How Automatic Measurement Improves Six-Side Mold Block Machining

A mold block has only six outside faces, but machining those faces correctly is not a simple facing job. The finished block must have the correct length, width, and height. Opposite faces must be parallel, adjacent faces must be square, and each finished reference face must be suitable for later cavity, hole, insert, and guide-pin machining.

The main problem is that the six faces are connected. After each flip, one or more previously machined faces become the references for the next operation. A small error made early in the process can therefore affect several later faces.

Automatic measurement checks the real block while it is still inside the CNC machine. A spindle probe can locate the block, measure available stock, check its angle, and set the work coordinate. A tool setter can check tool length and detect major tool damage. The CNC program can then decide whether to continue, measure again, make a small approved correction, or stop.

NIST describes on-machine measurement as a useful way to check workpiece position, tools, stock, process conditions, and finished features. In suitable applications, it can reduce wasted machining time, scrap, and the need to remove and realign a part for inspection.^[1]

For the basic order of machining the six faces, see the six-side mold block machining guide.

Why the Six Faces Affect Each Other

On a three-axis machining center, a common process is:

1. Machine the first large face.
2. Turn the block over and machine the opposite large face.
3. Stand the block on a finished side.
4. Machine the next pair of side faces.
5. Turn the block again for the two end faces.
6. Check the final size and the relationship between the faces.

Each setup uses surfaces produced in an earlier setup. If a chip is trapped under one corner, the block may tilt. The machine can still cut a smooth face, but the face will be in the wrong direction. If that face is later used as a reference, the same error moves into the next setup.

The final length, width, and height may appear acceptable while the block still has poor parallelism or squareness.

Size

The block must remain within the drawing tolerance. If the finished width should be 300.00 mm but measures 299.70 mm, the block is undersize. Removed material cannot be restored.

An oversize block may still be corrected if enough material remains. This is why measurement before the final cut is more useful than measurement only after the block is finished.

Flatness

Each reference face must be flat enough to support the next setup. A face may be high in the center, low at one end, lifted at one corner, or locally damaged by an uneven cutter.

One probe touch cannot describe a whole face. Several points are needed.

Parallelism

Opposite faces must stay parallel. If the top and bottom are not parallel, the block thickness changes across its length or width.

This can affect cavity depth, mold closing, insert position, plate assembly, and the position of holes machined from different sides.

Squareness

Adjacent faces must meet at the required angle. Poor squareness can shift later holes, pockets, and guide components when different faces are used as references.

Available Stock

The total raw size is not enough. Material must be available at every required corner.

A block may have enough average width but still have too little material at one corner because a saw-cut face is angled.

Angle Error

A small angle creates a larger height difference on a long block.

The basic calculation is:

height difference = block length × tan(angle)

For a block that is 1,000 mm long:

1,000 × tan(0.01°) ≈ 0.175 mm

This is only a geometric example. The real result also depends on the fixture, machine condition, support points, and stiffness of the block.

What the Measurement System Does

Spindle Probe

A spindle probe is loaded into the spindle like a cutting tool. Its stylus touches selected points on the block.

It can be used to:

• Find an edge or corner.
• Find the center between two faces.
• Measure the position of a surface.
• Check workpiece rotation.
• Measure raw-stock size.
• Set or update a work coordinate.
• Check selected finished dimensions.

The probe measures only the points selected by the program. It does not automatically scan every part of a face.

Tool Setter

A contact or laser tool setter may check:

• Tool length.
• Tool diameter, when the system supports it.
• A missing tool.
• A completely broken tool.
• A large change caused by wear or incorrect assembly.

A tool setter cannot identify every wrong tool. If two tools have similar measured lengths and diameters, a simple measuring cycle may not know that the wrong one is installed.

Software and CNC Logic

The measuring cycle must do more than collect a number. It should compare that number with an expected range and choose a safe action.

The program may:

• Write an approved result to a work offset.
• Update a tool-wear offset.
• Repeat a measurement.
• Clean the surface and measure again.
• Repeat a finishing pass.
• Store the result.
• Stop and show an alarm.

The controller, probe program, offset system, and machining program must use the same datum and sign rules. For more detail, see the guide to CNC controllers used in mold machining.

ISO 230-10:2022 provides test procedures for touch probes, scanning probes, contact tool setters, and laser tool-measuring systems that are built into CNC machine tools.^[2]

What the Results Can Prove

Not every probe result has the same purpose. It helps to separate measurement into three types.

On-machine probing is strongest for location and process control. It can also support final inspection, but it does not automatically make the CNC machine equal to a calibrated coordinate measuring machine.

The CNC machine uses the same axes, guides, spindle, and position-feedback system for both cutting and measuring. An error that affects the cutting motion may also affect the measurement motion. This shared error must be considered before the probe result is used for final acceptance.^[3]

ISO 230-10 also states that testing a machine tool for use as a full coordinate measuring machine is outside that standard's scope. That task needs additional checks for machine geometry and measurement traceability.^[2]

Check the Raw Block Before Cutting

Raw blocks may have saw-cut taper, forged surface variation, scale, bent edges, or local low areas. Measuring the block before roughing can prevent the machine from spending hours on material that cannot produce the required finished size.

The probe should answer:

• Is the correct block loaded?
• Is it in the correct direction?
• Is the block large enough?
• Does every corner have enough stock?
• Is a raw face badly angled?
• Can the finished model fit inside the real block?

If raw material is being ordered or prepared for a job, see the custom mold steel blocks page and the guide to calculating mold block size and machining allowance.

Separate the Allowances

Do not combine every allowance into one number. A process plan may need separate values for:

• Cleaning the raw surface.
• Rough machining.
• Finish machining.
• Grinding.
• Heat-treatment correction.
• Clamping or handling space.

Suppose the final width is 400 mm and the block must retain 1 mm of finishing stock on each side after roughing.

The required width after roughing is:

400 + 1 + 1 = 402 mm

This does not mean the original saw-cut block should be only 402 mm wide. The raw block also needs enough material to clean up rough, damaged, or angled surfaces.

Measure More Than One Point

A single touch on a saw-cut face may hit a high ridge or a low area. Measure several points and judge the smallest usable stock.

For example:

• Left front allowance: 1.4 mm
• Left rear allowance: 1.3 mm
• Right front allowance: 1.2 mm
• Right rear allowance: 0.1 mm

The average looks acceptable, but the right-rear corner may not clean up.

Do Not Always Center the Finished Block

Automatic measurement can find the center between two raw faces, but equal stock on both sides is not always correct.

There are three common choices:

• Equal stock: divide the material evenly when both sides are equal in importance.
• Fixed reference: hold one prepared face and remove most of the material from the opposite side.
• Unequal stock: leave extra material on a side that will later be ground, corrected, or used for an important feature.

For production pre-machining, a duplex milling machine can machine a pair of opposite faces in one clamping and reduce the number of manual flips.

Set the Work Coordinate Safely

A basic probing cycle may:

1. Touch the main support face.
2. Touch the side reference.
3. Touch the end reference.
4. Calculate the block position.
5. Compare it with the expected position.
6. Write the approved result to the selected work offset.

The program should never write a measured value without checking it.

If the expected X position is within 5 mm of a known fixture point but the probe reports a position 40 mm away, the cycle should stop.

Possible causes include:

• The wrong block is loaded.
• The block is reversed.
• The wrong probing cycle is running.
• The fixture has moved.
• The result is being written to the wrong work offset.

Automatic entry removes typing work, but it can also enter a wrong value quickly if the program does not include checks.

Check Seating and Rotation

Seating

A seating check looks for a block that is resting on a chip, burr, damaged support, or uneven clamp.

The check must use a known machined surface, prepared reference pad, locating step, or fixture sensor. A rough face cannot be used directly because its own unevenness may look like a seating error.

Assume four known reference points measure:

• Front left: 100.003 mm
• Front right: 100.008 mm
• Rear left: 100.006 mm
• Rear right: 100.084 mm

The rear-right point is much higher. The block may be sitting on a chip or burr, or the support may be damaged.

There is no single seating limit for every block. The allowed difference depends on:

• Block length and width.
• Required parallelism.
• Distance between the probe points.
• Surface roughness.
• Probe repeatability.
• Block stiffness.

After an alarm:

1. Stop the cycle.
2. Release the clamp safely.
3. Clean the support and the block.
4. Check for burrs and damage.
5. Reload the block.
6. Repeat the measurement.

A hydraulic clamping system can make repeated clamping more consistent, but excessive or uneven clamping force can still bend the workpiece.

Rotation

The probe can touch two separated points on one side and calculate how far the block is rotated.

If the side changes by 0.05 mm over a length of 250 mm:

angle = arctan(0.05 ÷ 250) ≈ 0.0115°

For roughing irregular stock, the CNC may rotate the work coordinate to follow the block. For a face that will become an important physical reference, it is usually safer to realign the block.

Stop and investigate when:

• Repeated angle measurements disagree.
• The angle changes after cutting.
• The top and bottom of the same side give different angles.
• The block may have moved.

Coordinate rotation changes the toolpath direction. It does not fix poor seating or fixture movement.

Check the Face Before the Final Cut

Remaining Stock

After roughing, measure the material left for finishing.

If the final surface is Z = 0 and the planned finishing stock is 0.5 mm, the rough surface should be near Z = +0.5 mm.

If one area measures +0.1 mm and another measures +1.2 mm, possible causes include:

• Uneven roughing.
• A tilted setup.
• Poor stock distribution.
• Movement caused by released material stress.

Measure several areas, not only the center.

Flatness Trend

A simple pattern may use four corner points and one center point. A large block may need more points.

The results can show a general tilt, a high center, a low center, or a lifted corner. They cannot prove that no high or low area exists between the measured points.

For this reason, call the result a sampled flatness check or flatness trend unless the method has been formally proved for final inspection.

Parallelism

Parallelism needs several matching points on the opposite face.

For a nominal thickness of 100.00 mm, assume the measurements are:

• 100.01 mm
• 100.02 mm
• 100.07 mm
• 100.08 mm

The average is close to 100.00 mm, but the 0.07 mm spread shows that the thickness changes across the block.

The cause could be poor seating, a tilted cut, block movement, tool loading, or machine geometry. The measurement shows the problem but does not always show its cause.

Squareness

A side face can be measured at two or more heights. A change in position shows an apparent angle.

The result is also affected by:

• Machine-axis squareness.
• Probe calibration.
• Stylus length.
• Touch direction.

Use this check to find clear setup errors. Use independent inspection when the drawing requires squareness tighter than the proven capability of the machine and probe.

Control the Cutting Tool

A correct work coordinate cannot correct a wrong tool length.

If the face-mill length is wrong by 0.08 mm, the finished face may also be wrong by about that amount even when the block position is measured correctly.

Measure important finishing tools:

• After the machine has warmed up.
• Before a final pass.
• After a long roughing cycle.
• After changing inserts.
• After a suspected collision or breakage.

A face mill may have several inserts. A tool setter may detect the highest insert but cannot prove that every insert is at the same height. Unequal inserts may leave lines, steps, or uneven cutting loads.

Tool choice must also match the material and cutting stage. See the available CNC cutting tools for common milling and mold-machining applications.

Use Compensation Carefully

A safe process can divide measured errors into three zones.

These are example values, not general industry limits.

If the drawing tolerance is ±0.020 mm, a ±0.030 mm correction range does not become the final acceptance limit. It only defines whether another finishing pass may be attempted. The corrected block must still meet ±0.020 mm.

Suppose the target surface is Z = 0.000 mm and the probe measures Z = +0.018 mm. The required correction size is 0.018 mm.

Do not copy a universal positive or negative offset sign. The correct sign depends on:

• The CNC control.
• The type of offset.
• The cutting direction.
• The program logic.

Conditions for Automatic Correction

Allow correction only when:

• Repeated measurements agree.
• The surface is clean.
• The block has not moved.
• The tool is not damaged.
• Enough stock remains.
• The error is a size or position error.
• The correction is inside a tested limit.

Errors That Should Stop the Process

Do not correct these errors with a simple offset:

• Twist.
• Poor squareness.
• Poor seating.
• Fixture movement.
• Missing material.
• Workpiece bending.
• Measurements that keep changing.

An offset moves the toolpath. It cannot repair the shape of the block.

Limit Repeated Corrections

The program should limit:

• The largest single correction.
• The total correction on one block.
• The number of repeated finish passes.
• The number of corrections in the same direction.

Repeated corrections in the same direction may indicate tool wear, heat movement, or calibration error.

Keep the Measurement Reliable

Check the Machine

The probe cannot be more reliable than the machine motion that carries it.

Machine errors may include:

• Axis positioning error.
• Axis straightness error.
• Squareness error between axes.
• Rotary-axis error.
• Changes caused by heat or load.

Modern machine-tool calibration separates errors within one axis, between different axes, and across the machine's working space. The correct test and correction method depends on accuracy needs, cost, and machine design.^[4]

Calibrate and Then Verify

Probe calibration normally finds:

• Stylus length.
• Effective ball diameter.
• Trigger position.
• Differences between probing directions.

Recalibrate after:

• Changing the stylus.
• A probe collision.
• Spindle or machine maintenance.
• A large temperature change.
• Unexpected measurement drift.

Calibration is not the same as verification. After calibration, measure a known sphere, ring, gauge block, or previously inspected part to check the complete system.

Repeatability Is Not Accuracy

A probe that reads 100.050 mm every time is repeatable. It is not accurate if the true value is 100.000 mm.

A shop may begin with 20 or more repeated measurements as an internal check, but this is not a formal proof of measurement capability.

Also test:

• The cold and warm machine.
• Different positions on the table.
• Different touch directions.
• Short and long styli.
• The result against an independent measuring device.

Use the Correct Probe Speed

Probe speed is not only a cycle-time setting. NIST tests found that probing feed rate can create a systematic measuring error because the machine continues to move during the probe's trigger and stop response.^[5]

Calibration and production measurement should therefore use the tested speed and direction.

Clean Before Measuring

A chip about 0.1 mm thick can create an error of a similar size. If it lifts one corner of a long block, the resulting angle can create a larger error at the opposite end.

Different contamination causes different problems:

• A chip usually gives a high reading.
• A burr may give a stable but wrong reading.
• An oil film may give a small changing result.
• Moving coolant may affect the trigger.
• Strong airflow may shake a long stylus.

A useful cycle is:

1. Measure the point.
2. Clean the surface.
3. Measure again.
4. Compare the two results.
5. Stop if they do not agree.

A spindle-mounted CNC chip blower can help clean open surfaces, but the second measurement must confirm that the surface is actually clean.

Choose Useful Probe Points

Avoid measuring:

• On a burr.
• On a chamfer.
• Too close to an edge.
• On a deep cutter mark.
• On an interrupted surface.
• Where coolant collects.
• Where the stylus touches at a shallow angle.

More points are not automatically better. The points must be placed where the expected error is most likely to appear.

Control Temperature

Steel changes size when its temperature changes.

NIST gives an example thermal expansion coefficient of 11.5 µm per meter per degree Celsius for gauge-block steel.^[6]

Using that value for a 1,000 mm steel block with a 5°C temperature increase:

11.5 × 1 × 5 = 57.5 µm

57.5 µm = 0.0575 mm

The actual value varies by steel grade, so the correct material coefficient should be used for close-tolerance work.

ISO 1:2022 defines the standard reference temperature used for dimensional and geometric properties such as size, position, angle, form, and surface texture.^[7] In normal dimensional metrology, this reference is 20°C.

Temperature can change:

• The block.
• The spindle.
• The cutting tool.
• The machine bed and column.
• The ball screws.
• The probe and long stylus.

Machine-tool thermal error is affected by internal heat, spindle operation, axis movement, coolant, and the workshop environment. These changes can move the tool relative to the workpiece during a long machining cycle.^[8]

For tight tolerances:

• Warm up the machine in a controlled way.
• Avoid measuring a hot block and accepting it as a cold size.
• Use similar temperature conditions for calibration and measurement.
• Allow large blocks time to reach a stable temperature.
• Record temperature when it can affect the decision.

Example Six-Side Process

Consider a block with these example requirements:

• Final length: 500.00 mm
• Final width: 400.00 mm
• Final height: 120.00 mm
• Approximate raw size: 504 × 404 × 124 mm
• Size tolerance: ±0.02 mm
• Parallelism between opposite faces: 0.03 mm

These values are only a process example. They are not recommended limits for every mold block.

First Face

Measure several points on the raw upper surface to find:

• The highest area.
• The general slope.
• The safe first cutting level.
• The available cleanup stock.

Machine the first large face. This becomes the main support reference.

Where the fixture allows it, also machine small reference pads or a narrow strip that the probe can reach after the first flip.

Opposite Face

Turn the block onto the first finished face.

The upper main face is still raw. Its four-point height difference cannot be used directly as proof that the block is seated correctly.

Check seating with one of these methods:

• Prepared reference pads.
• A probe-accessible finished side feature.
• A fixture contact sensor.
• A defined mechanical check.

Rough the opposite face and leave an example finishing allowance of 0.3 mm.

The 0.3 mm value must be adjusted for the material, block size, tool, heat-treatment plan, and final process.

Measure the remaining height, check the finishing tool, complete the final cut, and then measure several points to check the thickness trend.

Third and Fourth Faces

Place the block against a square fixture using the first finished face as the support.

Check:

• Contact with the support.
• Side rotation.
• Expected corner position.

Machine the third face. This becomes the second reference.

Turn the block onto the third face. At this stage, there are enough finished surfaces for a valid multi-point seating check.

Measure the remaining width, apply the planned stock split, and finish the fourth face.

Fifth and Sixth Faces

Use the first and third finished faces to locate the block.

Check the position, rotation, and remaining length stock.

Machine the fifth face and use it as the final end reference.

Turn the block against that face, measure the remaining length, and machine the sixth face.

Example Seating Failure

Assume a later setup gives these readings:

• Front left: 120.004 mm
• Front right: 120.007 mm
• Rear left: 120.006 mm
• Rear right: 120.074 mm

The rear-right point is about 0.067 mm higher than the other points.

The machine stops. The operator finds a thin chip under that support area, cleans the block, and reloads it.

The new readings are:

• Front left: 120.005 mm
• Front right: 120.007 mm
• Rear left: 120.006 mm
• Rear right: 120.009 mm

The block is now seated consistently, so machining can continue.

Match the Machine to the Block

Three-Axis Machine

A three-axis machine may be enough when:

• The block is easy to turn.
• The fixture is stable.
• Batch size is low.
• The required geometry can be held through controlled flips.
• The extra cost of a multi-axis machine is not justified.

Automatic measurement reduces setup risk, but it does not remove the need to flip and reclamp the block.

Five-Axis Machine

A five-axis machine can reach more faces without manual repositioning. This can reduce handling and keep more surfaces in one coordinate system.

It also adds new errors that must be checked:

• Rotary-axis center error.
• Pivot-point error.
• Rotary positioning error.
• Postprocessor error.
• Fixture-height error.

See the GZXC-2000 five-axis machining center for an example of a machine configuration with rotary axes and spindle temperature control.

Five axes do not automatically guarantee a more accurate block. The rotary system must be calibrated and proved.

Large Blocks

Large blocks need extra attention to:

• Support position.
• Bending under their own weight.
• Table load.
• Fixture bending.
• Crane loading.
• Temperature stabilization.

Do not perform the seating check while a crane or lifting strap is still carrying part of the block's weight.

A heavy-duty horizontal machining center with a rotary table can machine several sides with less reclamping, provided its table load, travel, fixture, and rotary accuracy suit the block.

Thin Plates

A thin plate may be forced flat by the clamps, machined in that bent condition, and then spring back after release.

The probe may show a good flatness trend while the plate remains clamped. That does not prove the free plate will remain flat.

Possible controls include:

• More support points.
• Lower clamping force.
• Balanced removal from both faces.
• Rough machining before stress relief.
• Light finishing passes.
• Measurement in the condition required by the drawing.

Material Condition

Material condition changes the process:

• Annealed steel: easier to cut but may move when large amounts of material are removed.
• Pre-hardened steel: avoids later hardening but can still contain residual stress.
• Heat-treated steel: may need extra correction or grinding stock.
• Forged or flame-cut material: needs more raw-surface measurements.
• Aluminum mold plate: cuts easily but is more sensitive to burrs, clamping marks, and temperature change.

Use the Measurement Pattern to Find the Cause

Record Useful Data

Useful records include:

• Block ID.
• Machine ID.
• Raw dimensions.
• Work offsets.
• Seating results.
• Rotation results.
• Tool measurements.
• Corrections.
• Final dimensions.
• Alarms.

Store the original measurement separately from the value after correction.

The records can show:

• A tool slowly wearing.
• A machine warming up in the same pattern each day.
• One fixture repeatedly causing a seating alarm.
• One machine needing more correction than another.
• A probe that has started to drift.

Introduce Automation in Stages

Automatic Location

• Find edges and centers.
• Measure rotation.
• Set work coordinates.

Setup Protection

• Confirm the block is present.
• Check raw size.
• Check orientation.
• Check available stock.
• Check seating and rotation where suitable references exist.

Process Measurement

• Check roughing allowance.
• Check finishing tools.
• Measure selected finished surfaces.
• Store results and alarms.

Closed-Loop Correction

• Update approved tool offsets.
• Repeat a finishing pass.
• Stop when a limit is exceeded.
• Track the total amount of correction.

A shop should prove the lower stages before allowing the machine to change offsets automatically.

Check the Real Return

Do not judge automatic measurement only by the extra probing minutes in the CNC cycle.

Measure:

• Manual setup time.
• Alignment time after each flip.
• Inspection handling.
• Rework.
• Scrap.
• Operator attendance.
• Unattended machine hours.
• First-pass yield.

For example, if probing saves 25 minutes per block and the shop machines 40 blocks per month:

25 × 40 = 1,000 minutes

1,000 ÷ 60 = 16.7 hours per month

This is only an example. The actual return must also include:

• Probe and tool-setter cost.
• Software.
• Installation.
• Training.
• Program development.
• Calibration tools.
• Maintenance and collision risk.

Full closed-loop measurement may not be necessary when the tolerance is wide, setup is simple, batch size is very low, or the machine cannot measure consistently.

The probe may still be useful only for workpiece detection, datum setting, stock checks, and broken-tool checks.

Final Inspection

Critical blocks may still need independent inspection because the CNC machine and probe share the same motion system.

Final inspection may use:

• A coordinate measuring machine.
• A surface plate.
• A precision square.
• A height gauge.
• A micrometer.
• An electronic level.

The inspection method must match the drawing tolerance and the accuracy that must be proved.

If milling cannot meet the required final flatness, size, or surface finish, a separate surface grinding process may be needed.

Conclusion

Automatic measurement improves six-side mold block machining by checking important conditions before more material is removed.

It can:

• Check raw stock.
• Find the real block position.
• Set the work coordinate.
• Check available finishing stock.
• Find setup movement and rotation.
• Check important tools.
• Allow small, tested corrections.
• Stop the machine when a result is unsafe.

Its main value is early error detection.

A setup problem found before cutting may take only a few minutes to correct. The same problem found after all six faces are finished may cause long rework or make the block unusable.

Automatic measurement works best with stable fixtures, clear reference surfaces, clean measurement points, calibrated equipment, controlled temperature, and independent final inspection when required.

Used correctly, it changes six-side machining from a process based on assumptions into a process based on measured results.