Accuracy on a high-hour Mazak VQC-family gantry mill depends on which hop in the position-feedback path is worn, and glass scales cover only some of those hops. Scales were a rarely fitted option on these machines, so the question is which feedback device the machine actually has and what the ways look like.
Which gantry layout is this machine: VQC, AJV, or FJV?
The three series share the gantry idea but move different masses on different guideways, which decides which axis wears first.
| Series | Gantry orientation | Axis motion | Guideways | Notes |
|---|---|---|---|---|
| VQC | Perpendicular, spanning the table | Table moves left/right (X); head moves toward/away (Y) | Box ways | Both ends of X and Y supported back to the main bed. A 20/50 carries a 50-taper two-speed geared head; a 20/40 is CAT40. |
Identify the series from the nameplate and the axis layout before reading any accuracy claim. A box-way VQC and a linear-way AJV fail differently on X and Y.
How does position feedback travel from the CNC to the table and back?
The position command leaves the CNC, drives the servo amplifier and motor, passes through mechanical hops, and returns as a feedback count. Where the feedback device sits determines which hops the control can see. Physical layer first: every hop below is a mechanical or electrical element that adds error before any parameter matters.
| Hop | What it adds | Inside a rotary-encoder loop? | Inside a glass-scale loop? |
|---|---|---|---|
| CNC position loop and servo amplifier | Following error, tuning limits | Yes | Yes |
| Servo motor and rotary encoder | Encoder count only | Yes (sensor) | Yes |
| X servo belt | Slip, stretch, backlash, breakage | Depends on encoder location: motor-shaft encoder does not see it; screw-end encoder does. Read the drawings. | Yes, seen as a position error at the table |
| Ballscrew and nut | Pitch error, wear, backlash, thermal growth | No | Yes |
| Table and head on ways | Vertical and lateral play, tilt, yaw | No | Only the component along the scale axis |
| Scale and read head | Contamination, cable or mounting faults | Not present | Sensor itself |
A scale closes the loop at the table, so ballscrew wear and belt slip show up as a position correction instead of a part error. A scale does not remove way play. It reads travel along the axis only, so lift, lateral shift, and tilt from worn box ways pass straight through into the cut.
Do glass scales fix a worn ballscrew, and did these machines have them?
Scales fix ballscrew and belt error but were an option that most machines did not have. An operator with experience on roughly a hundred VQC and AJV machines could count on one hand the ones with scales installed. The scale option was Sony Magnescale. A 20/40 with the factory encoders, and a 1986 machine that spent its life at an aerospace shop, both worked fine without scales. Do not assume scales on a used machine because the brochure lists them.
| Criterion | Rotary encoder (semi-closed) | Glass scale (full-closed) |
|---|---|---|
| Screw wear | Grows into positioning error; correct with pitch-error compensation and reversal-error settings | Compensated automatically along the scale axis |
| Belt slip or breakage | Invisible if the encoder is motor-mounted | Visible as position error |
| Way play | Not corrected | Not corrected |
| Failure exposure | Low; sealed in the motor or screw end | Contamination and cover damage along the axis; extra sensor, cable, and parts sourcing |
| How to identify | Look for a scale cover strip and read head along each axis and check the machine's option documentation. Do not infer from the model name. | |
Where do 10,000 hours show up on a box-way gantry?
They show up in the ways and chip-handling, not in the structure. The gantry casting and large box ways are rated by owners as built to last; the damage on neglected machines comes from contamination.
| Finding | Mechanism | Test |
|---|---|---|
| About 0.080 in of play on X ways of a 2050 that still ran | Way covers destroyed and never replaced; chips, abrasive, and plasma slag in the air ground the ways. The machine kept cutting while grossly loose. | Lift/pry test with an indicator on X; inspect for missing wipers and covers |
| X servo belt breaks repeatedly | Chips pack the cavity under the table until the belt fails. Cleaning and replacing the belt without restoring the covers repeats the failure. | Open the cavity, look for packed chips, check belt tension and tooth wear |
| Table sliding surface worn | A 20/50 running three shifts, hard duty, needed the table re-lined at about 10 years | Inspect table sliding surfaces, check gibs and lubrication |
| Fixtures or workpieces sheared in the pallet pocket on smaller units | The pallet dives into a pocket during X moves; setup clearance can be exceeded | Check clearances before any program dry run on a small VQC/AJV |
| Slow tool drum | Cycle time issue, not accuracy | Time a tool change; factor into part-time cost |
A machine that still cuts is not a machine that still holds size. The 80-thou example proves the machine can run while grossly worn, so a running demonstration is not acceptance evidence.
Which approach to the purchase check should you commit to?
| Approach | What it detects | What it misses |
|---|---|---|
| Trust catalog accuracy and assume scales | Nothing | Scales are rare; catalog specs describe a new machine |
| Check feedback type, then measure axis error | Screw and belt condition | Way play if the machine was only measured along the axis |
| Inspect covers, cavity, and belt first, then measure play and axis error | Contamination damage, way wear, feedback errors | Little; add a cutting test for final proof |
Use the third approach. Cover and cavity condition predicts way wear on this class of machine, and way play is the one error neither an encoder nor a scale corrects. Treat scales as a bonus that reduces screw-related tuning, not as the reason the machine holds accuracy.
What is the inspection sequence on a VQC or AJV?
- Identify the series and axis layout (VQC, AJV, FJV) and record the taper and model designation.
- Determine the feedback type per axis: scale cover and read head present, or rotary encoder only. Note the encoder location if the drawings show it.
- Open the cavity under the table and inspect the way covers and wipers. Look for chip packing, belt cracking or missing teeth, and slag or abrasive deposits.
- Run a lift/pry test on each axis with an indicator on a fixed reference. Record vertical and lateral play at several positions and compare against the original Mazak specification for the model.
- Run a reversal test in both directions at multiple points. Compare commanded to indicated travel to find backlash and screw wear on encoder-only axes.
- Capture a laser positioning run on each axis, both directions, and compare with the pitch-error compensation table stored in the control. Read the stored values from the control; do not assume them.
- Run a ballbar circle test in XY to expose reversal spikes, backlash, and squareness error.
- On smaller units, check the pallet pocket clearance before any dry run.
How do you verify the machine holds accuracy after warm-up and under load?
- Warm up the spindle and all axes to a stable temperature and repeat the positioning and reversal measurements from the inspection sequence.
- Cut a test part with a bore pattern at full production load. Measure it on a CMM or with gauge blocks.
- Repeat the lift/pry play measurement after the cutting test to confirm the ways did not shift under load.
- Repeat the laser positioning run and ballbar test one more time. Accept the machine only if results match the first run and stay inside the original specification.
FAQ
Why does a Mazak VQC without glass scales still hold accuracy?
Position feedback comes from rotary encoders, and the control compensates screw pitch error and reversal error. Accuracy holds as long as the belt, screw, and ways are mechanically sound; encoder-only machines from aerospace shops have run without scales.
Why does the X axis on a VQC develop play?
Chips, abrasive, and slag get past missing or destroyed way covers and grind the box ways. One 2050 had about 0.080 in of X-axis play and still ran, so measure with a lift/pry test instead of judging by running behavior.
Why does the X servo belt keep breaking on a VQC?
Chips pack the cavity under the table until the belt fails. Replacing the belt only resets the clock; restore the way covers and clear the cavity to stop the repeat failure.
How do I tell a VQC from an AJV or FJV?
A VQC has a gantry across the table with box ways. An AJV keeps that layout but uses box ways only on Z and linear ways on X and Y. An FJV puts the gantry behind and parallel to the table, moving the table for Y and the head for X.