Resolving Yasnac J300 G107 Corner Feed Rate Slowdown on Matsuura

Resolving Yasnac J300 G107 Corner Feed Rate Slowdown on Matsuura MC660VG

The Yasnac J300 control fitted to the Matsuura MC660VG vertical machining center protects arc-segmented motion (G02/G03) from servo-lag-induced dimensional error by automatically clamping the programmed feedrate at corners. When the function is active, the operator display flashes G107 and the axis feedrate is reduced to a fixed floor (commonly 50 mm/min) regardless of the F-word in the part program. On small pockets, bores, and free-form toolpaths this slowdown inflates a 20-minute Fusion 360 simulation into a 90+ minute cut, which is the symptom operators first notice. This reference documents the root cause, the parameter set that governs the function, the safe way to disable or retune it, and the commissioning verification used to confirm recovered cycle time without losing part accuracy.

1. Problem Details: G107 Behavior on Yasnac J300

Yasnac (Yaskawa Electric) J-series controls implement what the documentation calls the Automatic Arc-Corner Feedrate Designation Function. It is engineered around a simple observation: any circular block (G02/G03) entered at a non-tangential feed angle, or chained with another arc or linear block, requires the linear axes to constantly accelerate and decelerate. If the programmed F-value forces more acceleration than the servo drive can deliver, the axis falls behind its commanded position (servo lag). The visible consequence is a "shallowed" toolpath — arcs that look correct in CAM come out undersized on the workpiece.

To eliminate that error mode, the J300 reduces the F-word to a safe ceiling whenever an arc-corner is detected. The operator sees two on-screen cues:

On a typical MC660VG the clamped corner feedrate resolves to roughly 50 mm/min for small-radius geometry because the parameter tolerance is set conservatively. The CAM estimator in Fusion 360 does not know about this clamp, which is why predicted and actual cycle times diverge dramatically.

2. Root Cause: Servo-Lag Compensation, Not a Fault

G107 is not an alarm. The control is not detecting an error — it is preemptively limiting feedrate to prevent one. The underlying physics is the same constraint seen on every closed-loop CNC: the axis drive has a finite torque/acceleration envelope, and the integral of commanded acceleration versus delivered acceleration produces position error. On the J300 the relevant envelope is defined by:

Maximum corner feedrate F_c = sqrt(2 · a_max · R · ε)

where a_max is the axis acceleration limit, R is the corner radius, and ε is the permissible arc-corner tolerance. As R shrinks (tighter corners), F_c falls below the programmed F. The control has no look-ahead predictive override (the optional HON/GON advanced look-ahead packages are not installed on stock J300 builds for the MC660VG), so it cannot pre-shape the velocity profile. The result is the G107 clamp to a conservative fixed value.

2.1 Why CAM time estimates disagree

Fusion 360 (and every modern CAM package) assumes a constant F-value across arcs unless it can detect a radius below a user-defined threshold. If the corner radius drops below the CAM "tolerance floor" the post-processor should insert a reduced F-word on the affected block. Many post-processors for vintage Yasnac controls do not perform this insertion, so the part program enters the J300 at the high F-value and the control takes over.

2.2 Why it gets worse on a small pocket

A 100 mm circular pocket programmed as a single helical ramp or as chained G02 blocks at incremental I/J values contains no large linear blocks to amortize the feedrate change. The clamp is therefore engaged continuously from the start of the arc to the end of the cut. A square pocket, by contrast, contains four linear cuts where the clamp is not active — only the four 90° transitions are, and each transition completes in milliseconds at the programmed feedrate.

3. Parameter Map Governing G107 / G108

Three parameters interact to control the function on Yasnac J300. They are accessible from the parameter page after entering the machine-builder password mode on the Matsuura MC660VG.

The function is only active when both conditions of the J300 specification are satisfied:

1. pm4031 D2 = 1, and
2. The active unit mode has a non-zero tolerance — pm0416 ≥ 1 in G21 (mm) or pm0417 ≥ 1 in G20 (inch).

If either condition is false, the clamp is bypassed and the display shows G108.

4. Solution: Choose Disable vs. Retune

There are two field-proven remedies. The right choice depends on whether the part program contains features whose dimensional accuracy depends on the clamp (small radii, tight bores, free-form contours) or only open geometry where servo lag would not produce a measurable defect.

4.1 Option A — Disable G107 entirely (recommended for time-critical roughing)

Use this when the operator has confirmed via test cuts that the machine holds dimension without the clamp, or when the part program is dominated by open arcs where any dimensional drift falls inside the print tolerance.

1. Place the control in MDI / Parameter mode.
2. Navigate to parameter pm4031.
3. Change bit D2 to 0. The byte value typically drops by 4 (e.g., 0x05 → 0x01).
4. Optionally set pm0416 (mm) and pm0417 (inch) to 0. With both bits cleared and tolerance at zero, the function cannot engage even if a future parameter write accidentally re-enables bit D2.
5. Cycle control power or perform a parameter-write reset as documented in the Matsuura parameter sheet.
6. Confirm on the modal display: G108 should now be lit in place of G107 when running an arc block.

4.2 Option B — Retune the tolerance upward (recommended for finishing operations)

If the part program requires the clamp for accuracy (small radii, precision bores) but the operator wants shorter cycle times, increase the tolerance so the control computes a more permissive corner feedrate. The function then stays on (G107 still displayed) but the clamp value rises with the tolerance setting.

1. Confirm G107 is desired — keep pm4031 D2 = 1.
2. Increase pm0416 in mm mode or pm0417 in inch mode to the largest value that still meets part print tolerance. A typical safe ceiling on the MC660VG is 0.05 – 0.10 mm for roughing and 0.01 – 0.02 mm for finishing.
3. Run a calibration cut on a sacrificial part: a 25 mm boss with a 5 mm corner radius is a representative diagnostic feature.
4. Measure the boss diameter and corner radius with calipers or a pin gauge. Compare to the programmed value.
5. If dimensions fall inside tolerance, leave the new value. If they fall outside, halve the tolerance and re-test.

4.3 Why both pm0416 and pm0417 matter

The control only uses the parameter that matches the active unit mode. A shop that programs exclusively in mm can leave pm0417 at zero without effect, and vice versa. Set both to your intended working value so a G20/G21 swap inside a long program does not silently change behavior.

5. Step-by-Step Commissioning Procedure

Follow this procedure on a cold machine with the door closed, in MDI/Parameter mode, with the control in a safe state (no program running, all axes referenced).

1. Document baseline. Record pm4031, pm0416, pm0417 values, the J300 software version on the parameter banner, and the Yasnac servo-pack firmware label on the drives.
2. Run a benchmark cut. Program a simple circular pocket, 100 mm diameter, 5 mm depth, F600, single-pass G02 with I/J incremental entry. Time the cut from cycle-start to M30.
3. Verify G107 behavior. Watch the modal display during the arc block. Confirm whether G107 is lit and the actual feedrate indicator drops to the clamped value.
4. Apply Option A or Option B as decided above.
5. Re-run the benchmark. Record the new cycle time. A typical MC660VG recovers from ~90 min to ~20 min on a 100 mm round pocket after disabling G107.
6. Dimensional check. Measure the pocket diameter at four positions with bore gauges or a coordinate measuring machine. Verify against print tolerance.
7. Lock the parameters. Engage the machine-builder parameter-lock mode per the Matsuura setup sheet so production operators cannot inadvertently re-enable the function.
8. Update the post-processor. If the symptom recurs on different parts, modify the Fusion 360 (or current CAM) post-processor to insert reduced F-words on arcs below the corner-radius threshold, so the part program never relies on the J300 clamp.

6. Verification

After the parameter change, three checks confirm the function is behaving as intended.

7. Edge Cases and Field Caveats

The following caveats have surfaced from shop-floor experience with Yasnac J300 controls on Matsuura MC660VG-class machines.

7.1 Inside corners on endmills chatter at high feed

When the G107 clamp is removed, the linear feed into a tight inside corner may exceed what the tool — not the control — can absorb. An endmill in a 90° internal corner sees roughly 1.4× the radial engagement of an open cut, and the chip load rises proportionally. If chatter or tool breakage appears after disabling G107, the answer is a CAM-side feed reduction on the corner block, not a re-enable of the J300 clamp.

7.2 Helical interpolation and small-diameter bores

Helical bores below ~6 mm diameter show the worst behavior of any feature: every arc block is short, every transition is a corner, and the clamp dominates cycle time. Disabling G107 is almost always the correct call here, with the F-word reduced at the CAM level to keep dimensional accuracy on the bore wall.

7.3 Parameter persistence after battery replacement

Yasnac J300 parameters are stored in CMOS backed by the control's internal battery. After battery replacement, parameters revert to factory defaults — which on the Matsuura MC660VG means G107 is enabled with the conservative tolerance. Reapply Option A or Option B and verify immediately after any battery service.

7.4 Conflict with macro-based feed overrides

Some shops wrap their post-processors with custom macro logic (M-calls or G65 blocks) that overrides the F-word. The G107 clamp is applied after the F-word is resolved, including after macro overrides. If the part program reports a different F-word than what the control displays on the modal line during the arc, the macro is fine — the clamp is independent.

7.5 Yasnac MX3 / J300M / I80 differences

The G107/G108 logic is consistent across Yasnac J-series and MX-series controls, but the parameter numbering and bit assignments may differ. Older controls such as the I80 used a potentiometer on the rear cabinet for fine adjustment of the clamp; the J300 replaces that hardware trim with a software parameter. Do not assume a parameter write that works on a J300 will work on a J300M, MX3, or I80 without re-checking the manual.

8. Related Functions and Look-Ahead Packages

For users who want the time savings of disabling G107 but cannot tolerate the loss of corner-error protection, Yaskawa offers the optional HON (High-speed O-Number) and GON (General O-Number) advanced look-ahead packages. These packages pre-compute the velocity profile across a configurable number of look-ahead blocks and dynamically adjust feedrate to maintain servo tracking without falling back to the fixed corner clamp. On J300-era machines the packages are optional and rarely installed on stock MC660VG deliveries, but they remain the long-term solution for shops that need both speed and accuracy on free-form geometry.

9. Quick-Reference Decision Tree

1. Does the part have features that demand tight arc-corner accuracy?
   If yes → apply Option B (retune pm0416/pm0417) and verify on a sacrificial cut.
2. Is cycle time dominated by small-radius arcs (helical bores, circular pockets)?
   If yes → apply Option A (disable pm4031 D2) and verify dimensions on a sacrificial cut.
3. After the change, do parts fall inside the print tolerance?
   If yes → lock the parameters and update the CAM post-processor.
   If no → revert the parameter and reduce the F-word at the CAM level.
4. Is the F-word the binding constraint (cut still slow after disable)?
   If yes → the part-program F-value is below the axis acceleration envelope; raise F at the CAM level and re-test.

10. Safety and Good Practice

Parameter changes on a Yasnac J300 are non-volatile but reversible. Document every change in the machine log with the date, operator, before/after values, and the rationale. Run a calibration cut after every change before committing production parts. Never disable G107 on a part whose print tolerance is tighter than the published axis following-error envelope — the clamp exists precisely to keep parts inside that envelope on arcs. If the part program and the axis cannot together produce a part inside tolerance, the right answer is a CAM-side feed reduction, not a clamp disable.