“There is no ‘minimum viable product’ in hardware,” he says, referencing the trendy notion in software that you should launch the first “good-enough” version of your idea. “One screw can put you out of business.”
Indeed; scary stuff I learned while working for Lisp Machines Inc. I heard one believable legend at the time of a company with one or more boards that used a particularly pretty blue type of bypass capacitor (https://en.wikipedia.org/wiki/Decoupling_capacitor). I think the original Multiwire Lisp Machine memory boards also used them, they were quite sharp looking, evidently a naturally popular choice.
Anyway, the guy in charge in procuring parts suddenly found them unavailable, or with too long a lead time, and was tearing his hair out before he asked the engineers what could be done. After noting it was chosen for esthetics and there were many less pretty ones that would work just as well, the problem was solved.
But if you've got something more complicated than ubiquitous capacitors and the like, not having it available at all halts you in your tracks, and that's true for every one of the zillions of parts in your products. There has been a lot of what appear to be subpar hardware designs (e.g. the DEC RK-07 disk system) that were made because of parts availability, which gets more acute the bigger the scale of your production.
When I am designing a PCB that uses a part that a) is either hard to source b) will be EOL or c) has a capability spectrum I design a pad that is the union of two pad outlines. It isn't always doable, but it makes putting in alternative packages into the same PCB a whole lot easier. One cannot do that for that the entire design, but simple stuff like supporting a 1206 and a 2512 at the same time, or a 14 and 18 pin version of a part.
What I'm hoping for is a new group of services that help with permission seeking and politics.
I don't really need help fining an expensive laser cutter to use or even a Chinese guy to make products for me. I need help getting an FCC cert for an intentional radiator. A UL listing for a power supply design. CE for Europe, whatever for Australia and Canada.
Then patent protection. Mostly protection from patents, not filing for.
Import/Export. There's countries that its a federal crime to ship electronic gizmos to. Which ones, what kinds? What forms do I have to fill out so that my products don't spend 6 expensive months in a bonded warehouse in New Jersey while I get permission to receive them?
Do I need product liability insurance? What kind? Where do I get it?
These are the thing I think makers that want to convert into businesses need help with the most. Making the actual product seems like the easy part.
First, two disclaimers. I'm an advisor to Highway1, and my only firsthand manufacturing experience is with a relatively straightforward boardgame, produced stateside (because after investigating overseas manufacturing I decided it was more complex than I could manage without help).
I don't think this is a very accurate assessment of the problems facing startups. Most of the problems you pose can be solved in a few conversations with a good lawyer or experienced operator. If you want protection from patents, you're likely going to be talking to the kind of folks who sell protection, like Intellectual Ventures; they'll be happy to take your money. There are companies that specialize in managing certifications, listings, etc. These are problems you can solve with a few phone calls and a check.
There is, however, a do-or-die problem for most startups: it is extraordinarily hard to create a strong working relationship with a great factory.
First, you have to find a great factory. That requires a lot of experience and a great deal of touring, QA, reference checking, and so on. And while the good ones are great partners, the bad ones are working very hard to deceive you in to thinking they're good. But the worst part is that most of the best factories will have nothing to do with you because you're too small. You don't have the volume to be interesting yet, they are concerned you'll go out of business, your team may not be experienced and so you'll make mistakes that they'll need to correct, etc.
Second, you need to create a great relationship with them. That means at least one person living there full-time. And that person needs to have expertise in manufacturing, QA, and so on. That person needs to feel great about their role over a very long period of time, and not everyone who moves to a new country finds it to their liking.
Third, you need to figure out a logistics supply chain. A good friend, having solved both the previous problems, nearly bankrupted his company when the shipping service used the wrong kind of pen and the address rubbed off all the labels. The return fees alone were more than the total profit on the product.
Companies like Apple rely on experts like PCH to do this for them. I believe most startups benefit from it too.
Also: PCH/Highway1 is willing and well equipped to help with most of the rest of the problems you list as well.
You may or may not find Highway1 to be a good fit for your needs but unless you're very experienced in overseas manufacturing and/or doing something truly trivial, I believe you'll have more challenges with manufacturing than you suspect.
I imagine it is quite easy by way of a few calls here and there when you know just who to call. For both regulations and for factories. Its always easy when you know who to call.
Just sayin, do both, and perhaps emphasize both. I knew just who to call to get my parts built and imported, I had no idea who to call to get permission to actually sell them. The part you don't know is always the hard part. Me, I didn't even know what I didn't know. I still mostly feel like I don't.
I guess since I already knew a guy and he just hooked me up with a small run manufacturer, I thought that it was that easy for everyone.
Dan, what's your view on your competitor circuithub, and specifically their shared part purchase(which can offer small companies volume discounts of big companies), and their own for all model?
This is awesome. I am really looking forward to the sensor technology moving forward and how the world will change when everything begins to be measured.
I just purchased a Babolat Play racket which tracks every shot you hit when playing tennis. For the first iteration, its very impressive. To walk over during a water break and see the longest rally and stuff is pretty cool. Plus, different players definitely bring up completely different stats, which is also good to see. The future for tennis I suspect is something like playsight[0], which is cameras and a base station that can record you play, give you live serve mph, and even call the lines. But for now, the Play is the most amazing thing to happen tennis. 1662 balls hit this week.
In tennis, consistency is far more important than peak performance. If you want actual improvement, simply recording where the ball was in contact with them racket is far more useful than MPH.
Babolat play gives you a breakdown of where you are striking the ball on the strings. Plus, it gives you a breakdown of the consistency in your technique. And actually doesn't tell you MPH. And although it doesn't say "40 shots in, 40 shots out" so far I think the software does give you a good understanding of your consistency. Average shots/min and longest rally, comparing that to other sessions, gives you a really good idea.
"Making Silicon Valley Go Crazy" is a bit of hyperbole, no? I would guess a quick poll amongst SV-dwelling HNers will reveal >90% haven't heard of this.
Indeed; scary stuff I learned while working for Lisp Machines Inc. I heard one believable legend at the time of a company with one or more boards that used a particularly pretty blue type of bypass capacitor (https://en.wikipedia.org/wiki/Decoupling_capacitor). I think the original Multiwire Lisp Machine memory boards also used them, they were quite sharp looking, evidently a naturally popular choice.
Anyway, the guy in charge in procuring parts suddenly found them unavailable, or with too long a lead time, and was tearing his hair out before he asked the engineers what could be done. After noting it was chosen for esthetics and there were many less pretty ones that would work just as well, the problem was solved.
But if you've got something more complicated than ubiquitous capacitors and the like, not having it available at all halts you in your tracks, and that's true for every one of the zillions of parts in your products. There has been a lot of what appear to be subpar hardware designs (e.g. the DEC RK-07 disk system) that were made because of parts availability, which gets more acute the bigger the scale of your production.