The veteran wire EDM that finally talks

The problem

The machine that cuts the horn slots forgets every part it cuts.

The workshop that machines horns often relies on a wire EDM that is fifteen or twenty years old: precise, critical for the horn slots, and fitted with a proprietary panel that has no connection. Cut time, wire breaks, parameters per part: it all appears on the panel and vanishes. Nobody can correlate that data with the horn's final frequency or with rework.

• The horn workshop often relies on a wire EDM fifteen or twenty years old: precise, critical for slotted and block horns, and fitted with a proprietary panel with no network port.
• Cut time, wire breaks, wire diameter, tension, current and offsets per part appear on the screen and vanish when the next program loads. Nobody writes them down because nobody has the time, and the operator is already setting up the next titanium blank.
• When a horn comes off the tuning bench with its resonant frequency off target, nobody can check whether that part had three wire breaks or a different offset, so the rework is blamed on design or on luck.
• Replacing the machine to get data is a capital request nobody wants to sign for a machine that still cuts within tolerance.

How it fits the IRIS system

Connect in photo mode on the panel — read from outside, the control untouched.

ILEAN Connect places a camera pointing at the screen. OCR with an anchored model reads every change and structures it per part and order. The machine stays as it is; its control is never touched. iLEAN links what it reads with the horn's work order in the ERP, so quality and engineering see the full history of every part.

The camera reads what the operator already sees, and nothing more. No retrofit, no network card, no change to the EDM's program: the machine keeps cutting exactly as it does today and its history stops disappearing. For the first time, each horn arrives at the tuning bench with its cutting record attached.

See the full IRIS architecture →

Before and after

The silent EDM versus the EDM being read

Aspect	Today	With iLEAN Connect
Parameters per horn	On the EDM screen, then gone	Stored per part and work order
Wire breaks during a slot cut	Remembered by the operator, if at all	Counted per part
Off-frequency horn at tuning	Cause never traced	Cross-checked with its cutting history
EDM control and program	—	Not touched
Engineering and quality access	Walk to the machine	Full history per horn
Replacing the EDM for data	On next year's capex request	Deferrable

Before: panel data lost. After: machining history per horn.

Impact estimate

Impact estimate — to be validated with your numbers.

The block below is an estimate to be validated against your plant's actual data. We put it forward so the committee has an order of magnitude; we refine it during the assessment.

• Estimated payback 5-10 months.
• Replacement capex for the EDM can be deferred, because the reason for replacing it was data, not precision.
• The link between cutting parameters and horn rework becomes visible for the first time, so the next design or the next offset change is decided with evidence rather than with the most senior operator's intuition.
• Engineering can compare the cutting history of good and reworked horns of the same design instead of arguing from memory.

Payback 5-10 months · replacement CAPEX deferred (estimate to be validated). Replacing the machine can be deferred, and the link between parameters and rework becomes visible. Estimated payback of five to ten months; an estimate to be validated.

And the fair question from the production manager

“A camera reading an old screen will make mistakes.” — the screen is fixed, the fields sit in known positions and every value has an expected physical range, which makes it an anchored task where the best models drop below 1.5% error [1]. A frame whose wire tension or current falls outside the possible range is discarded rather than stored, and the first week is checked against manual readings before anyone relies on it.

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