Step 1: have capital. Almost everything related to hardware has either a minimum order quantity or is uneconomic at small volumes. So you need to plan to make 1,000 units and pay for them before you receive any customer money.

Don't forget this is the supply chain crisis era, so you need to make sure you can get 1,000 units of all your parts that are actually in stock and not on 24 month back order.

The other comment that says "Scaling a hardware product is probably 10-100x harder than scaling a software product" is absolutely correct.

Once you have a prototype you can Kickstarter, but a lot of hardware kickstarter projects fail. https://help.kickstarter.com/hc/en-us/articles/115005134554-...

Step 2: you need to be either a very fast learner across a lot of subjects, while spending your own money, or hire some consultants who know what they are doing. I used to work for a product development consultancy, but it's a pretty niche market.

This includes minefields like UL and CE certification.

Step 3: you need to either pick "local" manufacturing, where you can build up a working relationship and they will genuinely help you, or "remote" manufacturing which costs less but you have to diagnose all the problems over the phone or get on a plane to them. The latter is much harder in the COVID era.

Step 4: assembly, shipping, taxes, payment, etc; some of the local firms will also handle this for you (and take a cut).

The first 15-20 years of my career were spent as a hardware/embedded engineer and I'd like to add a few notes to the list above:

5. Unless you've got very specialized instrumentation, the power supply and case will dominate your costs. Per step 2 above, these will also dominate your timeline if you include an internal power supply.

6. FCC certification is also probably needed if your device is consumer oriented.

7. Expect development, time to first article, etc to be much slower. In my day, we couldn't have customers update their Flash-based firmware so we had to guarantee the hardware was perfect AND that the firmware was perfect. In your day, you still have to make sure the hardware is perfect and the reality is that very few of your customers will ever update the firmware (so it has to be perfect with each version.

8. For a small business, hardware returns are not only painful but have real costs. A recall will put you out-of-business (and if you're not careful with your corporate structure could put you into personal bankruptcy.)

5. Which leads to: for small volumes (or MVPs or early-stage prototypes), if you can buy your power supply, you should!

6A. Most products with a microcontroller will need verification that they are compliant with FCC regulations on unintentional radiators. That's relatively cheap to do. If you bought your power supply (see 5) you are probably compliant. The problem is... by the time you get to this point, you are behind schedule, money is getting tight, and any respin or even adding ferrites is going to be a huge headache. Also, what got you here... might not help you solve an EMC problem. Good luck!

6B. If you need a transmitter (i.e. bluetooth or WiFi): for low volumes, if you possibly can buy a prequalified module, do so. Then certification is just paperwork at most. But don't listen to me, do a chip-down design because your industrial design requires it. YOLO and there is no other way to learn. 2.4 GHz at WiFi SoC powers is pretty forgiving. BLE with its low duty cycle practically handles compliance for you. Buy your matching networks and filters and don't go crazy with your antenna design

7. In general expect to be surprised by how the engineering - hardware, software, mechanical, etc - involved is much more exacting and demanding than you might expect from say, how conceptually difficult your problem domain might be. Especially if you are a software engineer and haven't worked with physical products

8. See #7 - you should be doing everything you can in design and procurement to minimize customer issues, from UX to field failures to any abuse of the product you can reasonably imagine. (Misuse you can anticipate has product liability ramifications also) If you are building 1000 units, any returns are going to cut into your margins. Look up typical warranty return rates for your industry and work that into your business plan and your pricing

9. They will all tell you hardware takes 10x-100x longer than software (I think this is a software engineering perspective where they expect to be able to deploy from a schematic or something), but then they will also tell you that if firmware isn't making you late, you aren't getting enough value out of firmware. You figure it out

I'm surprised to hear power supply and case are the biggest issues since those are needed by everybody. What keeps them expensive?

Economies of scale, basically.

Your options for a case are to go full-custom, which is expensive in small volumes, or to go semi-custom, which involves buying a case and having holes machined, which is expensive.

With power supplies, you probably should buy off-the-shelf, and in small quantities that is going to be relatively expensive.

I've never seen the case and power supply dominate costs on any of the designs I've shipped, though. The poster might have very particular power requirements.

Some folks also get hung up on how much cable harnesses and connectors cost. I think it's because they conceptualize the design in their mind without them.

> Step 1: have capital. Almost everything related to hardware has either a minimum order quantity or is uneconomic at small volumes. So you need to plan to make 1,000 units and pay for them before you receive any customer money.

An interesting corrolary to this point is that somewhat counterintuitively it is often easier to aim for the highest segment of the market you target with a niche product than for broad appeal.

Having a very high markup and selling price means that even small batch become feasible. Sourcing for high quality part sellers generally means you will have less issue with quality control. Selling less means shipping and inventory management is simpler. Marketing is easier because your customer base is more narrow. As a bonus, high margins give you more leeway to make mistakes.

> you need to be either a very fast learner across a lot of subjects, while spending your own money, or hire some consultants who know what they are doing. I used to work for a product development consultancy, but it's a pretty niche market.

Would it possible to learn this stuff by getting a job in one of these consultancies somehow? And if so, how might one go about doing this?

It's also possible (and possibly faster!!!) to learn by trying to bring a product to market yourself. There are many small engineering consultancies around the US; getting a job there is like getting a job anywhere but they seem to be disappearing. I worked for one for five years then moved on as it was clear that I didn't have a growth path there.

Here's one of the better-known ones in my area: https://www.linkedin.com/company/logic-pd/

Usual process of "find them and ask for a job, or work your contacts", I suppose. I know that's not terribly helpful.

I have remembered about Bunnie Huang though, who wrote a book on the subject: https://www.crowdsupply.com/sutajio-kosagi/the-essential-gui...

He blogged extensively about the "chumby" product he did about a decade ago, and the "precursor" he's working on at the moment. https://www.bunniestudios.com/blog/?cat=6

I work for exactly one of these product development consultancies / contract engineering firms - we are a team of ~10 EEs and SWEs who work with customers from idea to design to development and then manufacturing. To the point of "lots of subjects", the above includes but is not limited to: sales, project management, electronic schematic capture and PCB layout, design for and manufacturing support, web development, and mechanical/system integration. Often we'll collaborate with our mechanical engineering counterparts who've opened their own product development and contract engineering consultancy.

I can't speak for all firms but the way we find our employees is through networks of current employees and through LinkedIn recruiting. You'll want to emphasize aspects of your resume/work that mark you as a self-starting generalist with key skills.

An example on the generalist side: we had a candidate say, "I don't want to do design, I just want to be handed a schematic and do PCB layout." They were passed over in favor of another candidate willing to dive in and do schematic capture if necessary, PCB layout, and speak directly to customers about requirements, timelines, budgets, and other project planning logistics.

An example on the key skills side: the weakness of generalists is of course lack of depth, and we'll often contract out to SMEs based on project needs. If you know of a product consultancy near you that's, say, mechanically-focused, you can offer e.g. SWE expertise on, for example, frontend/backend development for a site that accepts RESTful HTTP requests from the IoT device they're building a mechanical enclosure for, whose customer needs software expertise that the MEs don't have.

To find these companies, useful search phrases are things like "product development", "contract engineering", "engineering consultancy" and so on.

You found a PCB layout designer, not a design engineer. A good one is a real force multiplier that can lay out boards faster and more professionally than pretty much any EE I've worked with that lays out his own boards. You might not have enough work to keep one busy, though, and plainly that wasn't what you were looking for.

I agree with all your individual points but I’m slow in understanding your overall message, sorry.

It's a worthwhile specialization, but I have no idea how broken your recruiting pipeline would have to be to consider a layout designer for a "EE generalist who lays out his own circuit boards" sort of position. The typical layout designer does not have a BSEE or a resume that would indicate proficiency at circuit design - let alone project and customer management skills - it has historically tended to be a technician or draftsman level career path.

This was indeed an interview for a tech position. I was exaggerating about the design aspect so I’ll retract that. We do however have techs who interface with customers and who coordinate projects. From your responses I’d guess this is very unusual and thus my experience is a bit of an outlier. Maybe you could give the original asker better advice, I think anyone reading would benefit if you did.

I thought your advice to look for contract gigs was pretty good. I'd only add that I worked in that space for a while and consultancies are generally either desperate for work or desperate for help. If they're desperate for help they will be busy enough you will have to call and call to get them to talk to you, but if they don't completely shut you down, keep trying.

Or go work for a company that makes products. You will learn a whole lot of things about bringing a product to production that you will not learn at pretty much any product development consultancy.

Note that right now in 2022 just about everyone is desperate for help. Be sure you like doing parts substitutions.

I've hired layout shops where we worked with the designer more or less directly but at "whole product" design shops, if they even have a layout designer, in my limited experience it's not usually a customer facing position.

I have a couple of friends who work here https://www.synapse.com/careers

Sure, all of that is true of mass market consumer electronics businesses/products.

But a lot of the world makes (or peices together) fewer, higher budget products/projects and sells them to more focused markets (boutique hotels, regional CVBs, breweries, etc).

All of the above, but consider if you are going international. Be aware of rules regarding exports from the US, ITAR, and EU regulations.

I've been doing this for the past 18 months with a Raspberry Pi-based network administration tool called TinyPilot. I originally launched it on HN in 2020[0], and there was enough interest that I turned it into a real product[1] and have been iterating on it ever since. Like you, my background is in software, and I had no hardware experience aside from using the Raspberry Pi as a cheap server on a few projects.

I've never found a good guide on how to get started in hardware or eCommerce, but I've been posting yearly[2], monthly[3], and weekly[4] updates about my experience learning as I go.

One thing I wish I knew at the beginning: it's difficult! Scaling a hardware product is probably 10-100x harder than scaling a software product. There's nothing like GCP or AWS that scales from one to thousands. Vendors that offer fulfillment services won't work with you when you're only selling tens to hundreds per month, so you have to do everything yourself. I built my own fulfillment center[5] and hired people to staff it, and that's complicated because that gets into inventory management, employee payroll and benefits, insurance, etc.

The difficulty is a double-edged sword, though. It prevents people from competing with you because nobody else wants the hassle. But the downside is that you're the one who has to deal with the difficulty.

[0] https://news.ycombinator.com/item?id=23927380

[1] https://tinypilotkvm.com/

[2] https://mtlynch.io/solo-developer-year-4/

[3] https://mtlynch.io/retrospectives/

[4] https://whatgotdone.com/michael/project/tinypilot

[5] https://mtlynch.io/retrospectives/2021/05/

I write this with true love for your effort, but do I read your 2021 annual retrospective correctly stating that you made $13K on $477K in revenue? Maybe some of your expenses are paying yourself for software development, and that's just your corporate income?

Consider reading this book:

https://www.amazon.com/Sell-Margins-Higher-Than-Competitors/...

I would really consider raising your prices. I found that when I first started my business, I had a subconscious desire to replicate my income at my previous job, which caused me to irrationally underprice my products. After 2 years, I got over it, and it transformed my life. You've done an amazing amount of work. You should get paid for it (or, if people won't pay for it, get out). My dad used to say, "I don't do work for practice."

>I write this with true love for your effort, but do I read your 2021 annual retrospective correctly stating that you made $13K on $477K in revenue? Maybe some of your expenses are paying yourself for software development, and that's just your corporate income?

Thanks for reading!

No, that doesn't include paying myself. But it's also worth mentioning I also had $59k in raw materials at year-end. If I had to liquidate everything, it would cost about $5k to assemble the raw materials into ~$200k of product and ship them to customers, so I'd theoretically walk away with $195k.

I've always had a hard time assessing the financial health of the business because I'm investing a lot into future growth. Because of the supply shortage, I have to stockpile a lot more raw materials, so as sales grow, I'm also buying more materials to meet the new scale. And a lot of my costs are one-time events that I expect to pay dividends over time (e.g., hiring EE vendors to design a new product, hiring a design firm to overhaul the website, paying developers to add features).

>I would really consider raising your prices. I found that when I first started my business, I had a subconscious desire to replicate my income at my previous job, which caused me to irrationally underprice my products. After 2 years, I got over it, and it transformed my life. You've done an amazing amount of work. You should get paid for it (or, if people won't pay for it, get out). My dad used to say, "I don't do work for practice."

Thanks! This is something I have trouble with because it's hard for me to measure the optimal price. My volumes are too small to really A/B test different prices effectively. Over the course of the business, I keep raising the $50 thinking the new price is crazy, and then people continue buying anyway.

> I keep raising the $50 thinking the new price is crazy, and then people continue buying anyway.

Yes! It is your business, and you know best, but this sounds promising to me and reminds me of my situation. That is why I recommend that book. It argues compellingly that you want to sell less at higher margins, all things being equal. That gave me the courage to raise prices, and I ended up selling just the same/more at higher margins. Raising prices takes courage, and clearly there is some point where you go too far, but it's worth really pushing yourself emotionally to consider it early in the business. If I had not radically altered my pricing 15 years ago (15 years of massively higher income potential and counting! crazy) my life would be so different than it is now.

Rather than thinking of A/B testing, I would try to think of how much value your product adds for your customers, and then capture some reasonable percentage of that. Think of it very broadly (again, see the book). There is more value here than just the hardware/software. Your customers may value dealing with you instead of dell, etc. Try talking to some good customers about how much they would really pay. Some won't feel comfortable, but others may blurt out "we spent $3k on your hardware last year, and the truth is it's a rounding error for us. We just want someone who delivers a solution and answers the phone in person (or whatever)"

Anyway, I wish you the best of luck. After writing my original comment, I realized you had only been operating for 18 months. That, combined with your inventory assessment (I am in software, and have no inventory, so I didn't think of it), seems very promising!

Sounds like you might have a software startup rather than a hardware one.

Most of the value may be in the software for sure, but TinyPilot is a KVM (I had to look it up; knew I recognised it from somewhere but couldn't think what it was. Think I saw Jeff Geerling review it.) so there's hardware involved, inventory, a product to ship.

Firstly, as you can see from many replies here, this being HN some people think hardware startups are companies that put software on computers and sell the computers (e.g. developing a raspberry pi based tool for something). Let's call this hardware-lite.

It's obviously a bit of a 'spectrum', but I'd say a 'real' hardware startup would be one that develops new hardware, like Tesla, General Fusion, a robotics company, etc - really anything where the innovation is in the hardware, and if software is developed (usually will be) then it sits on top of that novel hardware. Let's call this case 'hardware-heavy'.

In the hardware-lite case, you're basically starting a software company that needs to set up a supply chain for a few parts, built by others, which provides the most cost-effective platform for your software to run on. You might then package that up, brand it, and sell it. This can be relatively capital nonintensive.

In the hardware-heavy case, you'll probably need quite a lot of capital. This will be needed for you to patent your invention(s), prototype your devices repeatedly until you've arrived at something commercially viable, set up a small batch manufacturing process to produce small volumes of prototypes. With hardware, when you ship something anyone can take it apart and reverse engineer it. Every time you replicate your product, it takes time and money - unlike software.

Tons of commenters here will be able to give you excellent advice about the hardware-lite model, because this is HN. I suspect there is less experience of hardware-heavy business models here.

Which fits your vision?

> In the hardware-lite case, you're basically starting a software company that needs to set up a supply chain for a few parts, built by others, which provides the most cost-effective platform for your software to run on. You might then package that up, brand it, and sell it. This can be relatively capital nonintensive.

True to a point.

Beware that even if you're putting your software on to someone else's hardware, it is ultimately you who becomes the "manufacturer" in the eyes of the consumer and the law.

So if your batch of Widgets that you bought from XYZ suddenly start setting fire to people's houses, that falls on your doorstep (yes you may have an upstream claim against the actual OEM, but it is you who will be responsible to the end users).

Similarly if you're getting involved in safety critical stuff, then buying OEM hardware will only take you so far. Regulatory approval for the final package will still be on your watch and your tab.

I've never scaled production, but having done hardware prototyping this sounds right to me. You can make some pretty clever thing with off-the-shelf components from DigiKey or Mouser, a soldering iron and a 3D printer. You scale 3d printing a bit with a partner like Xometry. Or even ordering custom fabricated plastic or metal. The team I had to build was primarily software with enough hardware knowledge to make things happen. The biggest gap was hiring an industrial designer who could do the details drawings of how to fit all the components and make sure the external dimensions fit the usage and had some style. We even managed to wrangle UL certification for a few things we built which is onerous and costly, but surprisingly not too complicated.

I agree with the wisdom that scaling software is orders of magnitude harder than scaling a software product. However, I would say B2B hardware is much more doable than B2C hardware. With B2B hardware you might find that you can sell equipment in the $10,000+ per unit range and if you can keep your BOM cost to <$2000, which is usually doable with low volume manufacturing, without needing scale, you can get to a "commercial" product a lot quicker.

> B2B hardware is much more doable

This is a point that usually gets lost in these discussions. The type of market makes a huge difference. My "side businesses" have always targeted B2B or niche B2C services and I have a lot of experience doing that. Businesses are far less price-conscious than individuals.

However, I also worked for a company making consumer wearable hardware and the approaches you have to take, along with the minimum viable volumes are night and day. Also, consumer products tend to require certification testing that industrial ones don't.

I did this with a COVID screening kiosk; we sold 120 units at $4,000 each, BOM costs of about $1200, about 1600 hours of engineering time and 4 hours each of fab time - you could buy Chinese kiosks cheaper than that by a large margin, but we integrated with customer RFID tags, time clocks, and door lock systems on a per-customer basis. It was helped by the fact that we had an industrial laser cutter and press brake to fab the mechanical aspects, a large network of existing customers, and a team of MEs, CEs, and fabricators to put everything together who were short on our usual B2B automation equipment work.

I recently chatted with the creator of TinyPilotKVM (https://tinypilotkvm.com/) which lets you remote control a server without installing any software on it. It's a KVM over IP device that you plug into your machine. He sells a few hundred of them a month after launching in mid 2020.

The podcast episode is at: https://runninginproduction.com/podcast/105-tinypilotkvm-let...

It covers everything from how he got started, starting with a proof of concept, 3D printing and refining the process as he went.

I've never done that myself, but I backed an SDR device called Flipper Zero on Kickstarter some time ago, and I found their updates on the project status very insightful.

Through their posts, you can make an idea on how complex and how much work there is on developing, manufacturing, certifying and delivering a hardware device.

Link: https://blog.flipperzero.one/

Also read about the saga of creating the Keyboardio 01:

https://www.kickstarter.com/projects/keyboardio/the-model-01...

They went through hell to get it manufactured. Shady companies, the whole nine yards!

Quick plug for one of my favorite podcasts right now that had an episode with the Keyboardio creator and their current struggle with the chip shortage[0]. Highly recommended listening!

[0]https://techpod.content.town/episodes/97-how-to-survive-the-...

Hi, I'm Jesse. I'm 1/2 of Keyboardio. We have, indeed, been through some stuff, but don't let that scare you off of hardware. We're currently in manufacturing for our third keyboard product.

We're generally pretty happy to chat with first-time hardware founders. Feel free to drop me a note at jesse@keyboard.io.

Me too, and because of it, I'm not even mad that they're a couple of years or so behind schedule now, the journey is fascinating. Fantastic blog.

I'm a web designer/builder and my wife and I invented a non-tech related, physical product that we brought to market. Feel free to email me.

After we built our prototype we hired a local machinist to make a "mold" that would produce higher quality (sellable) examples.

We then made (and sold) those in our basement for months, until we had enough traction to hire a CAD designer to make a CAD file for the purpose of finding a manufacturer.

We started cold calling local manufacturers in the US and were basically told to "f*ck off" as we were too small.

We finally got a lead by contacting another business with a "similar" product (or at least process). They referred us to their manufacturer in China, and we jumped on it.

We ordered generic samples of similar products they make and were happy. We then had samples made (after paying for the "mold" to be machined) and those looked good (we lucked out) and then finally ordered our first 750, and then subsequently higher quantities.

We then had another business owner who had experience with importing help us with customs/importing etc.

Ask for help! Find other small biz owners, they're usually eager to assist.

I've brought a couple of low-tech physical products to market in the last few years. There is no foolproof playbook, so I'll give you some advice in no particular order.

- Be prepared to spend 18 months bringing even the most trivial product to market.

- You find out literally everything having to do with manufacturing by talking to manufacturers. Start Googling, calling and emailing.

- B2B bares almost no resemblance to B2C. No one will respond to your emails. Almost no manufacturers will want your business. Even if you are buying components that require customization with cash, they still won't want your business. All you can do is keep annoying them until you find a salesperson that takes pity on you. I was able to get a key component for my first product made because I happened to find a brand new salesman that needed something to do.

- Don't worry about building the thing right until you are very, very sure you are building the right thing. Failure to do so will result in wasting of at least $10K. Spend at least $2K validating any product idea (landing pages, Facebook ads, product renderings, product video). Spending $2K to kill a bad idea you would have wasted $10K on later is good ROI.

- Be as public about building your prototype as you can be. Get involved in communities of people, wherever you find them, that you think would be your customers and post there. If they feel invested in what you built, they will like you and maybe buy it. If you show up after it is done and try to sell it to them, they will hate you. If no one in these communities seems interested as you're prototyping it, then absolutely no one outside of the community will want to buy it.

- Don't start trying to produce the finished product until people are begging you to sell it to them. I am not exaggerating when I use the word begging. If someone tells you they would buy it if it was available, they probably won't. If that person is your friend or family, they are likely lying to you because they don't want to hurt your feelings.

- Do not, under any circumstances, think you can pay someone you've never met to solve a problem for you. (See other comments about you needing to become an expert in everything.) They will to a horrendous job and charge a fortune. If you find a person you can trust, hold on to them like a hungry octopus.

- You need to have customers waiting when you first ship. Build a pre-sales mailing list. There are countless blog posts on how to do it so I won't rehash it here.

This is sound advice for pretty much any product, not just hardware.

Indeed. The main differences are 1) lead times 2) financial risk. Everything is slower and costs more when you go from software to hardware. Like 10-100x.

The difference that most shocked me is just how unmotivated manufacturers and resellers will be to get your business. You have to audition, and sometimes plead, to give them money. The reason is because the cost of taking on a customer for hardware can be very high. This is much less true for software.

It depends entirely on the item (and how it is manufactured) and the number of units you plan to sell.

There is quite clearly a difference between selling 100 items that are designed to be simple enough for you to manufacture in your garage, to selling 10,000 items which require injection moulding, custom PCBs long assembly steps, finishing, QC/testing, certification etc, are clearly two completely different challenges.

But the first step is to build a prototype that people want, and then think about how you can make it easy for yourself to assemble (because chances are that will also make it easier for someone else to assemble).

Any of Ben's posts before he left Bolt VC. https://beneinstein.medium.com/

Especially:

1. The Complete Guide to Building Hardware Startup Teams

2. The Illustrated Guide to Product Development

Not sure if he is still writing at Eclipse

I wouldn't worry too much about manufacturing right now. Just build prototypes for now, you're likely to go through many iterations anyway at first. You can sell and market prototypes just fine, people are surprisingly willing to put up with a lot if it solves a real problem for them.

Once you're at the point you can't fulfill all your orders just with prototypes, you can start thinking about scaling. At this point you will have much more experience. It's also possible to outsource quite a bit of the whole design. It's quite possible to mostly do software despite selling hardware.

MIT's How to make (almost) anything is a good place to start if you are new to hardware prototyping: https://fab.cba.mit.edu/classes/863.20/

I would also recommend that you join any local Hackerspaces in your area if you need access to special equipment: laser cutters, CNC machines, 3d printers. Peeps who frequent these spaces are very knowledgable and surprisingly helpful with n00bs.

As others have already mentioned here, don't worry about going into production for now because A LOT of stuff is in flux due to the pandemic. Most people who are serious about building hardware either end up moving to Shenzhen or spending a huge chunk out of their year supervising the manufacturing in person. If you are not ready to make this commitment yet then it's better to focus on making your prototypes locally with what you already have on hand.

Cool to see this! I took one of the first-ever MAS.863 classes in 2001... looks like those archives are lost to time. Probably for the best.

Keep in mind that prototyping is completely different from design-for-manufacturability (DFM), which is its own engineering discipline.

I've gone down this path quite recently and got pretty far. This is for a low-volume, hand-built, high-margin electronic device that is likely to find a home mostly among hobbyists. I've tackled most challenges: learning how to build a PCB, learning how to solder, figuring out how to build the enclosure for the device and programming the firmware. I'm effectively ready to sell it.

The remaining hurdle appears to be CE certification (or FCC if you're in the US). I know of some competitors that are doing similar things to me without CE, but I am really not comfortable doing it this way. While self-certifying for CE is doable it does not appear easy and if you screw it up it sounds like you can get in a lot of trouble. I'm still not sure about the best path forward but am likely to do my best with self-certification and set up a LLC in case things go awry.

If anyone has any tips regarding certification specifically please do share.

A relative designed and built a physical product and when through that problem.

According to his experience the hardest requirements were about high voltage and electromagnetic interference (both how your device interferes with others, and how others affect yours). He also had mechanical requirements but that was fairly easy to test himself.

He was able to design his device to not use any high voltage. Basically buy the power supply off the shelf and don't enclose it in your product.

For EMI, it's hard and expensive. Specialized companies will test and certify your device, but this is too expensive. He found a laboratory in a local university for a reasonable price who was able to help design, define some minimum testing to reasonably ensure compliance without breaking the bank, and realize the tests. It was long and a significant budget.

A lot of his BOM are components from China. His conclusion is that you cannot trust in any way their CE marking. Some pass, others don't. They are all marked. You need to test yourself.

For low volume "artisanal" products, it looks like a lot of people just don't perform rigorous (professional) tests, they just test whatever they at home, mark it and hope they never get into trouble.

> A lot of his BOM are components from China. His conclusion is that you cannot trust in any way their CE marking. Some pass, others don't. They are all marked. You need to test yourself.

yeah, that's tricky. Testing whether an ESP32 complies with CE sounds very difficult and pricy. I suppose if they are bought in the EU though that it's up to the supplier that you buy it from to make sure it's CE compliant.

> For EMI, it's hard and expensive. Specialized companies will test and certify your device, but this is too expensive.

Yeah, I actually did find a UK-based test lab that does it more cheaply. Indeed I planned to send my device there and mainly just verify EMI for CE conformance of my device. It's good to get some confirmation that this is the main test required.

My relative's device was controlled by an Arduino so indeed that was easier to trust. The untrustworthy Chinese components were more actuators, pumps, and so on, so easier to test than an MCU.

One advice concerning EMI is that it's one thing to certify the finished design, but it's tricky to design it right if you're not experienced with EMI. Before testing it's worth having some advice and design review with the specialists to see if they think the tests will pass and how to improve it.

Two possibilities. Depending on what you are doing you might be able to purchase modules that are pre-certified and avoid certification that way. For example, if one wanted to have a product that uses WiFi normally you need FCC certification for that - but it’s also possible to purchase WiFi modules that are already FCC certified and avoid having to go through that process yourself.

The second tip if for some reason you have to certify yourself is that when it comes to bureaucratic processes that you don’t want to get wrong and that you aren’t super familiar with I’ve had a lot of luck with finding people on Upwork or Fiverr that specialize in that particular process.

Thanks for the advice! Indeed, I was planning to use pre-certified WiFi modules (ESP32 specifically). There are a couple of things here though:

A) How can I trust that my supplier actually did the right tests? Do I just trust that they follow whatever spec (that passes CE) Espressif gave them and that's enough?

B) From what I've read, if all the components you use to build an electronic device are CE approved, that doesn't mean that your device is. This seems to be boiled down to as "CE + CE != CE" a lot of the time I've seen it. Is it wrong?

Fiverr/Upwork is definitely not something I've considered. Will give these a try.

RE ".....How can I trust that my supplier actually did the right tests? Do I just trust that they follow whatever spec (that passes CE) Espressif gave them and that's enough?..."

This is a very important point. It is not unknown for products to have completely fake certifications including a fake lab report and fake certificates. What you need to do is ask for certification report and certification certificate. These documents will show the testing lab. It should be accredited to do the test. You need to contact the claimed testing lab and quote the approval number and check all the approval details match . ie the company, product is correct. Could also get a test lab of your choice to review the accreditation documents and product to verify them ( not retesting) just review to check accreditation.

For A, yes, you need to ensure the vendor is reputable and that their testing is trustworthy. Espressif, however, moves a lot of chips and I think it is safe to assume the manufacturer is reputable if they are supplying other well known ESP32 hardware - can maybe check importyeti to see who other people are using.

For B, yes, you are correct, but my understanding is that as long as you don’t have any other “intentional” emitters (for example Wifi or BLE) you only need the unintentional if your device is battery powered or uses a certified power supply. Then it should be quite cheap and simpler to get the full certification (think $500-$1000 instead of an order of magnitude more)

I'm in a somewhat similar position. I have built and programmed several prototypes. Currently I'm building a website and plan to present the working prototype of one such project as a MVP. Based on what I have been reading on various blogs, I plan to conduct several email and social media marketing blitzes.

I have only recently joined YC Startup School and hope that I will be able to make some suitable connections. If you'd like to connect, please take a look at my HN profile page.

I produce these Kokonauts http://instagram.com/kokonautinc

Started using development boards and designed the pcb around it with the sensor and sold as a mvp. Then later the entire board and now iterating on that.

Use CAD to design the board such as KiCad, Diptrace and then get the board produced at jlcpcb or oshpark. You can assemble the pcb with the components at Pcbway, Seeedstudio, Aisler, etc.

Started manually flashing the firmware and every step was done by hand. The firmware flashing and testing is now automated. Continuing working on further automation. The board house also offers to do this but it's done by hand and I thought it would be fun to work on automation plus it saves a few dollars on each board.

The software follows the usual software development lifecycle.

The enclosure initially used 3dprintuk (jlcpcb also offers 3dprinted enclosures too) to produce. Now use a Prusa Mk3s+. FDM 3d printers are getting really good so not sure when to get injection mold. Quality of injection molded part is still the best. You can get a resin 3dprint (jlcpcb offers that) then vacuum cast for a few hundred for a few pieces before injection mold.

Find someone who has done it before for your first idea, unless someone has written a book on all that (has anyone done a hardware development bootcamp?).

When you have a prototype, don't go to a trade show and display it without reviewing UC/FCC labeling guidelines. We were across the isle from a major modem company and the inspector came by, picked up a display prototype and it was unlabeled.

They got a $10 or $20K fine right there.

> I have various ideas, not especially "high tech" more assembling of off the shelf components to create novel products that I think may be useful (I have a need for these myself). So I could build a prototype myself (despite being a software person). After that what?

I'm exactly on the same boat for Butt Mover[1], A butt triggered productivity-health game.

So I'm taking a novel approach; I'm not going to build the input hardware i.e. Butt Trigger, instead going to rope-in individual builders across the world to build and sell the Butt Triggers based on my design(without any commission to me) for their region while I provide the SAAS for Butt Mover.

This in theory should address the manufacturing, shipping, support woes in selling a hardware product which even on normal times was ludicrous to do from India(Shipping fee over ocean is at least 6X that of product). But commercial success of this process depends upon several variables, Guess we'd find out soon as I'm planning an early access for the game next week.

[1] https://buttmover.com

I highly recommend you check Michael Lynch's blog. He built a hardware product and documented his progress from the very start on his newsletter: https://mtlynch.io/tinypilot.

It has been pretty fascinating for me in comparison to al the digital product stuff I'm used to.

Thanks for the kind words! I shared a little more in another comment: https://news.ycombinator.com/item?id=30586841

It is challenging. I’ve been part of a project that took a product from idea to brick and mortar consumer retail in about 18 months and it was a lot.

You’ve got to become an ad-hoc subject matter expert in many different areas, or be able to find and hire those people. This is everything from electrical, RF, design for manufacturing, mold design, packaging, logistics etc.

In my experience finding a manufacturing partner is a challenge, big ones are usually not interested in a low volume product taking up their assembly line space. They’ll usually want some contractual guarantees about the number of units and it’s going to make your eyes bulge the first time you see it.

Call a lot of people, ask a lot of questions, and get introductions from every possible person you can. A lot of manufacturing is old school business and relies on past experience and transitive trust to start a relationship.

I really loved the challenge, but this kind of thing is not something to step into lightly.

Not a personal recommendation, but I've seen this book recommended a few times: https://www.oreilly.com/library/view/prototype-to-product/97...

Learn the following skills, or hire people who have them. Details depend on what you're building.

  - Electronics
  - PCB design, with software like Altium or KiCad
  - CAD software like Solidworks or Fusion
  - Program emebedded firmware in a low level language like C or Rust
  - Learn how to read datasheets, and interact with hardware at the register level
  - How to select parts and manage BOM
  - How to choose manufacturing techniques, like injection molding, extrusion, milling etc
  - Regulatory compliance. Eg, are there safety concerns? Does it contain lead? Does it radiate?

Short answer is, you need to figure out what your project requirement are, and learn the relevant skills.

Certification costs eg. FCC and licensing fees eg. Bluetooth will make anything small quantity not viable. Unlike software where you can have a very niche piece of software for hardware you need to sell a lot to cover the initial costs.

If you're using a pre-certified BT module, the costs can be dramatically lower than you might expect. Ditto doing your lab testing "over there" rather than in the US.

If it's not an intentional radiator, you can get multiple rounds of FCC+CE testing for USD650 in Shenzhen these days.

Even with using a pre-certified BT module there is a $9600 deceleration fee to SIG.

I had a hardware company a few years back. It was a disaster even though we sold it to companies like CNN, Princeton University, Stanford, UCSD, UC Berkeley. If you want to do it as a hobby business, go ahead. But it's incredible how quickly your hardware supplies will change and impact your offering and you have to consider returns and the margins you will make.

I would recommend offering a software package and let the users buy their own hardware if I did it again. That's still a difficult business but at least you don't have to carry physical inventory.

Ask me more if this is relevant to you and you have questions.

Most of the comments touched on what you need already, but one thing I will add:

I don't know if I can advertise other startup accelerators here, but there is at least one startup accelerator that involves companies with manufactured products. I participated in this program, and they work closely with participating companies on this process. The accelerator is also very well connected, which could help with your marketing thing too.

Same boat, with differences:

Have a product, have a device, have 20 of them garage-made.

Here is what we were doing 3 years ago with them - they have evolved since, but this gives a good idea (we were rejected for YC with this submission, got into startup school but need help bringing this to market.

We have a lot more sensor capability for these and we have a customer that wants to license our medical version, and I am struggling how to write up a partnership agreement.

https://youtu.be/FXkNY0vTATA part one, employee presence

https://youtu.be/c7qvfm6vhF0 part two, product disposition

The premise of the above videos:

We came from tracking weapons/chems for DoD among others whilst at lockheed.

This is an expression of "Slot Level Visibility" in an automated fashion...

We have iterated over the years from Lockheed based on how we saw the deficiencies of what RFID/Slot tracking..

So we have a spatial sensor that tracks:

Temp

Humidity

Particulate (Co2, and more)

Altitude

Wifi/mesh

Presence/ location of people (we can map them out, the partner company wants to license our base to add their Lidar on top (even though we can do lidar easily)

A badge that allows for awareness of people in the space...

---

We are talking to a national paintball org such that we can provide arena awareness of everyone on the field. (We may eventually actually track shots)

This is separate from our medical uses (tracking health of independently-living elderly in their homes) (did they open their medecine cabinet, have they fallen, how much movement activity did they do today, whats the air quality in the home)

Integration with voice "Alexa, did grandpa take his medications today"

We have working devices, patents and pedigree... and yet getting HW funds has been a wall for us.

We are a team of two, with support from a small group of contributing engineers from our network.

The Bolt VC blog has a ton of good articles on this, e.g. https://blog.bolt.io/casper-glow/. Check out the 'Teardown' category for the best ones.

I'm currently facing similar challenges/goals. I've found the book by Alan Cohen to be very helpful, even to increase the coverage of what I know I don't know.

> Prototype to Product: A Practical Guide for Getting to Market

Highly recommend it

Talk with Jabil. Their suite of services can help you with all aspects of the design, build and fulfillment aspects. You bring the idea, funds and go to market strategies.

Haha definitely do not do this, unless you have a million dollars to burn.

Seconded. I also strongly believe that if your hardware product is your core competency, it is designed by your team primarily, of course a consultant here or there can always do a good job spicing things up. I personally feel like turnkey product designs services are best suited for companies that have a specific accessory or something that needs to be designed, not the core product.

Thirded. But they probably won't talk to you anyway.

You might find Prototype to Product: A Practical Guide for Getting to Market by Alan Cohen useful.

My (naive) idea -

1) build any kind of prototype 2) Go to Shenzhen, China and find a manufacturer / partner.

3) compete with your own product on alibaba

Never trust someone that you don't have a good working relationship with to keep your IP safe.

There was a selfie stick phone case Kickstarter where you could purchase Chinese clones on Amazon before the original project finished funding.

There's a classic process here:

1) successfully crowdfund a hardware concept

2) Wait for the productionized version of that concept to show up on AliExpress for 10% of the unit cost if you manufactured yourself

3) Buy a few pallets off AliExpress, ship them to your backers.

Like a new version of Cunningham's Law:

1) Design it wrong 2) Wait for knock-offs with improvements 3) Buy those

That's a pretty common pattern. To a first approximation, you will never outcompete clones on cost or distribution. Chinese patents and trademarks are actually helpful if you need to enforce, as are good contracts written by a Chinese lawyer in Chinese.[1]

You can outcompete them on customer service, community, brand, and quality.

For most consumerish products, I would not be worried about a clone stealing your design files. It is incredibly cheap to clone the mechanical parts of a product from photos or a single copy bought retail. For most consumer electronics, the clones would be unlikely to use the expensive western parts and designs you would. Instead, you're likely to see really fascinating cost engineering around parts you didn't know existed.

[1] I have sued a supplier in China, who did some sketchy stuff. It was a fascinating process. Contrary to the conventional wisdom and to our expectations, the court actually helped us get back our tooling and inventory.

how do you keep people from stealing your doohickey?

I'm going through this right now. As a full stack developer, I found Hardware Academy (paid) to be good, and they have guides and forums you can ask questions. [1]

Over the last six months, I have built about five prototypes ranging from 12x10 inches to about 4x4 inches in size. Right now, I've it all fitting in a 2x4 PCB and 2x4x1 3D printed enclosure for customer evaluation. I didn't build the last piece, someone with expertise in hardware built it for me.

I don't have much time (hardware project ;) ), so I will jot down some of my observations and learnings in no particular order.

I'm building an IoT product (ARM Cortex-M4 microcontroller, a few sensors, cellular connection, and GPS). I'm US-based, so some of my comments may not apply to you. I'm also a software developer transitioning to hardware, so some of these may be too basic if you are experienced with building electronics

1. Understand why you are doing this? Is this a must-have or a nice-to-have product? In my case, it looks like a nice to have, so chances of success are low. I'm doing this because I want to build something and learn new skills (marketing, customer development, hardware). I'm saying this because there are times when I feel like I'm digging myself further into a hole as the investment is high and the chances of success are low.

2. Find someone you trust with all the equipment and skills to build it. The best way is to ask them to make small things first and build on the relationship. I've had success on upwork, don't hire based on hourly rate. I'm self funded so money is tight and I initially hired based on rate, which was not great. That said I've found good people at both medium-low and high rates. Sweet spot is around $60 - $100.

3. It's going to take a lot of time and money. For reference, I reached out to two turnkey companies, one gave me a quote for $125,000, and another one wanted about $7,000 to do a prestudy before giving me a quote. I decided to go at it myself and spend about $18,000 and counting. At the end of this $ 18,000, I'll have 12 customer test units (only 12 due to supply chain issues)

4. Understand certification requirements for your target market

5. Use pre-certified modules, which can reduce certification costs. But more importantly, you can do a sales test with small quantities without FCC certification, so using a pre-certified module and classifying my product as an unintentional radiator allows me to sleep better at night.

6. Incorporate firmware updates over the air early so you can fix issues when it is shipped

7. Talk to everyone about this to learn about challenges, especially adoption. In my case, I learned about people's fears that I must overcome and also new market segments.

8. Beware of people who are excited about your product. In my case, I can find people who love it and think it's a fantastic product with lots of potential. They mean well, but they might only be looking at it through a narrow lens. Also, beware of people who are against it. Understand where everyone comes from and make your decisions accordingly.

9. Read through the datasheets of every component and make sure they all work well, understand the software and hardware thoroughly (this sounds obvious, but I've missed so many things the first few times about hardware)

10. Debugging hardware is complicated. Instead of looking at a screen and running logic through your mind, you'll be looking at lots of wires and solder joints when running the logic. It gets harder as you shrink in size.

11. Learn to use kicad and LT Spice

12. This is obvious, but get a multimeter and test for shorts after each soldering session.

13. Always be consistent with wire colors

14. Spend time learning electronics and magnetism understanding RF interference in circuits. I used to think it would be easy to create a circuit board with multiple antennas, but I was very, very wrong.

If you have any questions, need referral or like to chat, please feel free to contact me, email in profile

[1] https://thehardwareacademy.com/

Edit: added info about upwork

We're going through this now as software engineers who are building EEG devices for sleep (https://soundmind.co)

We've taken as much software engineering methods as we can and applied it to hardware. So where many people are focusing on how you get your product manufactured, I think that is jumping ahead at this point.

Our process, and what we'd recommend to somebody else in our shoes is.

1) build your prototype with off the shelf components. Make it rough, make something that works, doesn't have to be polished and ready for production.

2) Show that product to your potential users, get them to use it. (hopefully you've been talking to users before this point, but it's nice to have something that is not complete vapourware). Learn from these prototypes, work on your software. Keep building hardware and iterating, and keep learning. Learn from them what they'd pay for the product, how much is it worth to them.

3) Do a back of the envelope quick math and see if you think you can make the product for less than they would pay for it. If not, maybe find another way to solve the problem. You don't need to be solid on your numbers, but you need to be able to ballpark and say "I think I can get this mass produced for $150, and can sell it for $180...hhhhmmm is that enough profit to make this a viable business??" You should be able to get help with this from some contacts who know about business if you're still uncomfortable, or show it to somebody who has done hardware and get their input.

4) When you think you've got things to a point where you are ready to get something to go to market, figure out if you can build it yourself (not mass manufacturing, can you build 1 or 5 or 15). If you don't have a vision for the aesthetic of what your product should look like, you can get industrial designers involved (we did), but from the vision point, see if you can build something yourself.

5) For hardware components, use as much as you can off the shelf. We're using esp32 which gave us the processing power we need, with ble AND a pre-certed antenna. We've designed our hardware such that our charging and power management is separate from our main PCB. This way we can get our power supply certed and (in theory) make changes to the main board with limited re-cert requirements. For batteries, see if you can use an existing battery pack. Camera batteries are available with protection circuit and plastic housing, which can save you on the cost of getting your own plastic molds for that component.

6) For us it was important to keep removing plastics and other molded components from our design. Mostly because we don't want to be creating more plastic waste, but also because molds can be expensive. See what other materials you can make.

7) At this point, you've got the BOM for hardware, you've got your enclosure or whatever you're packaging is. You should be able to get a bill of materials for these items. Getting PCBs built up is not that expensive, you can go to many sites and get quotes. Speak to local manufacturers about what is needed for your enclosures and get those priced out. Don't forget packaging, shipping, insurance, etc. I'm not sure of your product, so not sure what else you'd need, but at this point, you know enough that you're getting pretty close to knowing what your costs are.

7) Re-do your forecasting and now deep dive into your costs further. Maybe you can continue to strip down your costs (we've taken our BOM (bill of materials) down by almost 20% on our latest iteration. You've likely spent a lot of time up to now, but you're hopefully getting close, and you hopefully haven't spent much $$ but lots of time in refining the business and hopefully the product.

8) If all systems look good, and you still think this is a valuable product to bring to market, now you can begin talking to manufacturers. A manufacturing engineer might be able to help you to further refine the design to make it easier to manufacture. Local manufacturers likely have designers they like to work with who can help get everything ready to manufacture.

At this point, you can start following all of the advice around minimum order quantities, managing manufacturing, shipping, certs, etc etc.

As a software engineer, the one thing I'll tell you that I think we've learned is that software engineers work much faster than hardware, and people will say "hardware is hard", or "industrial design is slow". I thin all of that is BS. We contract with an EE who works at our speed. We do hardware iterations very quickly. It only takes a few days to get boards built up and sent from China (seems to really be the only place making up boards) and with early prototypes we were building (adding components, resistors, caps, etc) ourselves.

Getting a local manufacturer to build up your boards doesn't take long (from our experience) you just need to manage the pipeline as they need more lead time. But if you're managing the things you need to do, and keep them in the loop, you'll be organizing product, components, etc etc. So far, I've found the logistics management to be a big part of the challenge.

The biggest disappointment for us was industrial design. Our designers gave us an excellent direction in product vision, which we didn't have initially, but they were far too slow to work with. We were advised to bring industrial design in house and do it ourselves. I thought that was crazy advice, but I started doing it. At first really rough prototypes, and just kept refining ,refining, refining. I'd say I got 10x further than our ID group did in about 1/5th the time. The guys we hired are great at what they do, but the industry doesn't move at software engineering time scales. We're not slowing down waiting for people, we need these slow moving pieces to move up to our speed.

Having said that, we're in neurotech, so our software dev cycles are slower than your average software development, but still....

Lots of info there, and we're still in this journey ourselves. Hope that helps.