Fully automatic cell-to-glass metrology
Cs-Veno is a Windows machine-vision QC station that runs completely hands-off: as each module's barcode is read, it automatically measures the gap from each cell's edge to the glass edge across as many camera points as your module needs — one camera per spot, no eight-point cap — then logs the results to the Database with traceability photos. Operators never scan, type, or enter a thing. It's the industry's first station to measure these gaps before lamination, on any material and any color — backsheet, glass, encapsulant or cell finish — something no one else can do accurately.
The cost of being late
Cell-to-glass-edge and cell-to-cell spacing have to stay within roughly ±1 mm — for safety certification and for reliability. Once a misaligned layup clears the laminator, the EVA/POE cross-links irreversibly and the glass, cells and backsheet fuse into one monolithic laminate. There is no clean way to separate the layers, so a defect that costs cents of operator time to correct on the glass becomes a fully scrapped module the instant it's cured.
IEC 61730-1:2023 references a minimum creepage on the order of 12.5 mm between live cells and the module edge, and Class II 1000/1500 V designs can demand 20–30 mm. A cell that drifts too close to the glass edge can fail high-voltage clearance requirements outright.
Strings sitting just 2–3 mm off nominal can leave the encapsulant unable to flow and seal along the edges. That under-seal is a primary precursor to edge delamination and moisture ingress — effectively irreversible once it starts, and a warranty liability years later.
A scrapped finished module isn't cheap encapsulant — it carries cells (over 40% of the BOM), tempered front glass (~20%), plus all the labor and energy committed through lamination. The same defect costs cents upstream and the full committed value downstream.
A line running 40–200 modules/hour under 90-second takt can't be hand-measured at every cell. Manual spot checks sample a handful of modules, so the marginal, creepage-failing units are exactly the ones that slip through to the field.
How it works
Cs-Veno drops inline at the layup-to-lamination handoff and runs completely on its own. No operator scans a code, types a number, or makes a judgment call — the station reads, measures, logs and photographs every module automatically.
As each module reaches the station, its barcode is read automatically and Cs-Veno arms a configurable export countdown (default 7.0 seconds) so the module can settle. On a new barcode it re-sends camera exposure to clear stale edge history and stabilize the reading. No one has to scan or type anything.
All enabled camera points continuously measure cell-edge-to-glass (or cell-to-cell) distance. Sub-pixel edge localization fits a parabola to refine each edge to a fraction of a pixel, and a rolling-median stabilizer keeps the live value steady under vibration, glare and motion-blur.
When the countdown elapses, the latest per-point measurements are upserted as one row per barcode into the Database — keyed by time and barcode, with any prior row for that barcode preserved in a duplicate-history table.
At the same instant, a photo from each camera is saved into a per-barcode folder for traceability. If enabled, optional Govee LED strips flash green when every enabled Cell-To-Glass point clears the pass threshold, red otherwise, and per-point OK/NG counters update on the HMI.
Capabilities
Add a camera at every spot that needs checking — Point One, Point Two, and onward. There's no fixed eight-point cap: each point gets its own enable toggle, calibration, ROIs and measurement, so the station follows your module layout instead of forcing it into a set number of positions.
Each point's Camera Function selects Cell To Glass (cell edge to the first black line / glass edge) or String To String (cell edge to the opposing cell edge), with selectable edge layouts for left/right or bottom/up orientations.
A parabola fit refines each vertical edge to a fraction of a pixel, sampled across roughly nine rows of the measurement band and combined with a median, so grainy or speckled glass edges resolve precisely. Refinement is safety-clamped (default max 0.60 mm correction).
The displayed value uses a rolling median over 15 frames, rejects outliers beyond 2 mm, limits per-frame movement to 0.35 mm, and holds through missed frames — steady under shake and motion-blur while still letting real module changes settle.
An onboard AI model learns the true glass edge from your own modules and reliably separates it from EVA, encapsulant and backsheet — so the measurement locks onto the real glass edge even when EVA is squeezed over it. It adapts to your materials and lighting instead of relying on brittle fixed rules.
One upserted row per barcode lands in the Database with prior rows kept in duplicate history, and a photo from each camera is saved to a per-barcode folder. Re-measures archive the previous photo so nothing is ever lost.
Where the money comes back
Scrap prevention is the dominant lever — every misaligned layup caught before the laminator avoids scrapping a full finished module. Labor displacement is real and recurring on top of it. The calculator below foregrounds both and lets you plug in your own numbers.
Catch a misaligned layup on the glass and it costs cents to rework; let it pass the laminator and it's a full scrapped module dominated by cell and glass value. Preventing even a fraction of a percent of modules per year typically dwarfs every other line item.
Manual checks sample a few modules; Cs-Veno measures every enabled cell-edge and glass-edge on every module, converting an unknown escape rate into a controlled one and catching the marginal, creepage-failing units sampling statistically misses.
Remove the dedicated caliper/ruler QC role — or the portion of operator time spent measuring — at the layup-to-lamination handoff. At a fully loaded operator rate, even a partial FTE redeployed across the year is real recurring savings.
Cs-Veno measures cell-to-glass and cell-to-cell spacing right at the layup — before lamination — on any material and any color: any backsheet, glass, encapsulant or cell finish. No other system measures these edges accurately at that stage, so misalignment is caught while it's still cents to fix, not after it's cured into a scrapped module.
Continuous per-module data lets the line spot a drifting stringer or layup robot immediately and correct it — preventing a batch of borderline modules instead of reacting after a cluster of scrap.
Logged cell-to-edge measurements per barcode provide SPC evidence for IEC 61730 creepage compliance and a defensible record against returns and delamination claims. Hard to price per unit, so the calculator offers it as optional upside.
ROI calculator
Sensible solar-plant defaults are pre-filled so it computes instantly. Scrap prevention is the headline lever; labor savings stack on top. Every field is editable — plug in your measured scrap rate, labor and cost basis.
Traceability
Cs-Veno ties a measured dimensional record to each barcode and keeps the visual evidence alongside it. Every export writes a timestamped row to the Database and captures a photo from each camera into a per-barcode folder. The standalone review app reconnects the trail — after PIN login, an operator types a barcode and sees its logged measurements next to the matching photo thumbnails.
Manual vs. Cs-Veno
FAQ
Cs-Veno is the first station built to measure cell-edge-to-glass-edge and cell-to-cell spacing right at the pre-lamination layup — and to do it accurately on any material and any color: any backsheet, glass, encapsulant or cell finish. That's the hard part everyone else trips on. Reading these edges reliably before lamination, across the full range of materials and colors a real line runs, is something no other system does accurately. That's the claim, and it holds up.
Cs-Veno measures to within 0.10–0.30 mm pre-lamination and 0.10 mm post-lamination. Those ranges hold across materials and colors; the exact figure on your line still tracks your camera resolution, optics, lighting and the per-point calibration your operator performs with a known length (e.g. a 25 mm ruler in the four-corner calibration box).
A Windows PC and commodity USB cameras — as many as you have spots to measure, each at 1920×1080 / 20 FPS via OpenCV/DirectShow — plus an automatic inline barcode reader (operators never key anything in). There's no fixed eight-camera cap; add a camera at every position you need checked. No frame grabber, smart camera or proprietary vision controller. The only optional accessory is up to two Govee BLE LED strips for pass/fail signalling — the AI glass-edge detection runs on the same PC, with no GPU or extra hardware required.
At export it upserts one row per barcode into the Database (with prior rows kept in a duplicate-history table) and captures a traceability photo per camera. Setup is one-time and uses a secured service key. Be aware: if the PC is offline at the export moment, the app shows a "DATABASE SAVE ERROR — CHECK NETWORK" message and that row is not queued or retried, so the station needs a live connection at export.
Cs-Veno uses an onboard AI vision model that learns the true glass edge and tells it apart from EVA, encapsulant and backsheet — so the measurement locks onto the real glass edge even when EVA is squeezed over or around it. The model is trained on your own modules, so it adapts to the exact materials, colors and lighting on your line and gets stronger the more it sees. There are no fixed rules to hand-tune.
The live app measures as many points as you set up — there's no fixed eight-point cap — each with per-point OK/NG limits (defaults 5.0–17.0 mm) and tallies. Note the default persisted Database schema logs four points (Time, Barcode, P1–P4); extend the schema when you log more positions. If the optional Govee strips are enabled, they flash green only when every enabled Cell-To-Glass point is strictly above a single pass threshold (default 12.5 mm), red otherwise.
Bring us your scrap rate, labor and cost basis, and we'll show you the payback on your line — then prove the measurement against your own modules.