Project update 15 of 36
Happy New Year!
Despite the holidays, we’ve managed to make tons of progress toward getting your units manufactured. Of course, in the Asia-centric hardware world, the looming deadline is not the Christmas holidays, but the Lunar New Year. While our South Korea-based factory only takes a few days off for the New Year holiday, much of the component supply base is still in China, which takes an extended leave. As the risk of COVID-19 is still very real, the logistical headaches that typically accompany the holiday season are likely to be spread out over a longer period.
Thus, we’re pushing to get certification-ready prototypes built before the factories start the process of closing up for the holidays, which may happen as early as the last week of January. The "certification" we’re seeking is electromagnetic compliance (EMC), often referred to as FCC or CE certification. It’s basically a set of outdated laws originally drafted to protect the function of now-obsolete AM radios. In modern times, this law is mostly an economic and trade barrier to prevent small companies and newcomers from competing with established interests. The process is hard enough to be expensive and time-consuming for small companies, but easy enough that big companies can shrug it off with a few full-time staffers and a million-dollar facility dedicated to the problem. Ironically, despite its apparent rigidity, the certification process is full of loopholes, so AM radios still get jammed and radio astronomers still have to deal with interference. So, in practice it does little to accomplish the intended regulatory goals, but is still quite effective as a moat to keep out small players who can’t afford tens of thousands of dollars for the test fees, or lack the specialized skills and industry connections to adequately prepare for such tests. Thus, there is minimal incentive for big companies to invest the effort to lobby governments to reform and update this archaic system.
Personally, I age ten years in two weeks when running the certification process. Sometimes I’m spending days chasing down a transient blip that’s just a decibel over an arbitrary line in a largely disused band. All this is done while paying hundreds of dollars an hour to rent a test chamber and knowing that if the product can’t be tweaked to pass, the penalty could be several weeks and thousands of dollars spent on a full product revision to fix the underlying issue.
Regardless, part of the rules is we must have a final prototype before entering the certification process. To that end, I’ve finalized all the small design-for-manufacturing improvements on the case and started the tooling for the injection molds. I’ve also put in the order for the final PCB revision and we’ll hopefully get samples of the stretch-goal upgraded 1300mAh battery before the end of the month.
A number of subtle changes were made to the design to facilitate a broader range of materials for the omakase case options. The standard production case is aluminum with an anodized finish. Anodized aluminum is non-conductive, which is great from a safety perspective: if a wire goes loose inside an anodized aluminum case, it’s not going to short out. However, many other metals do not feature the requisite surface chemistry for anodization and so some other form of protecive coating is required, which may or may not be as robust as anodization. So, in the run-up to final production, I did an audit of every internal connector and tighted up the design and tolerances so as to minimize the chance that a live wire could come in accidental contact with the inside of the case.
We also recently got all the necessary firmware modifications done for certification testing. The Silicon Labs WF200 chipset was one of the easiest chips I’ve worked with from the perspective of certification. It turns out everything I needed to build the compliance firmware image was already in their GitHub repo, despite their FAQ indicating I had to reach out to an FAE! Usually it’s like pulling teeth to get proprietary and minimally documented certification firmware from the vendor. While I had to do a little bit of reverse engineering work to adapt Silicon Labs’ protocol to Precursor’s bespoke hardware-firewalled environment, that was about it. I’m now able to command the radio to produce continuous transmission streams in each of the required modulation modes and in every one of the required channels for certification testing.
With any luck, we’ll have certification-quality prototypes built just before the Lunar New Year shutdown. The plan is then to use the downtime while the factories are on holiday to thoroughly test the certification samples and to build test jigs, so when the factories come back from holiday in early March, we’re ready to start certification testing and, assuming a favorable test outcome, push the button for production!