Inter-turn short in a motor — once it goes into the impregnation oven, it is scrap.

The problem

The defect is visible before the oven — afterwards, it is not.

An inter-turn short is born in small pieces of damage to the enamel on the magnet wire during winding or slot insertion. The classic system has three leak points:

1. The operator's eye — a pinched wire, a badly formed end turn, a speck of metal between turns: these are things the veteran spots. The day they retire, that sensitivity walks out with them.
2. The surge test read by hand — the operator compares three waveforms on a screen. If the asymmetry is obvious, they catch it; if it is mild (microseconds of deviation), it depends on their eye and on how many hours they have been at it.
3. The timing of the defect — before impregnation the stator can be rewound; after the varnish, it is scrap. The window is a matter of minutes.

The marginal stator gets through, is impregnated, ships to the customer and fails in the field months later. The classic system works 99% of the time. That 1% is an insulation claim, and on critical motors (industrial ventilation, process pumps, electrical machines for rail) the field cost is several times that of the stator.

How it fits the IRIS system

iLEAN does not add a fourth system — it seals the cracks between the winding machine and the impregnation oven.

The problem is not a lack of information: the surge test and the operator's eye generate plenty of data. The problem is that nobody can hold that level of attention on every single stator through a whole shift, nor capture the veteran's sensitivity before they retire. iLEAN acts as the putty that fills the gap between what the bench measures and what quality needs.

The camera watches the winding. iLEAN Vision assists the surge test against the batch pattern. The doubtful stator is pulled aside before the oven. The person signs — never the other way round.

The iLEAN pieces applied to inter-turn short detection:

• iLEAN Vision (Edge with a CNN trained on windings) — external visual inspection of the finished winding: pinched wire, badly formed end turn, visible metallic debris, enamel damage. If the image does not match the batch pattern, it flags the stator before the surge test. It works with no network.
• Vision + assisted surge — a CNN over the surge tester's screen in real time. It compares the waveforms of the three phases against the pattern learned for the batch, not against the generic standard. Mild asymmetries the eye would miss are flagged for a retest or for human inspection. The person decides — the system only prepares the decision.
• iLEAN Tracer — links the doubtful stator to the magnet wire batch, the winding operator, the end-turn forming tool, the impregnation oven and the serial number of the finished motor. When a field failure appears, the trace back to origin comes out in seconds — and it is learned for the next batch.

See the full IRIS architecture →

Before and after

Surge test read by hand vs. assisted surge with iLEAN Vision

Aspect	Classic inspection before impregnation	With iLEAN Vision
Visual inspection of the winding	The operator's eye, the veteran's sensitivity	CNN trained on the batch pattern
Reading the surge test	Operator compares three waveforms on screen	CNN over the screen, sub-threshold asymmetries flagged
Mild inter-turn defect	Frequently gets through, fails in the field	Stator flagged for a retest before the oven
Capturing the veteran's knowledge	Walks out the day they retire	The learned pattern preserves the sensitivity
Tracing the defect back to the batch	A handwritten note from the winding lead	Linked to the wire batch, the tool and the operator
Recovering the stator	Only if the operator catches it in time	A higher share of stators recovered before the oven

Impact estimate

Impact estimate for your plant — to be validated with your numbers.

The block below is an estimate to be validated with the specific data of your plant. We put it forward so the committee has an order of magnitude; we refine it during the diagnostic.

• Mid-sized manufacturer of asynchronous or synchronous motors, with a winding line + stator insertion + surge test before the impregnation oven.
• iLEAN Vision pilot with a fixed camera over the finished winding + a CNN over the surge tester's screen + integration with the MES. First value expected within a few weeks: the first recovered stator that a hand-read surge test would have let through.
• Indicative payback between 4 and 9 months, depending on the average cost of a defective impregnated stator and the frequency of insulation claims in the field.
• Expected reduction in customer insulation failures of ≥ 30%, a defensible floor. The hard lever is the stator recovered before the oven (wire + varnish + hours) and the field claim avoided.

And the winding lead's reasonable doubt

“What if the AI flags a good stator as doubtful?” — hallucination is a problem of free generation, not of anchored tasks. In tasks where the AI compares a real image or a real waveform against a pattern learned for the batch, the best models brought error below 1.5% ^[1]. And even so, what is critical is not decided alone: iLEAN flags the doubtful stator and the person decides whether it is retested or rewound. The three safety rings exist precisely for this.

^[1] OpenAI paper “Why Language Models Hallucinate”, 2025 — on the reliability of AI in anchored tasks.