This article is written by:
Jake Felser - Mechanical engineer and project manager at Cooper Perkins. Background in startups, manufacturing, and hardware design.
Chris McCaslin - Mechanical engineer and project manager at Cooper Perkins. Fifteen years of experience bringing consumer goods and new technology to market.
Developing hard goods involves many disparate and potentially overwhelming aspects, all of which need to be considered as you transition a prototype into a marketable product. The general term for designing a physical product to be made in volume is design for manufacturability (DfM) which covers all aspects of designing to reduce the difficulty (and therefore cost) of manufacturing something. An often-neglected subject within DfM is design for assembly (DfA) which covers how your product is actually put together during manufacturing.
In a perfect DfA world, key variables about the production supply chain would be understood early in the process and design decisions would be structured around that knowledge. Each part would be optimized to perform as many functions as possible, every part would only go in one way (poka-yoke), and the entire assembly would be put together intuitively, quickly, and with obvious success or failure during and after each assembly step. This would all happen at the target production volumes and with the right mix of humans, fixtures, and robotic assembly.
In this less than perfect world, many large, mature companies have trouble with this, let alone startups. Engineers at many companies now work in a timing-critical, minimum-viable-product focused world, with investors, customers, and backers itching to get finished goods in their hands. So how much DfA is appropriate? It does not need to be a big burden, but it cannot be something you decide on only at the beginning or address only at the end. DfA is an integral part of the design process from start to finish.
A useful DfA process includes thinking about how your product will be built, assessing potential failure modes, and then prototyping to ensure that the planned assembly works. In practice, the process is iterative, starting simply and increasing in resolution as the design advances. If a contract manufacturer (CM) has been identified, their early involvement in DfA can be instrumental in building the necessary tools for a successful product launch.
Essential Design for Assembly Tools
At Cooper Perkins, a technology development firm, we use a number of DFA tools. Here are three that can be easily applied:
1. Build a production flow (PFlow)
A PFlow is a flowchart that shows each step of the assembly. It is a working document, starting out as a conceptual guideline or sketch, but as the design moves forward, it should begin to include as much detail as possible (eg. how the assembly is done, where, by whom, how to verify each step is done correctly). Mapping the process concretely often quickly reveals inefficiencies, places where parts can be or need to be combined, required sub-assembly steps, and any impossibilities in the assembly process.
A PFlow is also a nice tool to use when identifying and selecting vendors for production, as it gives an idea of the capability, capacity, and repeatability that will be required of them. It also provides a structured way to identify assembly steps and provide feedback in both directions.
2. Iterative assembly failure modes and effects analysis (aFMEA)
Often built from the PFlow, the aFMEA is a list of the possible failure modes associated with each assembly step and their downstream effects. For instance, if a bolt is not sufficiently tightened during assembly, the possible effect is the in-use failure of the product. For each failure mode associated with a high-risk effect (for instance, product failure), mitigation should be identified to eliminate the source of the failure (for instance, test bolt tightness with a torque wrench, and add a thread locking compound). Common mitigations are adding fixtures to ensure correct assembly, testing and verification steps, and poka-yoke features that prevent incorrect assembly. What you get out of an aFMEA is very dependent on the resolution of the design, so it should be started early and improved as parts and features are refined.
An aFMEA can and should include other relevant considerations such as how assembly is verified and tested, whether or not it can be disassembled, and how the product is serviced. It can often be a useful tool to help evaluate obviously challenging parts/components early in the design process, and to figure out which aspects of the assembly can be eliminated, and what must be addressed through mitigations. At every substantial design change, walk through the assembly to see if the aFMEA is still applicable.
3. Prototype and Test
While building a PFlow and executing an aFMEA are great ways to mitigate risk in the assembly process, nothing takes the place of prototyping all the parts and putting them together. This can expose practical concerns that were not apparent in other analyses, such as the relative size of various components, accessibility for fingers and tools, and whether features are too big or too small to work effectively as verification or mitigation. Can you see the verification features, or the components themselves? Building a fully assembled 3D prototype can also shed light on assembly tolerances and give valuable intuition on the strength and feel of the finished product.
Another important and often overlooked part of assembly is purposefully trying to put the product together incorrectly. Can you insert a part backwards and get it stuck? Can you break off assembly features? If a part is assembled upside down, can you tell in the completed assembly? Visual clues to correct assembly can be very useful, but the inability to mis-assemble a set of parts can have positive downstream ramifications.
DfA Planning and Strategy
Rigorous DfA can be very effective, but it can also be time consuming; because it involves details of both the design and the supply chain, it often has to be re-done with every major design iteration or change in supply chain or volume. A critical decision in the product design lifecycle is whether the hardware is intended to be early, test hardware, or a platform for multiple generations of refinement and product evolution. In the startup world, the pressure is often to do both: get something into the world for immediate feedback while not detouring from a long-term product platform.
Whether the design is intended to be test hardware or a long-lived platform drives the answers to the basic manufacturing questions (how is it assembled, where, by whom, and how many). If the goal of the product launch is to first build a small number of units domestically and later move production to Asia in much higher volume, the DfA process should focus on the builds in Asia from the beginning to prevent a major re-work. The extra cost incurred in the build of the small batch by designing the product for Asia will most likely be offset by the minimal re-design time once production is ready to move overseas.
Whether or not your product is designed well for assembly can be critical to its quality, price and design longevity, yet DfA is often a neglected step of the design process. Designing verifiable and unmistakable assembly steps can improve the yield of the product and improve the quality of the finished product. Eliminating scrap by improving yield of each step lowers the cost, and making each step easier can cut down labor expenses significantly. Often a large portion of the cost of goods sold is labor, even in Asia.
It is important for startups to internalize a DfA strategy early on in development to prevent a re-design of the whole product in the future. Using basic tools such as a PFlow, aFMEA, and strategic prototyping of the assembly early and iteratively can go a long way towards the development of successful designs that live on as platforms for further development in years to come.