Manufacturability is the Product Designer’s Responsibility

If you design a product without consideration for manufacturability, you are setting yourself up for problems you’ll have to fix later. Without consideration for Design for Manufacturing (DFM) starting with the first prototype, a long list of problems will show up at New Product Introduction (NPI). This includes everything from test fixtures that don’t exist when needed to lack of support firmware for production test. In the bad old days it wasn’t uncommon for products to get to the factory floor only to find that required manufacturing steps couldn’t be completed.

At F3, we treat manufacturability as a core aspect of the design process. Manufacturability must be addressed while drawing the schematic and layout for the initial prototype, not after the design is handed off.

Optimize for programming and branding

At F3, the first prototype is intended to be a production-ready design. This includes written test plans for Design Verification Test (DVT), certification, and manufacturing. A key aspect of DFM/DFT is ease of programming any parts that need it as well as “branding” into the device any per-unit specific information like security keys or calibration factors. This is also typically where unit-specific labels get printed and attached to the product.

For each programmable device, you need to determine:

• In-circuit programming versus pre-programmed parts. Programming on the line needs accessible headers or pads and a defined electrical path. At F3, we don’t recommend pre-programming unless it’s absolutely necessary.
• Access. Select dedicated programming connectors or test pads and determine whether they are shared across multiple devices or need isolation.
• Bootloaders and sequence. Which part must be live first, what programs what, and whether one processor provisions its neighbors.
• Time per unit. Programming and functional test time costs money, and line time correlates to unit cost. A design that takes two minutes to program instead of ten seconds carries that penalty to every unit built.

Generic firmware is the same in every unit. Branding is not. Branding, in F3’s terms, is the act of writing unique, per-unit information into each individual device at production time. That includes serial numbers, unique IDs, MAC addresses, cryptographic keys and certificates, calibration data, and regulatory identifiers. Two units off the same reel run the same firmware but must not share the same identity.

At F3, we develop end-of-line production testers for products we develop as well as for customer designed products. These testers complete all the programming and branding for each unit.

Wires are good

You need a test interface to control the behavior of the radios and other system elements that require testing. No matter what kind of test, design verification, certification, or manufacturing, you can’t have the test interface be a radio. Instead use UART, USB, or Ethernet because it needs to be a closed system without any RF interference. All radios have interference. At F3, using a wired test interface is a requirement.

Relying on the product’s own wireless link for testing is also a trap. The wireless interface is frequently one of the things being tested. It may not be functional until after branding and calibration, and it introduces variability that a production line cannot tolerate. Design the wired test port in from the start and include the firmware commands the tester will need to read sensors, drive outputs, report identity, and return measured results.

Test points are important. Test coverage is not.

Test points exist for two reasons: you are using them for design verification test (DVT) and/or production functional test. Too many unnecessary test points can be a liability. They can potentially bloat the board area; add placement and fixture cost; and create stubs that act as unintended antennas, degrade signal integrity on fast nets, and become reliability and EMI problems. As we say in our end-of-line production test blog, a production tester should verify function, not diagnose every possible failure, and the test-point strategy should follow that same discipline.

As with any modern product development, AI should also be included in the process for both quality and efficiency reasons. AI has the ability to review every data sheet, and errata for every part in the schematic and give valuable feedback. This does not, however, replace actual human reviews at each review gate.

Don’t stuff parts “Just in case”

Parts added out of worry make the product more expensive and fragile, not safer. A common example is dropping series zero-ohm jumpers onto a UART simply because someone is worried about swapping TX and RX. Get the pinout right in design review and the jumpers will be dead weight on every unit forever. The goal is to get the design right, not to hedge every decision with a part that someone might need someday.

Electronic design best practices

F3 has curated a checklist of more than 200 specific RF design best practices for reviews. It is reflective of our 30 years of electrical engineering experience. DFM and DFT is included in every one of these reviews.

We conduct several reviews throughout the process, including overall architectural reviews before schematics are created, reviews of schematic capture symbols and footprints, reviews at the end of schematic capture, placement reviews, mechanical reviews of the PCB, and PCB placement reviews. Best practices should be applied to all these reviews, which is standard operation for F3.

What’s the difference between manufacturing readiness and a prototype?

The simple answer: Documentation.

If the objective is to have the contract manufacturer complete assembly, you can empower them with the information and tools they need to do so. Full documentation is key, and it typically includes assembly instructions, prints for every custom component, bill of materials (BOM), end of line tester instructions, user manual, assembly process and expectations. Prototypes have almost none of this. They have a skilled engineer who understands the entire product and they’re the only ones who can test and verify the proto does what it’s supposed to.

So, what is time to volume?

F3’s main sales channel is referrals from customers and our many partners. Nobody makes any money, however, until the actual product ships in volume. That means passing certifications, DFM, supply chain, and ensuring the device’s performance meets customer expectations. We are creating custom solutions using established, known technologies for reliable products and cost, providing exceptional value. In volume. Our customers and partners know this from working with us and that’s why F3 is an industry leader in custom electronic products.