← Back to technical FAQ
Anjelio Automation · Technical article

Multi-SKU Changeover on a CCD Vision Labeler — What Takes Time

What actually drives changeover time on a CCD vision labeler—fixtures, vision re-teach, materials—and what to ask before RFQ.

Author: UAT Administrator · Published: 2026-09-23T07:23:20.834132296+00:00 · Substantively updated: 2026-09-23T07:23:20.834625776+00:00

Definition (quote-ready): On a CCD / vision-guided labeler, multi-SKU changeover usually means three layers—fixture/nest, vision recipe (recall or re-teach), and a short sample validation before the line is released. How long that takes depends on how different the next SKU is, not on a single catalogue hour. Anjelio locks fixture kit, recipe plan, and any changeover acceptance criteria only in a mutually confirmed technical proposal after samples.

Buyers in high-mix electronics and connector cells commonly ask about 换产 (changeover) burden—fixture swaps and debug time—alongside catalogue specs. That is market intent framing, not a timed Anjelio case study. Anjelio (anjelio.com) supplies industrial vision label application; confirm domain and model codes such as VL-SS before RFQ (brand note).


What “changeover” includes (three layers)

Changeover is not “fixture swap minutes” alone. Treat it as a stack:

Fixture / nest

Mechanical locate tooling that holds or indexes the board or part for place. You may swap a dedicated nest, adjust a family fixture, or share tooling across similar outlines. Large geometry jumps usually need dedicated nests; similar outlines may share adjustable family tooling. Fixture strategy is a design choice in the proposal—not a free universal kit.

Vision recipe

Saved camera/locate parameters for a SKU: which feature to find, ROI, lighting notes, and place offset relative to that feature. On a clean switch you recall a known-good recipe and confirm. When geometry, finish, or lighting changed enough that recall fails, you re-teach locate and sometimes place offset.

Sample validation

Agreed sample parts run after the mechanical and vision steps. This is the usual gate that ends “debug” before production release. Live Anjelio PDP language already ties fit to samples: engineering confirms location, tooling, and rate from samples; catalogue rates are not a substitute for a trial; final scope follows the written technical proposal.

Reserve the word re-calibration for cases where camera/robot frame or tool offset is formally re-established. Do not treat every SKU switch as a full frame re-calibration.


Do you always change fixtures when the SKU size changes?

Not always.

SituationTypical fixture approach
Large outline / thickness jump, tight place windowDedicated nest per family or SKU
Similar outline, different label or featureAdjustable or shared family fixture + recipe change
Same nest geometry, only label artwork changesOften keep nest; update recipe / material path

High-mix (高混线) cells care about this as much as peak UPH. The proposal should list which SKUs share tooling and which need a kit change—before anyone argues about stopwatch times.


When vision needs a re-teach (and when a saved recipe is enough)

ConditionUsual vision action
Feature still clear, lighting and nest stableRecipe recall + confirm place on samples
Feature moved, finish/contrast changed, nest or lighting shiftedRe-teach locate and/or place offset
Camera/robot frame or tool offset in doubtFormal re-calibration only if that scope is in the proposal—not every SKU by default

Recipe libraries and operator training shrink the “mystery debug” buyers worry about. They do not create a universal minute count for every plant.


What actually drives changeover duration

Honest drivers—none of which become a website guarantee:

  1. SKU similarity — how far size, nest, feature, and label jump from the last job
  2. Fixture kit readiness — dedicated nests staged and labeled vs hunting hardware mid-shift
  3. Recipe library quality — named, versioned recipes vs ad-hoc re-teach every time
  4. Operator steps — who may change nests, who may re-teach, who releases the line
  5. Acceptance definition — what “done” means (fixture checklist, recipe checks, sample place criteria)

That is why a catalogue page cannot publish a fixed changeover hour without your SKU set. Industry articles that discuss connector multi-SKU changeover or whether acceptance should include debug time reflect buyer questions—not Anjelio timed results. Ask for changeover criteria to be written into the mutually confirmed technical proposal (and into FAT only when the proposal says so).

Related but different decisions:

This page stays on fixture + recipe + validation.

Catalogue rates (for example VL-SS 1200–2000 labels/h, subject to sample validation) describe throughput under stated conditions. They are not a changeover-time proxy. The VL-SS PDP also asks you to confirm changeover needs before sample validation—bring that into the RFQ explicitly.


What to send so we can propose a multi-SKU plan

First-look families when nest or line flow shapes the plan: VL-SS, VL-DD, VL-EM, VL-TR; VL-4A only when pose/reach drives nest strategy. Full catalogue: /products. Vision labeling basics: I1.

Final configuration—including fixture kit count, recipe count, and agreed changeover acceptance criteria—is subject to the mutually confirmed technical proposal and quotation.

Frequently asked questions

How long does multi-SKU changeover take on an Anjelio vision labeler?

There is no honest one-number answer on a catalogue page. Duration depends on fixture strategy, vision recipe readiness, SKU similarity, and the sample-validation gate you agree in the proposal. Do not treat industry case minutes as Anjelio guarantees. Ask for changeover criteria in the mutually confirmed technical proposal.

Do we have to swap fixtures for every size / SKU?

Not always. Large geometry jumps often need a dedicated nest; similar outlines may share an adjustable or family fixture. The proposal should list which SKUs share tooling and which need a kit change.

Do we re-calibrate the vision system on every changeover?

Usually you recall or re-teach a recipe for the next SKU’s feature and place offset. Formal camera/robot frame or tool-offset re-calibration is a different, less frequent event—confirm scope in the proposal rather than assuming it every switch.

Can changeover time be part of acceptance?

Yes—if both sides define what “done” means (fixture steps, recipe recall checks, sample place criteria). That belongs in the mutually confirmed proposal / FAT language when written in, not as a website stopwatch claim.

How is this different from choosing heads (I3) or RFQ scope (I9)?

I3 picks config family (single/dual/robot, etc.). I9 covers commercial quote scope and lead-time start when published. I12 is the high-mix operations question: fixture + vision recipe + validation when SKUs change. Use I8 if you are still choosing inline vs offline layout. MES / Mark / protocol depth stays out of this page (I7).

Related products

Need a recommendation for your product, label, or line speed?
Talk to Anjelio engineering