Transformer and Coil Manufacturing KPIs: Test Yield, Winding OEE and Copper Scrap

Transformer and coil plants run on a handful of linked KPIs: first-pass test yield, winding OEE, copper scrap, rework, impregnation loading, thermal design margin and labor productivity. Use the ranges below as reference points for spotting a gap worth investigating, not as targets to copy, because your own trend says more than any single figure.

Every figure shows its basis. A superscript number links to a public source in the Sources list, and the sentence says what that source covers. A figure marked no cited source is kept exactly as originally published because no public source stating it has been found, so treat it as illustrative.

At a glance

First-pass yield at electrical test

First-pass yield at electrical test is the headline KPI. Typical shops run 90 to 95 percent no cited source, while world-class magnetics lines hold 98 to 99 percent no cited source.

Every point below target compounds, because failed units already carry full copper and labor. Treat any drop below 96 percent as a line-stop investigation, not a monthly review item no cited source.

How to measure

• Divide units passing Hi-Pot and turns-ratio tests on the first attempt by total units tested.
• Report it per shift and per product family.
• Trend it weekly from the test station's pass counts.

What moves it

• Winding tension control, since inconsistent tension causes most turns-ratio and short-circuit failures.
• Fast feedback from the test station to the winding station.
• Treating a falling trend as a line-stop event.

Size test throughput and track pass counts with the Hi-Pot Test Capacity calculator.

Winding machine OEE

Winding machine OEE separates good plants from busy ones. OEE is availability times performance times quality.

Typical coil winding sits at 45 to 60 percent OEE no cited source. No winding-specific data was found, but cross-industry measurements land in the same area: a study of Swedish manufacturers, with data from 2006 to 2012, found average OEE of 51.5 percent¹, and Vorne, an OEE monitoring vendor, says most manufacturers are closer to 60 percent⁴.

A disciplined cell reaches 75 to 85 percent no cited source. For a cross-industry reference, the world-class OEE figure from total productive maintenance is above 85 percent, built from availability above 90 percent, performance above 95 percent and quality above 99.9 percent².

Performance usually drags because effective speed is far below rated RPM once threading, taps, and tie-offs are counted. Availability improves most from bobbin and wire-spool changeover reduction, targeting under 8 minutes per changeover no cited source, inside the general SMED target of single-digit minutes, less than 10³.

For example, lifting effective speed from 40 percent to 65 percent of rated speed is often a bigger win than buying a faster machine.

How to measure

• Multiply availability, performance and quality for each winding machine.
• Compare effective winding speed with rated RPM, counting threading, taps and tie-offs.
• Time every bobbin and wire-spool changeover.

What moves it

• Bobbin and wire-spool changeover reduction.
• Cutting threading, tap and tie-off time.
• Raising effective speed before buying a faster machine.

Benchmark effective against rated speed with the Winding Machine Output calculator.

Copper scrap rate

Copper scrap rate is both a cost KPI and a process-health signal. Typical winding operations lose 5 to 8 percent no cited source, and the best hold 2 to 3 percent no cited source.

High scrap points to tension breaks, wrong-gauge issues, or excessive reject coils. A rising scrap trend predicts a yield problem before test data shows it.

For example, moving from 6 percent to 3 percent scrap on a line consuming 5 tonnes of copper a month returns roughly 150 kg of conductor a month, since 3 percent of 5,000 kg is 150 kg.

How to measure

• Divide scrap copper weight by copper issued to the floor.
• Weigh scrap by line and by week.
• Log the cause of each reject coil.

What moves it

• Tension control that prevents wire breaks.
• Gauge verification when wire is issued.
• Keeping bright clean wire segregated from enameled wire to protect reclaim price.

Estimate the copper each coil should consume with the Copper Wire Length calculator, so scrap has a baseline.

Rework rate

Rework rate and scrap rate are different numbers, and both deserve targets. Rework is units pulled for rewinding, re-taping, or re-impregnation, divided by units built.

World-class magnetics keeps rework under 2 percent no cited source, while 5 to 10 percent is common in shops with manual layer insulation no cited source. Rework is expensive because it doubles the labor content of an already finished unit.

For example, a 6 percent rework rate at 6 dollars of winding labor adds 0.36 dollars of hidden labor to every unit built, since 0.06 x 6 = 0.36.

How to measure

• Divide units pulled for rework by units built.
• Count every pull, even when the unit later passes.
• Split rework by type: rewind, re-tape or re-impregnate.

What moves it

• Mistake-proofing at the winding station.
• Layer counters on manual insulation steps.
• In-line turns verification before the coil leaves the machine.

Confirm the turns a design should carry with the Turns Count calculator before setting up in-line verification.

Impregnation batch utilization and cure yield

Impregnation throughput and cure yield govern the varnish line, often the plant bottleneck. Benchmark batch utilization, the units per cycle divided by chamber capacity, and target above 85 percent no cited source.

Void content after cure should sit under 3 percent for vacuum pressure impregnation (VPI) work no cited source, and cure reject rates below 1 percent are achievable no cited source.

A common failure is scheduling small mixed batches. For example, if varnish and cycle cost are fixed per batch, a 3 hour cycle run with the chamber 40 percent loaded carries 1 / 0.40 = 2.5 times the per-unit cost of a full chamber, or about 2.1 times the cost at the 85 percent target, and it starves downstream test.

How to measure

• Divide units per cycle by chamber capacity, every batch.
• Record void content and cure rejects per batch.
• Track the queue in front of the chamber and behind it at test.

What moves it

• Consolidating small mixed batches into full loads.
• Scheduling compatible products into the same cycle.
• Right-sizing loads to the chamber.

Right-size each load with the Impregnation Batch Size calculator.

Thermal design margin

Thermal design margin is a quality KPI you can benchmark before units ever ship. Track the fraction of designs whose measured temperature rise leaves at least 15 K of headroom below the insulation class limit no cited source.

World-class programs keep more than 95 percent of active designs above that margin no cited source. Thin-margin shops routinely release designs with under 10 K, then see field failures and warranty returns no cited source.

A second rule of thumb targets a rise-to-class ratio under 0.65, so a 130 C class part rises no more than about 85 K over ambient under full load no cited source. The two rules do not always agree, so add ambient and a hot-spot allowance to the rise and check the total against the headroom rule.

For example, at a 40 C ambient, an 85 K average rise, about the most the 0.65 ratio allows on a 130 C class part, puts the winding at 125 C. That is only 5 K under its class before any hot-spot allowance, so the design fails the 15 K headroom rule.

How to measure

• Measure temperature rise at full load on every released design.
• Add ambient and a hot-spot allowance before comparing with the class temperature.
• Report the share of active designs that meet the headroom rule.

What moves it

• Flagging marginal designs at the design review gate.
• Using measured rise, not predicted rise, for released designs.
• Core and conductor loss choices, which set the rise.

Estimate the core loss that drives temperature rise with the Core Loss Estimate calculator.

Labor productivity

Labor productivity closes the KPI set: good coils per labor hour, and its dollar twin, labor cost per good unit. A manual winding cell might produce 6 to 10 finished coils per operator-hour, while a semi-automated cell hits 20 to 40 no cited source.

Benchmark against your own product mix, not a headline number, because turns count drives the range. Normalize by turns and complexity, so a 700-turn part is not compared against a 30-turn part.

The strongest lever is batching similar products to hold setup under 10 percent of run time no cited source, which alone can lift throughput 20 to 30 percent no cited source.

How to measure

• Divide good coils by operator hours, per cell.
• Normalize output by turns count and complexity.
• Track setup time as a share of run time.

What moves it

• Batching similar products to cut setups.
• Semi-automation on high-volume parts.
• Labor cost per good unit reviewed next to output, not apart from it.

Model output per hour for your own mix with the Winding Machine Output calculator.

One weekly scorecard

Tie the KPIs into one weekly scorecard rather than chasing them singly, because they move together. Rising copper scrap precedes falling first-pass yield, low OEE performance usually means effective winding speed collapsed under changeover churn, and poor batch utilization starves test and drags availability downstream.

Example hard targets

• First-pass yield of 98 percent no cited source.
• OEE of 75 percent no cited source.
• Copper scrap of 3 percent no cited source.
• Rework of 2 percent no cited source.
• Batch utilization of 85 percent no cited source.

Review the trend, not the single week. Plants that hold four of those five at once typically sit in the top quartile for cost per good transformer no cited source.

How to build your own baseline

• Write down each KPI's definition and keep it fixed.
• Collect several months of shift and batch data before setting a target.
• Compare against your own trend first and published ranges second.
• Log what changed whenever a KPI moves, such as a new wire supplier or product family.