Checking Fixture Or Vision Measurement: 20,000 RMB Per Fixture Versus 50 Models

Overview

Dimensional inspection of a stamped or cast part is usually done with a checking fixture: a hard gauge that holds the part in a defined datum and presents pins, blocks and go/no-go features for each controlled characteristic. Fixtures are fast, tactile and well understood by production teams. They are also expensive per variant, occupy floor space permanently, and cannot measure anything they were not built to present.

Vision measurement does the same job with cameras and no contact. It is slower per feature on a single part, but the same hardware can be reprogrammed for another variant without a new gauge. The trade-off is not simply old versus new; it depends on how many variants the line runs, how many characteristics are controlled, and how the accuracy requirement compares with what the process can hold.

Big Bird Industrial (formerly Shimada Big Bird Industrial) has been building cleaning and deburring systems for automotive powertrain components since 2001, and the measurement discussion here uses the same numbers that appear in production specifications rather than laboratory comparisons.

1. What A Checking Fixture Actually Costs

A single checking fixture for an automotive structural part typically costs between 20,000 and 200,000 RMB, depending on the number of features, the datum scheme and the build tolerance of the gauge itself. That figure is per variant. A programme covering several hundred variants therefore carries a fixture cost that scales linearly with model count.

The Cost That Does Not Appear On The Quote

Each fixture also needs a storage location and a working area. A large structural fixture can occupy a footprint of three metres by three metres or more, and a plant running hundreds of fixtures needs a dedicated gauge room with controlled temperature. The recurring cost is space, handling and periodic recertification rather than purchase price, and it continues for the life of the programme.

Where Fixtures Remain The Right Answer

Fixtures are strong where the characteristic is tactile, where the inspection happens at the line side by an operator, and where the number of variants is small and stable. A go/no-go gauge that an operator can load in seconds is still the most efficient way to confirm a handful of critical features on a high-volume part.

2. What Vision Measurement Actually Costs

A multi-camera vision station is a larger single purchase than one fixture but a smaller purchase than the fixture set it replaces once the variant count passes a threshold. The economics turn on that ratio rather than on the absolute price of either option.

Coverage Per Station

A single vision system can be programmed for 50 to 80 vehicle variants, because switching variant is a matter of loading a different inspection recipe and, where necessary, a different fixture nest. The hardware does not change. This is the property that makes the comparison non-linear: fixture cost grows with model count while vision cost grows with the number of stations.

Speed Per Feature

A static multi-camera system can measure around 100 feature points in 30 seconds, because all cameras expose simultaneously and no mechanism has to move between features. A two-camera configuration with a robot repositioning the sensor is slower, in the order of three seconds per point. That difference is one reason static multi-camera layouts dominate where cycle time is tight.

3. Where Each Method Reaches Its Limit

Both methods have honest failure modes, and the useful comparison is where each one stops working rather than where each one performs well.

Vision Limits

Highly reflective curved surfaces return specular highlights that hide the edge being measured, and the usual mitigation is to project a laser speckle pattern so that the surface has texture to triangulate. Edges at abrupt angle changes produce missing points in the reconstructed cloud, which means a sharp upstand or a tight fillet may need a different view or a different sensor. Neither limitation is a defect of the method, but both have to be designed around.

Fixture Limits

A fixture measures what it was built to measure. If the drawing changes, the gauge is scrap or requires rework. If a new variant is introduced, a new fixture is needed. When a programme runs for years and the model mix shifts, the fixture population becomes a maintenance burden with no residual value.

4. The Mixed-Model Problem

Mixed-model production is where the comparison usually resolves. A line running a handful of high-volume variants can justify a fixture per variant. A line running tens of variants, each at lower volume, cannot: the gauge room grows faster than the production area.

Vision handles the mixed case because the marginal cost of the next variant is engineering time, not capital. That advantage becomes decisive when the variant list is still growing at the time the equipment is ordered, which is common on new energy vehicle programmes.

5. The Accuracy Budget Nobody Writes Down

Both methods are usually compared against the drawing tolerance rather than against the measurement uncertainty they contribute. A characteristic with a tolerance of plus or minus 0.5 mm and a measurement uncertainty of 0.1 mm consumes a fifth of the tolerance band before the process has done anything. The conventional rule of thumb asks for a gauge uncertainty of a tenth of the tolerance; in practice, structural parts often run at a quarter and the difference is absorbed by the process capability study rather than by the gauge.

This matters in the comparison because the two methods fail differently. A fixture is a hard gauge: its uncertainty is fixed at build and does not drift with part surface condition, but it wears at the locating pins and at the clamps. A vision system's uncertainty depends on surface finish, lighting stability and calibration, and it can degrade slowly as lamps age. Neither is automatically more accurate; both need a recertification interval written into the maintenance plan.

Datum Consistency Is The Shared Dependency

Every dimensional inspection depends on the datum scheme, and both methods hold the part in a defined orientation to establish it. Where the datum features are small, rounded or poorly formed, the same part measured on a fixture and on a vision station can differ by more than the tolerance being checked. That disagreement is a datum problem, not a hardware problem, and it is worth resolving before either method is purchased.

6. A Decision Sequence

The practical sequence is to count controlled characteristics and variants first, then compare the fixture set cost against one vision station, then check whether the accuracy requirement is inside the vision capability and whether the surface finish permits optical measurement at all.

If the answer is close, a hybrid layout often wins: a fixture or a small gauge for the two or three characteristics that must be checked every part at the line side, and a vision station for the full dimensional audit at lower frequency. That split keeps the fast tactile check where it belongs without building a gauge for every variant.

Frequently Asked Questions

Is Vision Measurement As Accurate As A CMM?

For the characteristics vision is designed to measure, agreement with CMM is within 0.1 mm, and positioning repeatability on a multi-camera system is below 0.1 mm. It is not a general replacement for a CMM on every characteristic, particularly deep bores and features that cannot be seen from any camera position.

How Long Does It Take To Add A New Variant?

Adding a variant to an existing vision station is a programming task measured in days, not a procurement task measured in months. The hardware, lighting and fixtures stay as they are unless the new part falls outside the station envelope.

Can Vision Inspect A Part With A Highly Polished Surface?

Yes, with a different illumination strategy. Projecting a laser speckle pattern gives the surface artificial texture, which restores triangulation. Without it, the highlight moves with viewing angle and the measurement becomes unstable.

What Happens To Existing Fixtures?

They usually stay for critical line-side checks and for customer audits that specify a tactile method. Retiring them entirely is rare, because a written inspection standard often names the gauge.

Conclusion

The comparison is not about which method is more advanced. It is about which cost curve applies: fixtures scale with the number of variants, vision scales with the number of stations. Once the variant count and the characteristic count are known, the answer is arithmetic rather than preference.

Where the two are close, a hybrid layout is usually the honest answer, and it is also the one that survives a model change without scrapping the inspection strategy.