated gear machine or a two-sided supported fixture instead.
Questions engineers ask about gear keeping setups
Can a Kinco servo hold a gear index without a gearbox?
Yes, if the load is light and the fixture is stiff. A direct-drive rotary axis removes gearbox backlash from the loop entirely, which is the main source of index error on a geared setup.
The tradeoff is torque. A direct-drive axis has less torque at low speed than a geared axis of the same frame size. For light cuts on small module gears that is fine. For heavy cuts it is not.
What index accuracy can I expect from this setup?
It depends on the encoder and the calibration, not on the drive brand. A table-side encoder with a calibration probe on the first tooth will typically repeat within a few arc-seconds on a stiff fixture.
Absolute accuracy is a separate number. The repeatability comes from the loop. The accuracy comes from how well you found the real datum before the run.
Why does my flank go wavy even when the index checks out?
Wavy flanks with a good index usually mean the axis is moving under the cut, not indexing wrong. Check fixture rigidity first, then check whether the position loop gain is high enough to resist the tooth-passing frequency.
A quick test is to reduce the depth of cut by half and re-measure. If the waviness drops, the loop was losing the fight against the cutting force.
Does the keeping machine replace a gear hobbing machine?
No. A hobbing machine generates the tooth profile from a synchronized hob and feed. A keeping machine holds an angular position for a separate operation.
They do different jobs. Some shops run both, with the hob cutting the teeth and the keeping setup doing the secondary features in one fixture.
How do I control thermal drift on a tight-tolerance gear run?
Warm up the machine for 30 minutes, hold the coolant temperature steady, and probe a known feature every hour. If the probe drifts, shift the program offset before the next part.
On a 200 mm diameter part, a 20 μm growth is normal for a dry cut. Wet cutting and a stabilization pause before the final pass keep it inside ±0.005 mm.
What module range suits this setup best?
Module 0.5 mm to 6 mm is the comfortable range. Below 0.5 mm the tool becomes the limit. Above 6 mm the cutting force starts to dominate the fixture.
That range covers most small gearbox, instrument, and servo-driven actuator work.
Send us the gear drawing
Upload the part and we will tell you whether the keeping setup fits, what fixture it needs, and how we will check the first article. Quotation and DFM feedback within 12 hours.
12-hour quote100% inspectionNDA on request
More gear and servo machining notes
We publish setup notes, tooling trials and inspection data from the factory floor.
