Backlash vs Repeatability: How to Set Acceptance Criteria Before PO

Many sourcing teams still treat backlash as a single proxy for axis quality. That shortcut looks efficient early and becomes expensive during pilot-to-MP transition.

Backlash and repeatability are related, but they answer different engineering questions and need separate release evidence.

Executive Summary

Use this three-gate rule before PO:

1. method lock gate: one protocol revision shared by both sides
2. feasibility gate: target achieved on representative samples
3. stability gate: variation controlled across lot samples

If any gate is missing, "pass" results are not dependable for production commitment.

1) Define both metrics as separate acceptance items

Use two lines in your RFQ, not one merged statement.

• backlash: directional lost motion under a defined reversal method
• repeatability: position return consistency under defined load, cycle, and thermal state

If one metric is explicit and the other is implied, supplier and buyer may both report "pass" on different test contexts.

2) Lock the measurement protocol before sample approval

At minimum, freeze these fields:

• method ID and test sequence
• fixture boundary and mounting condition
• preload and output-side load condition
• measurement instrument class and resolution
• thermal state and warm-up rule
• sample size and data-record format

Without a frozen protocol, two labs can produce numbers that are both valid but not comparable.

3) Method lock flow (to prevent post-PO disputes)

4) Two-layer acceptance model

Layer A is design feasibility. Layer B is process stability.

Do not release recurring PO based on Layer A only.

5) Practical protocol template (copy into your test sheet)

This table removes ambiguity before the first dispute appears.

6) Worked example: method mismatch that reversed ranking

Illustrative buyer-side case:

Initial supplier reports:

• Supplier A backlash: 7 arcmin
• Supplier B backlash: 6 arcmin

Procurement assumption: B is better.

Method details discovered later:

• A used reversal travel 1.0 degree with 0.5 Nm preload at 25 C
• B used reversal travel 0.2 degree without preload at 35 C

After unifying method to one locked protocol:

• A measured 7.4 arcmin with stable repeatability spread
• B measured 8.1 arcmin with wider repeatability spread

Decision changed from B to A because method context was normalized.

7) Decision thresholds and release logic

Use explicit release logic in your internal gate:

1. method lock pass: all protocol fields frozen and signed
2. feasibility pass: target met on engineering samples
3. stability pass: lot variation within agreed window
4. deviation governance: any protocol change requires written re-approval

If step 1 is missing, steps 2 and 3 are weak evidence.

8) Threshold matrix with owners

9) Commercial clause that prevents re-interpretation

Include one concise clause in commercial terms:

Acceptance for backlash and repeatability is valid only under Method [ID-REV],
with defined load, temperature, sample size, and instrumentation.
Any change to method or boundary conditions requires written re-approval before shipment release.

This avoids late claims based on changed test context.

10) Buyer checklist before PO release

• protocol ID and revision signed by both sides
• sample size and lot rule approved
• pass-fail threshold documented with units
• data-record template and archive ownership defined
• nonconformance response SLA agreed

If any line is missing, schedule risk remains high even after sample pass.

11) Related pages for pre-check

• Product Families
• Application Notes Hub
• OEM Quality and Delivery Control

For project-specific review, start from Contact / RFQ.

12) 20-minute acceptance meeting script (copy template)

Use this script before PO release to avoid test-method disputes.

Meeting title: Backlash and Repeatability Acceptance Freeze
Project:
Protocol ID-REV:
Chair:
Date:

Agenda (20 min)
1) Protocol completeness check (5 min)
- Method ID, load, temperature, sample size, instrument all present? Y/N

2) Feasibility result review (5 min)
- Engineering sample results:
- Any boundary deviation:

3) Stability evidence review (5 min)
- Lot trend:
- Variation summary:

4) Decision and actions (5 min)
- PO gate decision: GO / NO-GO
- Re-test needed: Y/N
- Owner and due date per open item

Final record:
- Signed by buyer quality:
- Signed by supplier quality:

13) Field Notes from Buyer Calls (Anonymized)

Q: Supplier says backlash passed, so repeatability should also pass. Is that valid?

Not necessarily. Backlash and repeatability respond to different test boundaries and should be accepted as separate items.

Q: Two labs produced different numbers. Which one is correct?

Both can be correct under different protocols. The useful question is whether both used the same method revision, load, and instrument class.

Q: We are under delivery pressure. Can we release PO before lot-level stability data?

Only if PO terms explicitly limit release scope and require stability closure before recurring lots.

14) Anti-Patterns to Avoid

• collapsing backlash and repeatability into one acceptance sentence
• approving sample data without method revision signatures
• starting recurring shipments before lot-level variation evidence is reviewed

15) Failure Postmortem: Sample Passed, Recurring Lots Failed

Observed pattern from a transfer-axis program:

• engineering sample passed backlash check
• recurring lot showed wider position variation during warm operation
• supplier and buyer both claimed compliance using different method settings
• PO dispute escalated because protocol revision was never co-signed

What fixed the issue:

• one method ID-REV shared across both labs
• mandatory lot-level repeatability trend review before recurring release
• clause added to block method changes without written re-approval

Sources and Last Verified

• ISO 230-2:2014 - Test code for machine tools, Part 2: Determination of accuracy and repeatability of positioning
• AIAG Quality Core Tools (APQP, PPAP, FMEA, MSA, SPC)
• ISO 9001:2015 - Quality management systems requirements

Last verified: May 11, 2026.

Final CTA

If you want a program-level review, email [email protected] or message WhatsApp +8618857971991.

To get a usable first response, include:

• application and axis function
• duty cycle and ambient conditions
• drawing revision and interface constraints
• target timeline and forecast quantity

FAQ

Are backlash and repeatability the same thing?

No. Backlash describes directional lost motion, while repeatability describes position consistency across cycles.

Why do disputes happen after sample approval?

Because method and boundary conditions are often not frozen, so results are not directly comparable.

What should be fixed before PO release?

Freeze method ID, instrumentation, sample size, thresholds, and nonconformance ownership.

What is the practical release trigger for PO?

Release only when method lock, feasibility, and stability gates are all passed under the same protocol revision.