Kilandu sales team exchanging business cards with a European PCBA customer
PCBA Answers

What does a DFM review actually check?

Short answer. A useful DFM review checks the board against the process it will actually run on, and returns findings in six categories: panel utilisation, paste aperture against land pattern, thermal mass imbalance, missing process features, component lifecycle risk, and impedance or stack-up gaps. Each finding is listed with what it typically costs if it is not addressed, because a finding without a consequence is just an opinion.

The six categories, and what each one costs

  • Panel utilisation. A board outline that leaves unusable panel area, or spacing below what the router needs. The effect is a higher bare board cost on every order, not just the first.
  • Paste aperture against land pattern. An aperture ratio outside what the printer holds consistently, usually on fine pitch. The effect is solder bridging or an insufficient fillet, and it is found only at AOI, after the boards are built.
  • Thermal mass imbalance. Large copper planes or heavy components on the same side as fine-pitch parts. The effect is cold joints on the small features or overheating on the large ones - a reflow profile problem created by layout.
  • Missing process features. No fiducials, no tooling holes, no test points, or a panel break-off that leaves burrs. The effect is hand alignment, slower assembly and no access for test.
  • Component lifecycle risk. End-of-life, not-recommended-for-new-design or single-source parts in the BOM. The effect is a build that cannot be repeated, at a premium that grows each year.
  • Impedance and stack-up gaps. Controlled-impedance targets without a fabrication equivalent, or a stack-up with none stated. The effect is boards that meet the drawing and miss the electrical requirement.

Why the review sits at the assembly stage

The findings that matter most cross the boundary between board and assembly. A stack-up decision changes the reflow profile; a component substitution changes the paste volume; a panel break-off rule changes the tooling-hole requirement. A review performed only against the board will miss exactly the findings that are expensive to discover after the order is placed.

Who runs it and how long it takes

A standard review is returned with the quotation, normally within one to two working days of a complete file set, and it is carried out by process engineers rather than passed to the fabrication house or the customer's design team. Where the review is a joint exercise, a call resolves most findings faster than annotated drawings exchanged over several rounds.

What the review is measured against

A finding only means something if it is checked against the limits the line actually runs at. These are ours, and they are the figures a finding is raised against:

  • 1 to 24 layers, 0.2 to 3.2 mm board thickness, 0.5 to 6 oz copper.
  • Minimum drilled hole 0.2 mm, placement down to 01005, fine pitch down to 0.25 mm.
  • IPC-A-610 Class 2 or Class 3 acceptance, with the class stated on the quotation rather than assumed.

A design inside those limits is reviewed for yield and cost. A design outside them is not automatically refused: the finding is raised at review together with the process or design change that would bring it inside, and it is raised before any tooling is cut. That ordering is the whole value of the review, because a stencil and a fixture ordered against an unbuildable panel are costs that cannot be recovered.

Where the boundary is

The review produces findings; the design belongs to the customer. Kilandu manufactures to a customer design and does not take over the product design. Where a finding comes back, the decision to change the design is the customer's, and the review states the cost of each option rather than making the change.

When not to change a mature design

A design that has been in production has usually earned its layout. If a finding would mean a re-spin of the board to fix something that has not yet caused a defect, the cheaper answer is often to change the process - a different stencil aperture or a revised reflow profile - rather than the design. That is a judgement call, and it is one of the reasons the review is done by people who run the process.

The six findings, and where each one surfaces

CategoryTypical findingWhat it costs if not addressedWhere it surfaces
Panel utilisationAn outline that leaves unusable panel area, or spacing below what the router needsA higher bare board cost on every order, not just the firstAt fabrication, on every subsequent order
Paste apertureAn aperture ratio outside what the printer holds, usually at fine pitchBridging, or an insufficient filletAt AOI, after the boards are built
Thermal mass imbalanceLarge copper or heavy parts on the same side as fine-pitch componentsCold joints on the small features, or overheating on the large onesAt reflow, as a profile problem created by layout
Missing process featuresNo fiducials, no tooling holes, no test points, burrs at break-offHand alignment, slower assembly, no access for testOn the line, during the first build
Component lifecycleEnd-of-life, not-recommended-for-new-design or single-source linesA build that cannot be repeated, at a premium that grows each yearAt quotation, or later at material release
Impedance and stack-upA controlled-impedance target with no fabrication equivalent, or no stack-up statedBoards that meet the drawing and miss the electrical requirementAt electrical test, after the boards exist

Each finding is raised together with the process or design change that would bring it inside the limits, and raised before tooling is cut - which is the only point at which the change is still free.

What Kilandu does

DFM is run in Shenzhen against the process the boards will run on, and a standard review is returned with the quotation. The full checklist and the findings list are on the DFM page, and the file set needed to trigger a review is on the quote requirements page. Send a design for review to alden@kilandu.com.