Engineering: Can We Design Around It?

On Wednesday morning of the first week, Dev Anand has two messages open on the same screen. The first, from Ravi Menon on Tuesday afternoon, is one line long: can we design around it? The second, from Hannah Wolcott at ten past eight, closes the door Ravi was hoping to hold open — she has run every source of CB-40 boards to ground, and nothing she has found arrives inside the twelve-week window.

So the question has narrowed overnight. Nobody is asking whether the CB-40 can be replaced; of course it can. They are asking whether it can be replaced by the middle of May. Saying no to that, with dates and reasons rather than a shrug, is much of what an engineering lead does in a shortage.

What lands on Dev’s desk

Ravi’s question is a good one and Dev treats it as one. “Can’t we just use a different chip?” is what any sensible person asks when one component stops a factory, and it earns a careful answer rather than a weary one. Nothing about it is naive; only the physics, the software and the law are inconvenient.

Hannah has closed the alternatives: Kestrel Micro needs ten weeks, brokers want $1,100 a board for parts of unverifiable provenance, of which she has taken one small traced batch and no more, and $85,000 of air freight claws back two weeks at the far end. She also sends what Vantor and Kestrel have told her about which microcontrollers can actually be bought this year. Ravi’s standing instruction binds everyone downstream: protect first the relationships that cannot be bought back at any price, revenue second, money only where it buys something lasting.

The rest is fixed: 520 boards, 80 ovens a week, 6.5 weeks of production then 5.5 weeks dark, quarter end in week nine. Dev controls the design, whatever can be recovered from the building, and the accuracy of his dates.

What an engineering lead actually does

An engineering lead owns the design of the product and the technical file that proves it is safe and legal to sell. Normally that means roadmaps, revisions and cost-downs; in a shortage it means being the company’s reality check, converting “surely we could just…” into weeks, dollars and consequences. He does not decide what Calder does; he is the person without whom nobody else can.

The software on Dev’s desk

The design tools answer whether a different microcontroller physically fits — pinouts, power, heat, clearances. The product lifecycle system answers something more valuable in a crisis: where else is this part used? That single query is what turns “a board is late” into “every product line stops”, and at Calder it is also what proves the CB-40 has been a single point of failure since 2020.

The last two screens are the ones non-engineers underestimate. Test records and the certification file tie a specific design to specific evidence that it is safe. Change the control board and that evidence no longer describes the product being sold, which is why fourteen weeks is not pessimism — most of it is retesting and recertifying rather than designing.

The decisions

Why “just use a different chip” takes fourteen weeks

Follow the chain and the fourteen weeks stop looking like padding.

A different microcontroller has different pinouts — which physical leg of the chip does what — different power characteristics and different timing, so it cannot sit in the CB-40’s footprint. The board is redesigned around it, about two weeks of schematic and layout, and prototype boards must then be fabricated and populated: three weeks in somebody else’s queue, no faster for being urgent.

The firmware — the software living on the board itself — is ported in parallel: cooking programmes, probe calibration, burner control and fault handling, all written against one chip’s peripherals, all rewritten against another’s and retested, because a board that mistimes a gas valve is not a cosmetic defect. Call that three weeks, plus two to build pilot ovens and validate them as ovens rather than on a bench.

Then the part nobody outside engineering expects. Those ovens go for electrical safety, electromagnetic compatibility and gas-appliance testing: four weeks in an external laboratory’s diary, booked in advance. And because Calder’s UL listing references a technical file describing the certified design, a changed control board changes the file, and the product must be recertified before it can be sold.

That is where impatience lands, so Dev is careful about it. Certification is not bureaucracy. It is why a commercial oven may legally be sold, why a hotel’s insurer will cover a kitchen containing one, and why Harlow’s facilities team is allowed to sign for it. An uncertified oven is not a slightly riskier oven; it is an unsellable one.

Starting the redesign anyway

Fourteen weeks is longer than the twelve-week disruption, so the redesign cannot save this crisis — and Dev says so in the first line of every conversation about it, because an engineer who lets people hope otherwise damages the plan.

He starts it anyway, and how he frames it to the board is the decision. Not crisis response: the removal of a single point of failure that has sat inside every product Calder sells for six years — one board, one supplier, one chip, in every unit across three lines. The shortage did not create that exposure, it revealed it. Sold as a response to Vantor’s email the programme dies the moment deliveries resume; sold as the closing of a structural hole it survives the return of normality.

Its engineering content is the principle of designing for supply. The new board is built around a microcontroller at least two manufacturers make in interchangeable form, so a future allocation cut is an inconvenience rather than a stoppage, and single-maker parts are avoided where an alternative exists and flagged in the bill of materials where none does. It would have cost a few dollars a board six years ago and saved roughly $1.7m of gross profit this quarter.

He asks for it to be funded, about $250,000, rather than staffed from goodwill: a programme run by borrowed people is the first thing dropped when the borrowing stops.

The smaller wins actually available this month

This is where an engineering lead earns his place, because none of it is glamorous and some of it works.

The first move is a hunt through the building. Control boards accumulate where no inventory report looks: service stock for warranty repairs, boards inside ovens returned from site awaiting rework, engineering samples on benches, units built for orders later cancelled. In a company with more slack that sweep is often worth two or three weeks of parts, which makes an afternoon of looking one of the best-value activities in the crisis.

At Calder it is worth less than a day, and Dev says so rather than dressing it up. The sweep finds 47 boards. Thirty are service stock and he ringfences them: raiding the pool buys two days of production at the price of an installed oven failing in a customer’s kitchen with no spare, which is exactly the damage Ravi is trying to avoid. Nine are older-revision samples needing firmware validation; five sit inside customer ovens awaiting repair. What is genuinely recovered is three finished Compacts built for an order cancelled in January — three ovens that ship this week without touching the 520.

The second move sounds cleverer and gets rejected. Calder has an older, simpler control board, still certified on a line discontinued in 2021 and available today. Could the Compact ship with basic controls now and the advanced baking programmes follow later? Dev says no twice over. Technically, a different board in a current Compact is a changed certified design, dragging most of the same certification tail behind it: about ten weeks, still outside the window. Commercially it is worse. The advanced functions need hardware the old board lacks, so “enabled later by a firmware update” is not true — it is a promise to send an engineer back to every customer’s site, and it leaves two variants in the field for Siân’s team to service. A temporary variant is never temporary — it ships, it works well enough, the return visits slide, and it is still out there in 2031.

What engineering owes the other desks

The most valuable thing Dev produces this week is not a fix but a set of answers firm enough to build plans on.

Exactly what cannot change: one CB-40 per oven, no substitution without recertification, no exception for a rush order. Exactly how long each option takes: fourteen weeks for the redesign, about ten for a reduced-function variant, never for a drop-in. And exactly what is interchangeable — the CB-40 is identical across Meridian, Compact and Rack, so any board can become any oven. That sounds trivial and is not: it is what lets Tomasz compare margin per board across the three lines in module five.

An honest “no, and here is why” is worth more in a crisis than an optimistic “maybe”, because a maybe is not free — it keeps a lane open in everyone else’s plan. Tomasz holds boards back against a rescue that never comes; Marcus sells dates on the strength of it.

Where this goes wrong

What happens in real companies is not an engineering failure but an attention failure. The board approves the respin in week two, while the shock is vivid. Around week thirteen the deliveries resume, the factory catches up, and the redesign engineers are the obvious people to pull onto the launch that slipped while they were busy. Nobody cancels the programme; it stops having anyone on it. A year later the same sole-sourced chip sits in every unit, protected by a memory of how unpleasant March was.

The defence is structural rather than emotional: a named owner, a funded budget line nobody can quietly repurpose, a delivery date reported alongside the sales numbers, and a risk-register item Ingrid will not let close. Willpower fades on a predictable schedule; governance is what you use instead.

What Dev hands on

To Tomasz goes the answer that lets allocation begin: no engineering rescue arrives inside the window at any price, and the plan is built on 520 boards and nothing more. With it go the three finished Compacts, the service pool marked as unavailable for production, and the confirmation that boards are fungible across the three lines.

To Ingrid goes the finding that outlives the shortage: one component, one supplier, one chip maker, in every unit of every line, unexamined for six years — the exposure her register should have carried.

To the board goes the funded redesign proposal: fourteen weeks, roughly $250,000, no help with the current crisis, and the permanent removal of the condition that made it possible.

Spot the decision

Read each situation and decide how an engineering lead should handle it, then tap a card to check.