How to Run Mold Sampling for PPAP: ISIR and Capability

A connector housing program stalled at PPAP submission when the ISIR full-layout report showed 9 of 214 dimensions outside tolerance. The same mold had passed T1 sampling six weeks earlier. The gap was not the steel. The trial report had used a best-fit alignment, while the submission layout was measured to datum A-B-C, and the 0.08 mm true-position shift on the terminal pads came from the two measurement methods. The customer rejected the file, the molder re-sampled, and the delay cost roughly USD 18,000 in line-down charges before the mold maker's CMM records settled the argument.

Mold sampling for PPAP is the paperwork step that turns molded samples into an approved part number. Technical trials T0, T1 and T2 decide whether the mold makes good parts. The PPAP submission decides whether the customer allows those parts into serial production. The two overlap at the sampling stage, and each has its own gate, evidence list and owner. This article covers the customer-facing side from a mold maker's view: ISIR dimensional data, material and capability evidence, and the measurement discipline behind it.

DieStrike is an IATF 16949-certified precision mold maker, so we support customer PPAP with ISIR-level dimensional data from our trial sampling. We do not sign the part submission warrant; the molder owns that signature. But most of the evidence behind the warrant is created while the mold is still in our hands, and that is where this guide concentrates.

The Snapshot

• AIAG PPAP Level 3 is the common automotive submission level, and it demands dimensional results, material and performance data, and process capability evidence, not just sample parts.
• An ISIR full layout covers every dimension on the drawing, commonly 150 to 300 features on a connector housing, measured on a CMM reporting to ±0.005 mm.
• Initial process capability is commonly requested at Cpk ≥ 1.33, and 1.67 or better on critical features; treat both as common requirements, never as universal law.
• A continuous 300-part sampling run is a common industry convention for the capability study, but the customer's own requirements set the final count.
• As an IATF 16949-certified mold maker, DieStrike delivers ISIR-level dimensional data, material certificates and cavity repeatability evidence within its standard 2-5 week tooling lead times.

What PPAP Level 3 Asks of a Mold Maker

PPAP, the Production Part Approval Process, is defined in the AIAG PPAP manual, now in its fifth edition. Its purpose is a single claim: the production process can hold the drawing at volume. The evidence package runs to 18 elements, from design records to the signed part submission warrant.

Most automotive programs run at Level 3, the AIAG default for production parts. Level 3 requests the warrant plus product, material, performance and process capability data. Level 1 requests only the warrant and supporting records. Level 5 requests a complete audit of the supplier site. The customer's own requirements can reset any of these choices, which is why the submission level is agreed before sampling, not after.

Of the 18 elements, a mold maker directly feeds a small group. Dimensional results come from measuring molded samples. Initial process studies borrow the stability the mold showed during trials. Qualified laboratory documentation covers the CMM and its calibration. The rest, from the process flow chart to the PFMEA, is built by the molder around its own production line.

The split is worth naming before a program starts. When a connector OEM such as TE Connectivity, Amphenol, Luxshare or Dongshan issues a PPAP requirement through the supply chain, the molder assembles the file. The mold maker's contribution is the dimensional and material evidence that makes the file defensible. Molds that ship without that evidence force the molder to re-sample, and re-sampling adds 1 to 3 weeks to an already tight launch.

The Submission Sampling Run: Parts, Press and Protocol

The submission run must look like production, because PPAP approves a process, not a prototype event. Use the production press, the production mold and the production process settings. A lab press with hand-tuned parameters produces parts that will not match the capability study run later on the floor.

Run continuously for the agreed part count. A 300-piece continuous run is a common convention in automotive sampling, long enough to expose wear, drift and operator effects that a 30-piece batch hides. The count is a convention, not a law. Some customer-specific requirements ask for 100 parts, others for 500, and the PPAP manual itself leaves the quantity to the agreed control plan.

Log the run like an experiment. Record the machine number, melt temperature window, injection pressure, cooling time and cycle time at the start and end of the run. Stamp or label parts by shot number so any later measurement can be traced to a moment in the run. Keep the first approved part as a master sample, because OEM quality engineers frequently ask to compare it against production lots.

Divide the parts into evidence groups before measuring. Pull dimensional samples from the beginning, middle and end of the run, commonly every 50th to 100th shot. Pull appearance and color samples separately. Retain 1 to 3 parts per cavity as archives. This article handles the submission paperwork; the guide to running a mold trial covers the T0 to T2 technical work that comes before it and produces the stable tool this run depends on.

ISIR: The Full-Layout Dimensional Report

ISIR stands for Initial Sample Inspection Report, and in practice it is the densest document in the package. A full layout measures every dimension on the drawing, including the ones nobody checks in production. A connector housing commonly carries 150 to 300 callouts once pitch, true position, form and draft are all counted.

The report is judged against the drawing's GD&T, not against a loose interpretation. Features are located from the datum reference frame, usually A-B-C on a connector, and a dimension that passes under a best-fit alignment can fail by a wide margin under the datums. That single mismatch caused the rejection in the opening example, and it repeats whenever the trial lab and the submission lab use different alignment strategies.

Measure multiple parts per cavity, because one part proves nothing about a mold. A typical ISIR layout measures 5 parts per cavity and reports each feature's nominal, actual, deviation and verdict. True position, flatness and profile callouts matter most, since they expose cavity-to-cavity and run-to-run movement that a single dimension cannot.

Condition the parts before measuring. Plastic parts move with temperature and humidity, and a freshly ejected sample can still be shrinking. Common practice conditions samples at 20 to 23 °C for several hours before the CMM run. Every measurement happens on the same calibrated machine so the dimensional report and the capability study share one source of truth. Our guide to reading a CMM report walks through the columns, datums and red flags that decide whether a layout really passed.

As a mold maker we add one layer the molder cannot see: the as-built cavity dimension. The cavity steel was machined to compensate for shrinkage, so the molded part and the steel do not agree by design. Supplying the as-built CMM data on the cavity, cut to ±0.005 mm with geometry features at ±0.002 mm, lets the molder explain that difference before the customer asks. A full layout that arrives with both steel and part data closes more submissions in one round.

Material Certificates and Heat Treatment Records

PPAP element 8 in the AIAG structure covers material and performance test results. For the plastic part, the resin certificate names the grade, the lot and the physical property test values. The molder pulls these from the resin supplier and keeps them lot-traceable, because the certificate must match the material actually molded in the sampling run.

The mold side carries its own certificates. Every steel grade in the tool should have a mill certificate naming the supplier, grade and heat number. Heat treatment records document the hardening cycle, soak and tempering. Hardness spot checks on critical inserts confirm the surface condition, and for hardened cavities DieStrike runs typical cavity steel to HRC 62 so the evidence file matches the machining reality.

These tool certificates rarely travel inside the PPAP warrant. They belong in the mold file that supports it. When an OEM auditor asks why a cavity wore faster than its neighbor, the answer lives in the steel cert and hardness record, not in the part data. Keep both sets current, and keep the mold-side file in the same quality system that produces the part-side report.

The Capability Run: Cpk ≥ 1.33 as a Common Gate

Capability studies answer the question the ISIR cannot: the ISIR proves the mold can make parts, and the study proves the mold plus process can keep making them. The index is Cpk, which compares the process spread and centering against the tolerance band. A common initial-capability requirement across automotive CSRs is Cpk ≥ 1.33, with 1.67 or higher reserved for critical, safety-related or customer-nominated features.

Those numbers are common practice, not universal law. Some customers accept 1.33 on everything, and some VDA-based programs demand stricter values on defined characteristics. The customer-specific requirement, not the industry average, is the binding number. Read it before the run so the study samples the right features at the right count.

A typical initial study collects 25 subgroups of 4 to 5 consecutive parts, roughly 100 to 125 measurements per feature, drawn from the continuous run. The molder charts each critical dimension in sequence to check for drift before computing capability. A drifting mold produces a low Cpk even when every part is in tolerance, and that is a tool-side signal, not a process-side one.

← swipe to scroll →

Table note: ppm figures are typical statistical conversions for a centered, normally distributed process. The customer's CSR overrides any table value.

When Cpk comes back below the gate, split the cause before touching anything. A wide spread with good centering points to process noise or tool instability. A centered process with an off-center mean points to shrinkage compensation or setup. The mold maker owns the tool-side share, and a stable cavity is the precondition for a stable number. If the study fails on a tool dimension, the correction loop is the same one our IQ, OQ, PQ validation guide describes for medical tooling, with the capability gate set by the automotive CSR instead of the FDA protocol.

Gage R&R: Why Measurement Comes Before Capability

Capability numbers inherit the error of the gage that produced them. If measurement variation eats 40% of the tolerance, the Cpk number is fiction. That is why PPAP evidence chains typically include a gage R&R study before the capability study is trusted.

R&R stands for repeatability and reproducibility, the variation of one operator measuring the same part and the variation between operators. A common AIAG-style study uses 10 parts, 3 appraisers and 2 to 3 trials each. The standard acceptance bands are widely used: under 10% of total variation is acceptable, 10 to 30% is conditional and depends on the application, and over 30% is unacceptable until the gage or the method improves.

Molded connector parts complicate R&R because they flex. A housing that bends 0.03 mm under the fixture reads as operator error when the part simply moved. Rigid fixtures, datum-located nests and consistent seating instructions cut that artificial spread, and the CMM's own accuracy matters as much as the fixture. A machine reporting to ±0.005 mm supports tolerances ten times larger, which covers most connector callouts at ±0.05 mm and above.

Run the R&R on the same CMM that will produce the ISIR and capability data. Mixing machines between the three documents reopens the alignment and uncertainty questions that cost the opening example its submission. One calibrated machine, one fixture strategy and one measurement program keep the whole package internally consistent.

IMDS and the Electronic Material Trail

IMDS, the International Material Data System, is where automotive material declarations actually live. OEMs require suppliers to submit a material data sheet, or MDS, for each part, built from the resin supplier's declared chemistry. Without an accepted IMDS entry, a PPAP warrant is commonly considered incomplete regardless of the dimensional results.

The MDS is assembled by the molder using resin data, and it flows upward to the OEM. The mold maker's direct input is small, but two details matter. First, mold release agents and process oils can contaminate a declared chemistry, so the sampling run should use the production release strategy the MDS assumes. Second, the declaration deadline often precedes the part approval, and OEMs commonly expect the IMDS entry to be submitted and accepted before or with the warrant, so schedule it alongside sampling, not after.

Beyond IMDS, European programs ask for REACH and RoHS evidence, with the EU SVHC candidate list now numbering past 200 substances and growing each revision. The molder tracks the list against the resin data. The mold file stays lighter here, but the discipline is the same: chemistry evidence is traceable by lot, and a certificate without a lot reference does not answer an auditor's question.

What the Mold Contributes: Repeatability and Window

The strongest mold-side evidence is repeatability across two separate runs. If the same cavity produces the same average dimension on two different days, the tool is stable and any remaining variation belongs to the process. DieStrike structures trial sampling so customers can compare run one and run two directly, with cavity-by-cavity averages in the dimensional report.

Cavity spread is the second witness. A 16-cavity connector mold that holds every cavity within 0.02 to 0.05 mm of the same nominal is a mold that will pass a capability study. A single cavity sitting apart from its neighbors is a tool problem, usually a cooling circuit or a gate land difference, and it must be fixed before the submission run rather than explained after it.

The process window is the third. A mold approved at one lucky parameter set fails the moment production drifts. Sampling should confirm the mold holds dimension and appearance across the agreed window, and the mold maker's trial record should show the window edges, not just the center point. Thermal balance belongs in the same record, because cavity temperature differences above a few degrees Celsius shift dimensions cavity to cavity.

Steel quality underlies all three. Cavities machined to ±0.005 mm with geometry features at ±0.002 mm, hardened to HRC 62 and thermally balanced, behave predictably run after run. Predictable behavior is what converts a good-looking ISIR into a passing capability study, and it is why the mold file, not just the part file, belongs in the submission evidence chain.

VDA, CSRs and the German OEM Difference

Not every automotive customer runs AIAG PPAP. German OEMs and their tiers operate under VDA 2, which structures production process and product approval as a PPA, with its own report forms, evidence hierarchy and terminology. An AIAG Level 3 file does not automatically satisfy a VDA-based customer, and vice versa.

VDA programs tend to place heavier weight on process capability for machine and process alike, with Cmk and Cpk values, sampling plans and approval stages defined in the customer manual. The dimensional and material evidence underneath is largely the same data, reorganized into a different format. A mold maker that can supply clean ISIR-level dimensional records with full GD&T context serves both systems without extra rework.

The rule that overrides every manual is the customer-specific requirement, or CSR. CSRs routinely tighten capability values to 1.67, change submission levels, demand specific IMDS timing or add appearance approval steps. Read the CSR before the sampling run, list its deltas against the AIAG default, and confirm the study plan against that list. Sampling to a guessed requirement is the most expensive guess in the PPAP timeline.

Who Owns Each Piece of PPAP Evidence

Submission failures often trace to ownership gaps: both sides assumed the other would produce a certificate, a study or a declaration. The table below assigns each evidence item to its primary owner and shows what a mold maker contributes from the sampling floor.

← swipe to scroll →

Table note: owners and typical values follow AIAG practice and common CSRs. Confirm every line against the customer's own requirements before the sampling run.

Assign a named person to each row before the mold ships. When the capability study is due in week 5, the molder's process engineer, the quality lab and the mold maker's sampling team should already know who holds the data and who signs the verdict. Ownership agreed in writing at quotation closes faster than ownership discovered at the audit.

Timeline: Fitting PPAP Sampling Into Mold Delivery

PPAP sampling sits at the end of the mold delivery chain, and it inherits every delay earlier in the chain. A typical precision injection mold from DieStrike ships in 2 to 5 weeks, with DFM feedback inside 24 hours and standard mold components such as ejector pins, core pins and mold bases shipped in 3 to 7 days when spares are needed. Sampling is scheduled against that delivery, not after it.

Sequence the work so evidence is produced once. T1 dimensional verification confirms the tool against the drawing, and the same CMM data feeds the ISIR when the customer's template is applied. The capability run follows tool approval on the production press. From tool release to a complete submission file, plan for 2 to 4 weeks of sampling, study, and documentation, and confirm the customer's PPAP due date against the start of production before the mold order is placed.

Keep the correction loop inside the schedule. When sampling exposes a tool issue, the mold maker's own machining floor should be able to fix it without a second logistics chain. DieStrike's 120+ machines keep that loop internal, and spare inserts cut from the original CAM programs let a worn cavity be replaced without re-qualifying the whole tool. A sampling plan that budgets one correction round, typically 3 to 7 days with standard components on hand, survives launch schedules; a plan that budgets zero does not.

Frequently Asked Questions

Q1. Does the mold maker submit the PPAP, or does the molder?

The molder submits and signs the part submission warrant, because PPAP approves the production process that includes press, parameters and operators. The mold maker supplies the tool-side evidence: ISIR-level dimensional data, material certificates, and repeatability records. A clean handoff between the two is what makes a submission pass in one round.

Q2. Is the 300-piece sampling run mandatory?

No. It is a common automotive convention because it exposes drift and wear that short runs hide, but the binding number comes from the customer-specific requirement and the agreed control plan. Some CSRs specify 100 parts, others 500. Confirm the count in writing before the run starts.

Q3. What happens when Cpk comes back below 1.33?

The submission cannot proceed at that value, so the cause is split first. A wide spread points to process noise or tool instability, and an off-center mean points to shrinkage compensation or setup. The molder fixes the process share, the mold maker fixes the tool share, and the study is re-run. Budget one correction round in the launch plan.

Q4. What is the difference between the mold trial's first article and the PPAP ISIR?

The trial first article approves the tool at T1 with a single article against the drawing. The ISIR is the submission-level document set built from the approved tool, covering multiple samples per cavity, material evidence, capability data and the customer's specific format. The same CMM data can feed both, but the ISIR template and sample counts usually differ.

Q5. Can a mold maker provide ISIR-level dimensional data on molded samples?

Yes, when the trial sampling is structured for it. As an IATF 16949-certified mold maker, DieStrike measures molded samples on CMM equipment at ±0.005 mm accuracy, reports as-built cavity dimensions, and hands the molder data formatted for its submission template. The dimensional file is complete before the mold leaves the plant, which is why our trial sampling service is specified in the mold order rather than discovered later.

The Bottom Line

PPAP sampling converts a good mold into an approved part number, and the conversion lives in evidence, not in shots fired. ISIR layouts, certificates, capability and gage data must all trace to one calibrated measurement source and one agreed plan.

Agree the submission level, sample counts and Cpk gates before the mold order, and assign each evidence item an owner. DieStrike supports the file as an IATF 16949-certified mold maker with ISIR-level dimensional data, and our DFM response lands within 24 hours. If you are planning a sampling program, tell us the customer requirement and the CSR; we will structure the trial data so your submission survives the first review.