[We asked an AI what it thought the pitfalls of electronic product development were. This is the fifth in a series of articles provoked by the AI’s response.]
We finally reach the stage of volume manufacturing in our journey through creating a new product. What “volume” means is obviously a major consideration here; manufacturing 30 units is a very different proposition to manufacturing 300,000 units. We will mostly be talking about medium to high volume manufacturing in what follows. Low volume manufacturing may have some of the same issues, but they often get raised and dealt with earlier in the design process.
The first potential problem the AI threw at us was that of underestimating supply chain risks. Factory fires, geopolitical tensions and global pandemics are just some of the things that have unexpectedly disrupted the supply of components for manufacturing. Access to raw materials and again the associated geopolitics is an issue in (for example) electric vehicle manufacture. Likewise in the wake of the COVID pandemic we did quite a lot of work quickly redesigning existing products to use the chips that were available at the time. The biggest issue here is that many of these risks are entirely unpredictable.
There are two fundamental strategies for dealing with supply chain fluctuations. You can either try to ensure that you don’t run out of components, or you can develop with alternatives in mind.
It’s hard to be certain you won’t run out of components in big production runs. For small or medium runs we recommend reserving or buying critical components very early in the process to ensure you will have them at manufacturing time, but that can be an expensive proposition for large runs. Having multiple suppliers is a good idea, but in times of global shortage the problem is that no one has the components you need. Still, maintaining good relationships with suppliers will help you out in these times.
Switching to alternative components can be as simple as a drop-in replacement on the production line, but often it is more complicated and expensive than that. We have certainly dealt with processors that were pin-compatible with the processors they were replacing, but which turned out to have strange and (naturally) undocumented properties that it took a while to shake the kinks out of. Obviously this is not work that anyone would want to do unnecessarily, so we normally recommend putting more effort into avoiding shortages in the first place. You should bear in mind what we said earlier though, that sometimes these shortages will occur with little to no warning.
The second issue that our AI prompter raised is an odd one: quality fade. The suggestion is that after the initial excitement of manufacturing a product dies down, manufacturers may cut corners to save costs. We can honestly say we have never encountered this. No manufacturer would ever want the reputation that would come with doing it.
What we have seen happen is related to the first pitfall. Manufacturers have been known to replace parts they can no longer obtain with “equivalent” ones. Unfortunately as mentioned above, parts being notionally equivalent doesn’t mean they actually behave in exactly the same way. Sometimes this causes problems out in the field, and can be hard to diagnose.
The process-driven answer to this potential issue is to implement strict QA protocols and testing, and do regular audits of the manufactured boards. Certainly QA testing is a good and necessary thing, but we prefer a more people-oriented approach. We keep open lines of communication with the manufacturers we work with, and recommend that our clients do the same. When they foresee shortages, we would much rather manufacturers tell us as soon as possible so that we can work together to identify alternatives. We can then do the testing in the lab that prevents problems from arising in the field.
The final issue the AI raised is, by way of contrast, sadly common. Especially if you have never manufactured an electronics product before, it is desperately easy to underestimate how much work goes into customer service, user documentation and after-sales support. While consultants can and do take on technical support issues, these support systems need to exist in your company even if you only do a small production run. For large runs, the systems need to scale up accordingly. Similarly, cloud-connected devices need their network infrastructure to be ready before they can be sold, and that too will be an ongoing cost and require technical support.
It is important to have these support systems in place before volume manufacturing gets fully underway. From the moment you make your first sale, you need to be ready to deal with customer enquiries and requests. Nothing seems worse to customers than not feeling they are being listened to when problems arise, and even not having a way to send thanks and praise can be disconcerting.
You also need to consider ongoing issues such as firmware updates; when such things become necessary, how are you going to distribute new firmware to your customers? The design considerations for updates will have been made earlier on, and making them available in the internet age is a good deal easier than it used to be, but you still need to have the facilities to do it.
Once again, you may have noticed that a common thread running through this article is communication. It is something that comes up time and again, building those relationships so that the people who need to aren’t deterred from letting you know when issues arise. Talking and listening are both key.