Cost reduction on a charger fails in a specific and predictable way. The savings are real and booked, and the warranty cost that arrives eighteen months later is attributed to something else entirely.
Avoiding that requires classifying parts before touching them, and scoring each candidate against its qualification cost rather than its unit saving alone.
Classify before you optimise
Attack in reverse order of risk. That is the opposite of what spend analysis suggests, because the highest-value components are usually the ones carrying the most certification exposure.
- Safety critical: protection devices, insulation systems, earthing hardware.
- Functional critical: metering, contactors, control components.
- Firmware coupled: meter interface, modem, flash, MCU peripherals, sensors.
- Commodity and mechanical: fasteners, brackets, generic passives.
- Cosmetic and packaging: labels, cartons, inserts.
Savings that require no design change
Commercial renegotiation at volume, consolidating part numbers across variants to increase order quantity, packaging redesign that pays back on freight, and process improvement that lifts yield.
Variant reduction is the most commonly overlooked. Every variant multiplies test, documentation and inventory, and many exist because a decision made for one customer was never revisited.
Firmware coupling is the trap
A meter, modem, flash device or MCU peripheral can be electrically equivalent and behaviourally different. Register layouts differ, timing differs, AT command handling differs, and the resulting faults appear in the field rather than on the bench.
This class of part looks like a straightforward substitution and behaves like a firmware project. Treat any part the firmware talks to as requiring integration validation, more than electrical comparison.
Score before substituting
Each candidate carries an annual saving against a qualification cost, a test cost, a certification risk and a supply risk. A saving requiring environmental requalification and a certification delta review may cost more than it returns at your volume.
Doing this arithmetic before the work starts is the entire value of running cost reduction deliberately rather than opportunistically.
When architecture is the cost
Past a point, part substitution stops yielding. Reducing component count, simplifying the power path or consolidating boards produces step changes that no amount of sourcing can match.
That is a redesign with its own certification consequence and its own programme. It should be entered deliberately rather than arrived at by accumulating small changes until the product is no longer what was certified.
The changes that cost more than they save
- Anything in the safety path, where the qualification and certification burden is highest.
- Metering components, where accuracy class governs billing validity.
- Contactors, where the failure mode is a live circuit and endurance testing is slow.
- Any part whose behaviour firmware depends on, without full integration validation.