When I bring up quality planning to new hardware startups, I usually get a combined look of confusion and fear. Facing creating a 30 page document full of test procedures is overwhelming. They are surprised by the amount of thought and work that needs to be done to create a quality plan that will ensure their product will work robustly in the field.
Startups make the false assumption that quality is just the final check at the end of the product line to make sure it turns on. However this last inspection is only the final check on what has been a long path of testing, verification, and validation from the earliest prototypes to ongoing production testing to ensure that production and suppliers haven’t drifted in their quality. The earlier and more robustly you can test your product, the less likely you are to have the customer find the failures. In addition, you want to have a strong hand in how your product is tested so it catches problems before they escape.
Let’s first talk about what we mean by “quality.” This is a term that is thrown about and has attached to a lot of aspects of the design. Quality can include the softer more elusive feeling of a product (i.e., why does the Apple iPhone look like high quality) to how well the product performs against industry standards (i.e., think consumer reports on the noise of your dishwasher) to the reliability of the product (i.e., think warranty and repairs).
Quality methods fall into one of two categories: validation and verification.
Validation makes sure that you are designing to the right criteria. Typically the criteria are defined in specification documents or other promises to the customer. These tests are usually more open ended user tests. Verification is a much more rigorous numbers based testing that clearly states what the goals of the product should be and whether or not it passes or fails. The quality plan is the document that outlines the second.
Prototype verification and validation testing. Engineering analysis, CAD models, and very early prototypes can be used to test the design; however some performance characteristics will only become apparent when you have pilot prototypes in hand. At this point in the project, time is very tight. Planning what you need to test long before the pilot run, will help you identify weaknesses in the product quickly and make the design changes in time for the next pilot run.
Regulatory and safety requirement. You will need to subject a certain number of units to certification and safety testing to satisfy the local laws and regulations (see article on certification).
Molded part quality. Many of your parts will be made from molded plastic or metals. Ensuring that the parts are of the right dimensions before the build ensure that any problems can be traced to the design rather than the production quality.
Durability and reliability. In the normal life of your product it will be dropped, pulled, sat on, left in a hot car, kicked, sprayed with water, and get dusty. You will need to test your product to ensure that your design is robust to these stresses and can survive without impacting performance and safety.
Transportation testing. Your product will be subjected to wide temperature swings, impacts, drops and vibration before it even gets to your customer’s door. There are standard tests that ensure that your product and its packaging can withstand the stresses of being shipped and transported.
Life testing. Many failures occur only after a large number of cycles. Fatigue, wear, UV damage, thermal stresses can have cumulative impacts on a design. In addition, some failures are random and just need enough cycles to ensure that the product has an opportunity to display the failure. Life testing runs the product through a large number of cycles on critical aspects of the design. Because the failures you are looking for are rare and random, larger number of samples and cycles are needed.
Incoming inspection. When you buy components or sub-assemblies from suppliers, you can’t assume that their quality system is stable and capable. Each batch (and in some cases each part) will be subjected to incoming quality checks to ensure that the supplier is delivering a consistent quality that meets the specifications you have set. There may be special features or aspects of the purchased parts that you may want to do more frequent sampling if that feature is very critical to the performance of your system or if it is a safety critical item.
PCB testing. Each of your PCBs will go through multiple tests including 100% optical inspection, Integrated circuit testing and functional testing. You want to be able to find failures before they get built into your product and be able to diagnose a failure before you have to send it for rework (which often does damage itself).
Functional testing. Your completed unit will have a series of tests that ensure all of the basic functionality work before it gets shipped. This can include camera tests, on/off speakers etc. These tests are different from the verification testing. The verification testing will be much more extensive and often not possible to do 100% on the production floor. Rather functional tests ensure that the design as built works as expected.
Aesthetic and packaging. You will have clear criteria about what is and isn’t acceptable to ship. These can be as detailed as the size of flash and as seemingly obvious as no hair or sharp objects in the packaging. It is fun to read older requirements documents and try to figure out what weird item ended up in a box that drove the requirement.
Shipment audits. Functional testing and inspection are not reliable nor will they catch 100% of defects. Operators get tired, new operators are trained and don’t know the process well or shortcuts are made. Shipment audits take a small number of product before shipment out from the factory and re-run the whole suite of functional and aesthetic checks. Any failures will require the CM to rework the batch to ensure conformance. It is critical to do this because it is very difficult and expensive to get product fixed once it leave the warehouse.
OPT. Things change and degrade over time. Suppliers will make small changes to processes, second source suppliers deep in the supply chain will be changed out, SOPs change or aren’t followed. Running comprehensive reliability tests throughout the life of the product gives you a higher confidence that these can be caught before a large number of products are out in the field.