C.03 · Checklist · 13 min read

Tolerance and Datum Checklist What the drawing has to state so that sintering shrinkage stays compensable — and which entries create false precision.

The most common cause of failed MIM first article inspections is not a manufacturing deviation. It is a drawing written for a machining process and handed to a sintering process unchanged. It then contains requirements that cannot technically be met, without anyone having objected — and omits information that determines the outcome.

This checklist walks the drawing field by field.

The figures the review is based on

Item Value Source
Standard tolerance ±0.3 % of nominal EPMA, guide value for the as-sintered condition
Tightest tolerance ±0.1 % of nominal, on a single fine-tuned dimension EPMA
Linear sintering shrinkage 14 – 20 % Industry guide value; a single measurement on a standard feedstock gave 14.3 ± 0.3 %
Batch scatter of shrinkage a few tenths of a percentage point Experience value; causes are powder distribution and binder content
Surface in the as-sintered condition Ra ≈ 0.8 µm EPMA; MIMA states "32 rms or better"
Density 95 – 98 % of theoretical density plant range; MPIF Standard 35 sets a minimum density per material
What does not exist for MIM A link to ISO 2768 and a link to IT grades per ISO 286. Neither EPMA nor MIMA nor ISO 22068 establishes one — ISO 22068 is a pure material standard and contains no dimensional tolerances at all. Putting "ISO 2768-m" in the title block of a MIM drawing states an accuracy nobody has verified. For press-and-sinter parts it is different: there IT grades are stated regularly, down to IT 8 to IT 7 for sized dimensions.

Twelve review points on the drawing

Datums and orientation

No. Review point Requirement
1 Complete datum system present primary, secondary and tertiary datum named, not just one face
2 Primary datum on the sintering support face the face resting in the furnace is the only one with reproducible position
3 Datums away from the gate the gate carries the vestige and the largest local density deviation
4 Datum targets spatially close together datum targets on widely separated regions add up direction-dependent shrinkage

Dimensional tolerances

No. Review point Requirement
5 Function-critical dimensions marked three to five at most, explicitly identified as such
6 Tight tolerances placed in one direction ±0.1 % is achievable along the flow direction, considerably harder across it
7 Remaining dimensions at standard ±0.3 % or wider; every unnecessarily tight dimension ties up inspection effort
8 Dimensions to be machined named including stock allowance, so the tool can carry it

Form, position and surface

No. Review point Requirement
9 Flatness set realistically it is hardest to hold across the longest span — check there first
10 Roundness and coaxiality demand only where function requires it; otherwise paid for expensively through sizing
11 Surface requirement split by face Ra 0.8 µm applies to flat faces, not to the gate vestige and support points
12 Gate position and support face released both belong on the drawing or in the release agreement

The three entries most often missing

The condition in which the dimension applies

A MIM part passes through up to four conditions: sintered, heat treated, sized, coated. Heat treatment changes dimensions and flatness; a coating changes outer dimensions at the micrometre level. If the drawing does not state the condition of measurement, the supplier measures as-sintered and the customer measures the coated part — and both are right.

The measurement method

At a tolerance of ±0.03 mm the method decides the result. Tactile, optical and computed tomography readings deviate systematically from each other at sintered edges, because edge rounding is captured differently. For first article inspection the method has to be named, otherwise the discussion is about instruments rather than parts.

The Cpk expectation per dimension

A blanket Cpk of 1.67 across all dimensions is usually not economically achievable in MIM. What is common and defensible is a split requirement: Cpk 1.67 and above on the named functional dimensions, Cpk 1.33 and above as the general standard. That split belongs before sampling, not in the discussion afterwards.

Worked example for point 6 A 20 mm dimension at ±0.3 % gives ±0.06 mm. The same dimension at ±0.1 % gives ±0.02 mm. The difference sounds small but changes the tooling strategy completely: at ±0.06 mm a compensation calculation is enough, at ±0.02 mm the tool needs a correction loop after the first sintering run — and that costs time in ramp-up, not money in the tool.

What actually makes shrinkage scatter

Compensation in the tool is uncritical as long as shrinkage is constant. It is not entirely. Four influences sit behind it:

  1. Feedstock batch. Powder distribution and binder content vary slightly from batch to batch. This is the largest single contribution.
  2. Fill level and packing pressure. They determine local green density and therefore how much a given area shrinks.
  3. Furnace position. Within one batch the temperature field is not exactly identical.
  4. Part orientation. Shrinkage differs parallel and perpendicular to the setter plate.

In practice: assuming a shrinkage scatter of a few tenths of a percentage point is correct — and demanding ±0.02 mm on a 30 mm dimension unintentionally demands batch-specific readjustment of the sintering parameters.

Our approach

We give tolerance feedback on every drawing before we design a tool. It names the dimensions achievable as-sintered, the dimensions that should be sized, and the dimensions that will need machining. That split is therefore fixed before the tool is cut — not during first article inspection.

Further reading

Questions about the topic of this article?

Our engineering and purchasing teams are available for technical enquiries — without obligation, without a form funnel.