Advice for Early-Stage Hardware Startups | Hacker News Reader

Advice for Early-Stage Hardware Startups

blog.ycombinator.com

90 pointsthisisbrians34 comments

Actually being in Shenzhen where the product gets made has been really important for us.

I'm a biomedical engineer by training. My gut reaction in most phases of product development is to dive in and engineer the problem myself. It took me a while to realize this: I'll never be able to ship on time if I tried to engineer everything. Turning myself into a product manager has increased the pace of our product development by at least an order of magnitude.

Instead of taking detailed measurements of my device and building a three-dimensional model to send over, I take my product designer's sketches and a prototype to a case manufacturer. He comes up with a rough version, we go back and forth, and a few iterations later it's what I want. Oh, and in the process we've figured out what could be made more efficiently and what isn't actually all that useful.

Hardware prototyping is slow when you're doing it all yourself. If you're willing to tug on and navigate the Shenzhen supply chain and concentrated manufacturing brain power, you can iterate at speeds that can at times rival software.

Can you elaborate on the case design process? What language is being spoken? Are you meeting in person? Is he showing you CAD or physical prototypes? How much are you spending during this process?

This is so refreshing to see on HN. People looked at me like I was out of my mind in the ancient days of 2011 when I told people I was working on a hardware startup. I would only take issue with the author's advice to reconsider what you're doing if your volumes are under 1000. You can bend the cost and risk of your product launch substantially by focusing on a narrower niche and growing out from there. VCs don't like this of course, but scaling a low volume product to a high volume product is orders of magnitude easier than going from zero to a million units in a year. Following good design for manufacture principles and keeping excellent documentation is also key regardless of production size.

I've been employed most of my life by companies that never made 1000 of any one product. A lot of specialized industrial or aerospace electronics are profitable despite not being mass produced.

Yeah, that's the wrong 1,000.

Reconsider if your unit price is under $1,000.

A $1,000 per unit device that you can sell can cover up a multitude of sins because it's likely profitable from the first sale.

A $10 per unit device won't be profitable until 10,000+ units.

As a sole founder, and yes I realize that is in itself a problem, hardware is hard.

I've taken a few "purely digital" projects to the masses, but doing that with hardware on your own is freaking hard.

On the one hand, we as starters are supposed to hack, grind, iterate until we have something we can show people and gain feedback. Somewhere in there we are supposed to find other starters that are willing to take the plunge with us.

In the digital realm this is easy because you can make fruitful progress in a few night's work.

The the hardware side you can work for months to demonstrate something that looks like dogshit and does what you say it does.

In the pure digital world, you don't usually have to separate "looks like" from "works like", unless you are just at the absolute earliest stages.

So is hardware hard? Hell yes.

Is it harder than "pure digital". Hell yes.

Are things making it easier, and does exposure to fab labs and maker spaces make it easier? Yeah, a little bit.

Nothing about "The FTC will sue you if you market an insecure device and put consumers at risk by using it."

If you're not aware, the FTC has put a tremendous amount of resources over the last ~2 years towards enforcement actions against insecure device makers and defining a minimum baseline of security that all device makers should meet. The likelihood that you will be named in one of these suits has risen quite a bit during this time period.

http://www.ftc.gov/news-events/press-releases/2015/01/ftc-re...

http://www.ftc.gov/news-events/press-releases/2013/09/market...

http://www.ftc.gov/news-events/blogs/techftc/2015/02/whats-s...

Excellent, practical advice, thank you!

Does anyone know if "#2 Don’t forget the certifications" apply to very small / simple hardware ideas for ex: harware products / designs that use open source kits like the arduino kit?

I am working on a product based on an arduino kit, and want to make sure I have the certifications part (if needed) covered.

I am guessing that these certifications are more for big complex hardware products like electric cars etc? or No?

I'd say it's a mixed bag of safety, liability, and legality that will vary widely between products. Some quick examples:

Safety: if you sell an appliance that operates on mains (hazardous) voltage, there are standards in place (UL, etc.) you should be following in order to prove your design doesn't allow that high voltage to easily come in contact with a person. So for your Arduino thing, as long as the power adapter that you source is already certified, you are probably good to go.

Liability: Closely related to above; what if your device is involved in a fire? There are flammability requirements for devices for different environments, and certification will help you prove you did everything you were supposed to in order to prevent your device from starting or propagating a fire. Again, not as big of an issue once the voltage is stepped down to <= 5 VDC.

Legality: mainly, FCC (in the USA) comes to mind. It's actually illegal to sell radio devices using certain frequencies unless they are tested and approved by the FCC. You can buy an FCC-certified radio modem or module to get by this one quite a bit easier. Medical devices are an entirely different can of worms (FDA, etc.).

So, it really depends on the use case for your product. What are you planning to build?

Even non-transmitting devices can be "unintentional radiators" as per the FCC. Using pre-licensed components doesn't exempt you from testing and appropriate certification -- you're on the hook for the entire product. Furthermore, submodules might only be certified for certain configurations which your device may violate, and thus you would require certification.

Basically, there are three kinds of procedures you need to worry about: verification (the least intrusive, requires that the manufacturer do tests but does not require approval); declaration of conformity (requires a certified test lab certify the product); and certification (the "FCC Code" -- requires an application to the FCC or a TCB, including technical specs and official test results). The only way to be sure is to have engineers on staff or on contract with experience in licensing issues.

By the way, if you're selling a kit product, you may be exempt from much of the approval process, which may influence your decision.

Not that I've gone this way, but I'm starting to believe that certifications are like patents - not worth the money if you're very small. The FCC fines are huge, but are they focused on the little guy? 3D Robotics seems to walk this line, by a talk I saw recently.

I'd consider waiting until you know there's demand past Kickstarter, at least. I see some worry about CE and the individual countries like Japan and ANZ, when global domination is a long ways off. You're probably going to have to iterate again anyway in order to get retail-ready margins.

A clue that it matters little is that when they need to put your unique product in a government-designated category, it's going to be something like "Computer Peripheral" or "Measuring Device". There is a lot of grey area. Hell, CE lets you self-certify.

Another debatable early-stage expense is liability insurance. Technically most distributors require it, including Amazon. But it can be nearly impossible to get for hardware with no track record, never mind what it costs. The best option is sometimes to let your LLC do its job and go bankrupt if someone wants a lot of money.

Does not sound like a terrific strategy to me. It probably will become very hard to bolt the requirements on after you've finished your product. And for people here in Germany, the pebble missing CE seals was a great deal as the shipped products got lost in customs. Pretty easy to piss off your backers like that.

I'm not saying to ignore the radio requirements at all. If you're diligent, make liberal use of reference designs, etc. your design will pass. I (and Chris Anderson) just question whether the time and expense of getting certifications that no one will notice is the right place for a startup to spend its resources. Fair game, when so many skip arguably more important things like security.

I agree on Pebble - the volume they were sending through, and a little inexperience with customs, raised flags that were hard to put down. But by all means, when you've raised $10mil, get your certs. I'm telling the Kickstarters who barely raised enough to execute their hardware, and then think that blowing $20K in order to cross all the t's for international markets, is not putting the cert before the horse.

It's not how simple the product is or whether it's open source that determines regulatory requirements, but what it is and where and to whom you're selling it.

For a quick primer on FCC requirements (which only covers US EMC), check out Sparkfun's "The FCC and Open Source"[1]. For more depth on other regulatory requirements, e.g. safety, "Global Certifications for Makers and Hardware Startups"[2] is pretty decent.

[1] https://www.sparkfun.com/tutorials/398

[2] http://www.emcfastpass.com/cert-ebook/

Kits and test equipment are generally exempted from FCC, if they are unintentional radiators [1]. Unintentional radiators don't get certified or get an FCC ID, instead it is based on self declaration. You should do the test (a couple of hundred dollars) to make sure it passes. There are a number of labs locally and abroad that will do it. It might make sense to use a local lab for pre-scan and then pay an off-shore lab to do the full test.

If you have any kind of radio transmitting part, i.e. an intentional radiator, then you should use a pre-certified module or otherwise pay about 5-10K to get it certified.

1. https://www.sparkfun.com/tutorials/398

> Always think about bringing your capabilities in-house, if still outsourced -- EE, ME, firmware, ID, apps, frontend & backend dev.

This. 1000x this. The temptation to outsource everything is huge in the beginning, but the margin stacking and 'time stacking' this creates can be fatal.

I'll add another power of ten to yours. Just like you gotta be a full-stack developer to be lean and mean in your early-stage web startup, you gotta do/learn as much of the hardware stack yourself to save money and keep hard-won knowledge in-house. Given how capital-intensive hardware is, you should be even more stingy.

Not that everything must be done yourself any more than you should reinvent AWS. But here's an extreme case of outsourcing: https://medium.com/@stevekreyos/the-rise-and-fall-of-kreyos-...

This is all solid advice. To me, this reads like page one of a hypothetical book "So You Want to Start a Hardware Company." There's a lot to learn about building a real hardware product. I'd love to read the rest of the chapters in that book, written by the people who have been through it.

Though I would say that you probably shouldn't spend the time creating a new hackerspace. As someone who has been there, starting up and running a hackerspace can be a real time vampire. Joining up with your local space is great - you'll meet a lot of smart people, have access to some useful tools, and maybe even find some potential future employees. But I really don't think it's a good idea to start a hackerspace at the same time you're trying to start a company.

Finding new hires from your local hackerspace is great advice. You get to observe and work with people before hiring them. BTW, If you are ever in Oakland,CA you owe it to yourself to visit the Omni Commons. I'll be there attending CCL's syn-bio lecture this Saturday at noon if anyone wants to discuss hardware startups afterwards.

I work in the bay area with all-local manpower, ask me anything. We bootstrapped a line of products with absolutely no help from banks or accelerators.

What kind of products do you make? What margin do you need to support local manufacturing? What is your main distribution?

Laser cutters and telepresence things, about 80%, my website and a few resellers. http://www.robots-everywhere.com

Here is a really good video describing how a low production hardware product got to market. https://www.youtube.com/watch?v=vIvTeHWfouA

This is a great article full of a lot of good advice. Hardware can be very hard for a lot for reasons and you can make expensive mistakes especially if it's your first time bringing a product to market.

We successfully bootstrapped some hardware to a shipping product and made it through the Apple MFi program.

If anyone wants to chat drop me a line. My email is in my profile and I'm happy to help If I can.

Is there a List out there with the best hardware-producing companies... I mean, what if i have the layout for a beagleboard 5.0 and want to order 2k of them...

I find it pretty hard to get an overview over hardware factories...

"Join or start a Hackerspace, work from Techshop, contact makers who post interesting projects: find people working on hardware like yours and ask how they dealt with challenges you’re facing. These conversations have led to me discovering faster and cheaper ways to make SMD stencils, casting aluminum parts from 3D prints, sourcing cheap components direct from China at in-country prices, and taught me everything I know about making things."

anyone have thoughts about how much IP protection someone should have before talking about their ideas at a Hackerspace or Techshop?

Does it make sense to file provisional patents before working on an idea at one of these spaces?

Nobody is going to steal your unproven hardware idea at the TechShop. Don't worry.

at what point in the processes of proving a concept does it make sense to get IP protection?

When you realize that you will make more than the patent costs, plus the legal fees to sue an infringer.

Before then, I'd say you have better uses for the money.

A provisional patent is around $400 and lets you slap Patent Pending on the manual, which makes you look more serious, at least.

[shrug] So you just paid $400 for the right to sue someone. Still gotta pay the lawyers.

It also makes you look a bit better to the sort of people who care about patents (The problem with lawyers is that they quickly get more expensive than siege engines, so their usefulness is limited).

A provisional patent is $65 if you are a micro entity, $130 if you're a small entity. You have one year after filing your provisional patent to file a "regular" (utility) patent based on the provisional, and your utility filing gets the benefit of the provisional filing date.

See fee schedule, refs of micro & small entities, etc:

http://www.uspto.gov/learning-and-resources/fees-and-payment...

Nobody gives a shit about your idea until it's worth something, usually because you've built it.

Don't be coy--nobody likes that guy.

Reply on news.ycombinator.com