cnx_vinyl/docs/quality-automation.md

# Quality Automation

The target is to automate as much as possible. For quality control this
means instruments and inspection stations that measure, decide, and log into
the [traceability system](traceability.md) without manual transcription.
Research findings from vendor sites and trade sources, July 2026; the
recommended architectures are engineering positions to be validated with
vendors, not purchased designs.

## Stamper, father, and mother metrology

What must be measured on the electroformed nickel parts (~0.2 to 0.3 mm
thick): groove geometry (grooves are 40 to 80 microns wide, ~28 microns
deep, 200 to 350 grooves per inch, inverted to ridges on a stamper),
thickness uniformity (total thickness variation of 10 to 20 microns is the
achievable target, by analogy with CD stamper practice), surface roughness
on the groove walls, and micro-defects.

A Keyence-class instrument stack covers this, in tiers:

- **Keyence VK-X 3D laser confocal microscope**: 20 nm vertical and ~120 nm
  lateral resolution, non-contact. The reference-grade instrument for groove
  cross-sections, wall roughness, and fine defect analysis. Scanning the
  full spiral (hundreds of meters of groove or more) is not feasible; the
  correct strategy is **sampled radial zones** (lead-in, outer, mid, inner
  radius). The optical disc industry verified its nickel stampers with the
  same instrument class, so the technique transfers directly.
- **Keyence VR-6000 wide-area profiler**: 0.1 micron height resolution,
  ~1 second per field, with an HDR mode built for glossy surfaces (bright
  nickel is near-mirror, the classic hard case for optics; manageable, but
  plan tuning time). Fast wide-area form, flatness, and gross defect
  mapping.
- **XRF thickness gauge** (Fischer/Bowman class): non-destructive nickel
  deposit thickness per serial, replacing the industry's manual 10-point
  micrometer practice. Thickness must be read on the outer land/shoulder
  zones, not across the groove field.
- **Stylus profilometer** kept as the traceable calibration reference.
- **Inline machine vision** for fast 100% cosmetic screening of every part.

Integration is real but not turnkey: the Keyence instruments export CSV and
raw 3D data and support scripting/REST layers, and the inline sensors speak
Ethernet/IP, but there is no off-the-shelf MES connector. Budget a light
middleware service that stamps each measurement with the part serial and
batch ID and writes to the genealogy database.

### Part marking and orientation fiducials

Every metal part gets a machine-readable identity, using two complementary
marks:

- **Dead-wax marks**: matrix codes scribed in the runout area are
  established industry practice and propagate through the whole chain
  (father to mother to stamper to every pressed record). Adding a small
  **angular fiducial** at a defined position in the dead wax costs nothing
  and makes the part's rotational orientation camera-readable at any stage.
  Mothers are playable positives and can be scribed directly at their
  generation.
- **Back-side laser marks**: a serial or Data Matrix code lasered on the
  stamper back (outside the pressing surface) identifies the physical part
  for metrology and genealogy logging, exactly as the optical disc industry
  marked its stamper backs. This mark does not propagate to records; it
  pairs with the dead-wax identity.

**Marking method per material** (safety-driven, not preference):

- **Lacquers: mechanical cold scribe only, never laser.** Nitrocellulose is
  highly flammable and thermally sensitive, and the lacquer is the audio
  master; a stylus scribe in the dead wax is the century-proven method and
  can be automated with a small CNC scriber if machine-consistent marking
  is wanted. The lacquer is destroyed when the father is separated, so its
  identity lives on through the propagated scribe, not the part.
- **Nickel parts: laser marking, freely.** Fiber-laser serials and Data
  Matrix codes on the backs, per optical-disc practice.
- **Pressed records: no laser marking.** Laser-burning PVC releases
  hydrochloric acid fumes, a worker-safety and equipment-corrosion problem.
  Records already carry the pressed-in dead-wax marks from the stamper, and
  per-record uniqueness comes from the traceability database (press, cycle,
  run position) rather than physical marks.
- **Labels and packaging: printed or inkjet codes**, which the packaging QC
  barcode checks already assume.

The galvanic process itself is rotationally symmetric, so orientation never
needs to be locked through the baths, only tracked, and the fiducial does
that. The fiducial also gives the inline optical inspection a fixed angular
anchor for reading matrix codes on finished records, and it would make
label-to-groove rotation measurable per record. Actively aligning labels at
the press (vision-guided rotation in the label feed) is feasible but custom
and currently deprioritized; the fiducial keeps that option open without
committing to it.

### Industry context

Even a respected stamper house checks thickness with a hand micrometer at
10 points on sacrificial blanks, with nothing logged.
Automated non-contact metrology with per-serial logging is ahead of current
industry practice, which is exactly the point: it makes every father, mother,
and stamper a data-backed accept/reject decision before the next plating
generation or press mounting.

### Hardness on the centre slug

From optical-disc practice, per the 2008 Blu-ray stamper working-group
paper in
[`equipment/sibert/sps-optical-disc-2009/`](equipment/sibert/sps-optical-disc-2009/):
stamper hardness (micro-Vickers, HV0.3) is measured on the disc punched out
of the centre, never on the part, and a freshly made-up nickel bath plates
soft at first (from ~150 HV0.3 upward against a ~200 HV0.3 target for
Blu-ray), which changes how stampers punch and finish. The vinyl centre
punch (Sibert VSP) leaves one slug per stamper; marked with the part
serial, it is a free per-stamper coupon for hardness and thickness, logged
against the bath ID. The paper's limits are for Blu-ray and do not
transfer, but its structure (thickness, flatness, hardness, back
roughness, centre hole, eccentricity, burr, visual defects, each with a
method) is the model for the stamper acceptance spec this plant still has
to write. Equipment context in
[galvanics and stamper prep](equipment/galvanics.md).

## Pressed record QC: three layers

Full automated audio checking of every record is physically impossible: a
12" record plays for ~40 minutes both sides while a press makes a new record
every ~25 seconds, so one press outruns roughly 80 to 95 full-time listening
stations. No commercial playback-QC robot exists anywhere. The credible
maximum-automation architecture is layered:

### Layer 1: 100% optical inspection, inline

Machine vision can inspect every record at line speed. Detectable with
darkfield lighting and line-scan cameras: non-fill (dull patches where PVC
did not fill the stamper), stitching and flow lines, warp (via structured
light or profilometry), off-center spindle holes (concentricity of hole vs
groove spiral), surface debris, and label errors (wrong label, off-center,
print defects). Non-fill and stitching correlate with audible defects, so
optical inspection is a strong proxy for audio quality without playing the
record.

No vinyl-specific vision product exists (search of October 2026); this is
a custom integration of general machine vision at the press unloader, on
the pattern of the old CD and DVD in-line scanners: one turn under a
radial line-scan camera with angled lights, plus laser distance sensors
for warp. The closest vinyl precedent is Precision Record Pressing's
in-house AQA (2023), which scans sampled records with cameras in the lid
of modified turntables before playing them. CNX would be an early
adopter, and it should be budgeted as an integration project. The device
options, a station design, prices and a cost estimate are in
[inline optical QC](equipment/inline-optical-qc.md).

### Layer 2: press telemetry, every cycle

The proven, deployed piece. Viryl's ADAPT platform demonstrates the model:
per-cycle nozzle pressure, mold temperatures, cycle time, cake weight,
yields, and drift alerts streamed from the press. The Pheenix AD12 runs a
Beckhoff PLC exposing similar data over standard industrial protocols, and
supports remote monitoring. Machine data access is already flagged as a
procurement requirement in the traceability spec.

### Layer 3: automated audio comparison, sampled

A needle-drop capture station with automated analysis, run on samples:
every test pressing / first-off per stamper, plus pulled samples scaled by
stamper age (stampers last roughly 1,000 to 2,000 records; sampling should
densify near end of life; industry rule of thumb is a spot-play around every
50th record).

The analysis pipeline is buildable from proven parts but is a first-of-kind
system as a whole:

1. Capture the needle-drop (a conventional tonearm; laser pickups exist but
   are dust-sensitive and impractical as line instruments)
2. Spectral-envelope comparison against the source master (the approach
   proven by Stamper Discs' Fonograph tool, which avoids the hard
   time-alignment problem by comparing whole-program FFT fingerprints)
3. Click/pop detection and counting (mature algorithms of the iZotope RX
   class, scriptable)
4. Noise floor from lead-in/run-out silence, channel balance from program
   RMS
5. Numeric scores and pass/fail to the traceability database

What only this layer can catch: wrong master cut into good-looking grooves,
EQ/mastering errors, and speed/pitch problems. Optical and audio layers are
complements, not substitutes.

## Raw PVC compound QC, per lot

Every incoming compound lot passes a gate before release to production, and
the lot ID follows every record pressed from it:

- **Paper first**: supplier certificate of analysis per lot, checked against
  the agreed specification; a retained sample of every lot is archived
- **Moisture**: verify the compound is within the moisture spec before it
  reaches the extruder (moisture causes surface defects and noise); a
  benchtop moisture analyzer is a cheap, automatable check
- **Melt behavior and thermal stability**: a melt-flow or rheology check per
  lot catches drift in processing behavior, and a heat-stability check
  matters specifically for PVC (degradation releases hydrochloric acid and
  shows as discoloration and noise); these are standard plastics-lab
  instruments with data export
- **Color and appearance**: spectrophotometer reading against the reference
  for black and colored compounds; granulate size and dust content by
  sampling
- **Release by test pressing**: a new lot, and especially a new supplier, is
  released only after test pressings from that lot pass optical and audio
  QC; this is the backbone of the local-supplier qualification protocol,
  since qualifying Thai/regional compound against European references is an
  explicit goal
- **Regrind stream QC**: the in-house 20% recycled stream is its own
  material source and needs its own gate: no labels, paper, or foreign
  material entering the grinder, contamination checks on regrind batches,
  and verification of the virgin/regrind ratio at the extruder; regrind
  batch IDs are logged like any other lot

### The PVC lab, in tiers

Prices are rough list-price ranges per instrument class (estimates, not
quotes; ask Thai distributors for local pricing and service).

**Per lot, in-house from day one** (benchtop instruments, all with data
export for the traceability gate):

| Check | Instrument | Estimate, USD |
|---|---|---|
| Moisture, minutes per sample; the most common cause of noisy surfaces | Halogen moisture analyzer (Mettler HC / Sartorius MA class) | 3k to 8k |
| Heat stability: oven strips at pressing-relevant temperatures, watching discoloration over time | Lab oven with timed strip racks | 3k to 8k |
| Density against the compound spec; feeds record weight consistency | Analytical balance with density kit | 4k to 8k |
| Color L\*a\*b\* against the reference lot | Spectrophotometer | 10k to 30k |
| Gloss on black compound | Gloss meter | 2k to 5k |
| Granulate physicals: particle size distribution, bulk density, dust and fines | Sieve shaker with sieve stack, bulk density cup | 3k to 8k |
| Ash content as a filler-loading drift check | Muffle furnace | 3k to 8k |
| Specks, gels and contamination on pressed plaques | Stereo microscope and light table (the method is shareable with the optical inspection investment) | 3k to 8k |
| Test plaques for the speck count, color and DSC, without tying up a production press | Lab hot plaque press or small two-roll mill | 20k to 60k |
| Sample conditioning (below) | Drying oven, desiccators, sealed retained-sample storage | 2k to 5k |
| Metal contamination in lots and regrind | Grate or bar magnets at the hopper and regrind intake | under 2k |
| **Day-one tier** | | **~55k to 150k** |

Two optional upgrades to the day-one tier: Karl Fischer titration for low
moisture levels and supplier disputes (10k to 20k), and the formal heat
stability method, ISO 182 Congo red / conductivity tracking of HCl
evolution (Metrohm PVC Thermomat class, 20k to 35k).

**Sample conditioning is part of the method, not an extra.** At 60 to 90 %
ambient humidity granulate picks up moisture between the warehouse and the
balance, so a moisture result is only comparable if the sample was taken
sealed, conditioned the same way every time and tested within a fixed
window. The lab sits in the conditioned QC zone ([scale](scale.md)) at the
HVAC's controlled humidity, and the retained sample of every lot is stored
sealed.

The plaque press makes the inspection plaques; the **release test
pressing** of a new lot still runs on a production press, because that is
the real test.

**In-house as the lab matures** (the PVC-specific workhorses, ~USD 120k to
240k together):

- Torque rheometry (Brabender class, ~60k to 120k): fusion/gelation time
  and torque, the standard predictor of how a PVC compound behaves in the
  extruder; plain melt-flow index is less informative for PVC than for
  other thermoplastics because the material is shear- and heat-sensitive
- DSC (~40k to 80k): glass transition, additive signatures, and the
  record-relevant application of measuring the degree of gelation of a
  pressed record, which correlates with surface noise and durability,
  turning DSC into a process-tuning tool as well as an incoming check
- FTIR with ATR (~20k to 40k): identity fingerprint; record compound is
  classically a vinyl chloride/vinyl acetate copolymer rather than pure
  PVC, and FTIR verifies the copolymer and catches unannounced formulation
  changes and foreign polymers in the regrind
- Option: handheld XRF (~25k to 40k) brings stabilizer and heavy-metal
  screening in-house for every lot instead of per supplier, worth it while
  Thai and regional compounders are being qualified

**Partner labs before buying the mature tier.** Pharmacy and science
faculties routinely run FTIR, DSC and Karl Fischer; Payap University's
Faculty of Pharmacy ([Payap partnership](payap-partnership.md)) and Chiang
Mai University's science faculties are the obvious places to ask
(unconfirmed which instruments they hold). Contract testing there defers
most of the mature tier until lot volume justifies owning it.

**External lab, per supplier qualification** (not per lot):

- K-value / dilute solution viscosity (ISO 1628-2): molecular weight, the
  headline spec for record-grade resin since it governs flow into the
  groove
- XRF elemental analysis: stabilizer system identification (Ca-Zn or tin
  acceptable, legacy lead rejected) and heavy-metals compliance for export
- TGA: full compositional breakdown (plasticizer, filler, carbon black)
- Volume/surface resistivity: static behavior, which becomes dust
  attraction on finished records

The tiers define the protocol: the per-lot gate runs moisture, heat
stability, color, density, and granulate physicals plus the release test
pressing; a new supplier additionally gets the full external panel and
pressing trials. This is the working method for qualifying Thai and
regional compounds against the European references. The day-one tier is
costed in [total investment](finance/total-investment.md) as added scope.

## Packaging QC

Packaging comes from local carton partners, so QC splits into an incoming
gate and an assembly gate:

- **Incoming cartons and sleeves, per delivery**: dimensional checks
  (records must fit without force; spine width per title), board grammage
  and thickness, print quality against the customer-approved proof
  (spectrophotometer Delta E for color-critical jobs), lamination/finish,
  and seam strength; results logged per packaging lot and fed back to the
  partner
- **Inner sleeves**: verified for scuff risk (surface finish matters more
  than looks; a bad inner sleeve scratches every record put in it)
- **Print inspection is automatable**: camera-based print inspection is a
  mature field, and the same vision investment as the record label check
  can cover jacket print defects
- **Assembly verification, automated**: the highest-value check is that the
  right record goes into the right jacket; scanning the runout matrix code
  against the jacket barcode at the packing station makes mismatches
  systematically caught rather than unlikely, and confirms inserts and
  extras per the order's bill of materials
- **Barcode verification** on every finished unit (a misprinted barcode is
  a rejected delivery at the distributor)
- **Shrink wrap and final visual** on a sampling basis

Both gates write to the same traceability chain: compound lot and regrind
batch attach to every press cycle, packaging lot and assembly checks attach
to every order.

## The traceability payoff

The natural record identity is **stamper ID + press ID + cycle timestamp +
side**. Against that key: press telemetry on every cycle, an optical result
on every record, an audio score on every sampled record, and the stamper's
own metrology history from before it was mounted. Audio scores trending down
against a stamper's play count become an automatic retirement signal, and a
customer complaint about one record resolves to a specific stamper, press,
shift, and compound lot.

## Honest limits

- No vendor sells a turnkey version of layers 1 or 3, or the metrology
  middleware; all three are integration/development projects
- 100% audio QC of every record is not achievable at press speed and should
  never be promised to customers; 100% optical plus sampled audio is the
  accurate claim
- Human listening stays in the loop for test-pressing approval; the
  automation makes sampling systematic and logged, it does not replace ears