Stress-screening every unit isn't quality control. It's a design margin you never had.

I've stood in enough plants to recognise the sound of a problem being dressed up as progress. It usually arrives as a new test cell, an extra inspection station, a 100% screen that everybody nods at and agrees is "just good practice." Controlar's announcement of its automated environmental stress screening system for high-voltage EV DC-DC converters lands squarely in that territory. Thermal cycling, vibration profiling, electrical stress – applied to every unit off the line. Read past the engineering language and you can hear what it actually says: our design margins and process capability are not sufficient to ship without proving each unit survives.

Controlar's stress screening is a process capability confession

ESS has real pedigree. Military and aerospace programmes developed it to precipitate latent defects – weak solder joints, die-attach voids, capacitor cracking – before they reached the field. The key word is latent. ESS is not a sorting tool for defects your process routinely produces. It is a safety net for the defects your process might produce at the tail of the distribution. When you apply it to 100% of production, you are making one of two statements, and probably both: your design margin is too thin to absorb normal process variation, or your process capability is too low to be trusted. Neither statement is something you would put in a press release. But the engineering logic is inescapable. If your Cpk on the critical characteristics of that converter were sitting at 2.0, you would not need to thermally cycle every unit. You would screen a statistical sample, monitor the data, and release on confidence. If your PFMEA had identified the real failure modes and you had engineered them out at the design stage – derating components, selecting fault-tolerant topologies, specifying materials with verified margin – the residual risk would not demand 100% screening to manage.

What 100% screening buys you – and what it doesn't

ESS catches the unit that would have failed in the field. That is real value. A field failure on a high-voltage DC-DC converter is not a warranty line item; it is a potential safety event, a recall exposure, a brand incident. The screening is not irrational. It is rational given the design and process constraints. But those constraints are the actual problem, and the screening obscures them. What 100% screening does not do is reduce the defect rate. It shifts the defect from the customer to the factory floor. That shift has worth – you would rather catch a failing unit in-house than on the road – but you pay for it permanently. Capital equipment depreciation, cycle-time penalty, floor space, energy consumption, calibration cycles, maintenance contracts, and the hidden cost of false failures: units that meet design intent but fail the screen because the screen itself introduces stress artefacts. At EV volumes, this becomes a structural per-unit cost on products that are already margin-constrained. The design margin you skipped in the programme plan is now a recurring line in your cost of goods sold. At SNOP, the 70% defect-cost reduction we achieved came from raising process capability until screening could be reduced, not automated. We tightened incoming material controls, stabilised critical process parameters, and used QRQC and A3 methodology to drive root-cause resolution on the top defect modes. The screening we removed was screening we no longer needed. That is the direction quality engineering is supposed to travel – toward eliminating the need to screen, not toward more sophisticated ways of doing it.

Aerospace engineered the margin first. EV is still paying tuition.

The aerospace industry went through this learning curve decades ago. Under AS9100 and the EASA regulatory framework, screening intensity is driven by criticality analysis, not applied as blanket coverage. Flight-critical parts receive rigorous qualification and ongoing surveillance. Lower-criticality parts receive proportionately less. The system functions because the engineering work was done upstream – in the design margin, in the process qualification, in the traceability architecture that lets you prove what you claim. In my current role at Airbus, we reduced EASA audit findings by 50% in a single cycle. We did not achieve that by adding inspection layers. We achieved it by strengthening the process itself – ensuring that design intent, manufacturing instruction, and verification method were aligned and auditable end to end. Screening exists where criticality demands it. It does not exist as a compensation mechanism for engineering shortcuts. The EV industry operates under IATF 16949, which provides the same toolkit – APQP, PFMEA, control plans, PPAP. But the culture has tilted toward detection over prevention. The playbook is: ship fast, screen everything, sort the fallout. That has a shelf life. It works while capital is cheap and while every competitor makes the same trade-off. The moment a competitor engineers the margin properly and ships without the per-unit screen, your cost disadvantage becomes permanent. We are already watching EV programmes stall and plant tenants rotate in Macomb County. Thin margins punished by thin engineering.

If your PFMEA says "100% screen" as the control, the failure mode isn't controlled – it's just expensive

This is the test I apply when I review a control plan. I look at the control method column for each high-severity or high-occurrence failure mode. When I see "100% inspection" or "100% screen," I do not see quality. I see an unresolved engineering problem that has been operationally accommodated. A controlled failure mode has been designed out, processed out, or verified through capable statistical methods. A screened failure mode is still happening. You are just catching it before the customer does.

Every unit you stress-screen is a unit your engineering could not guarantee.

Controlar built a capable system. The engineering behind the automation is genuinely impressive. But the most honest reading of this announcement is not that the screening industry is advancing – it is that the EV converter supply chain cannot yet ship without it. The companies that will dominate this market are not the ones with the most sophisticated screening lines. They are the ones who engineer products and processes that make those lines unnecessary. Every euro spent on 100% screening capacity is a euro that should have gone into design margin. The longer the industry treats the symptom as the solution, the more expensive the correction becomes.