So you want to build a hardware company,

and you’re a software engineer…


A major part of my career has revolved around being an information beacon. I have found great success through learning and then regurgitating back to the community that I have grown with.

Since starting work on Jewliebots a little over a year ago, I have learned an incredible amount about hardware, engineering, and manufacturing. None of this information is readily available online. There is no exhaustive how-to. I have gotten much of it by speaking to people who have done it before, learning from my mistakes, and from the incredible staff and mentors at Highway1.

To continue the spirit of disseminating my knowledge and learnings to the larger community, I’m going to write a series about how to navigate the world of building a hardware product when you are a software developer.

The series is mapped to rollout as such:

Part 1: Prototyping

Part 2: Debugging

Part 3: User Testing

Part 4: Proving Traction

Part 5: Design for Manufacturing

Part 6: Manufacturing

Part 1: Prototyping

For a hardware company, the prototyping phase is just as important as it is at a software company. It is done for similar reasons: proof of concept, user testing, iterating.

The big difference, and this applies to just about everything in a hardware company, is not the prototyping phase; it’s the fact that that software is pretend. I don’t say that to diminish all the amazing things we’ve done with software, but the complexity lies in the infinite variations on the arrangement of 1's and 0's. Hardware is real life and requires real life objects to exist.

The prototyping phase has significantly more complexity as you are often working with things that are very small components of a larger whole. There are a few approaches to prototyping, I’m going to list what I have seen work below.

There is a way to split up your prototyping to make it easier to manage in the beginning. You can start with two different prototypes that you eventually merge into one when you are doing design for manufacturing.

“Works-like” prototypes are ugly, and they are rarely something you are proud to show off. I’ve seen super polished ones, and I’ve seen prototypes held together by tape and construction paper. The purpose of the “works-like” prototype is to prove your concept, “this is something that can exist”.

A “looks-like” prototype is made to prove design. The priority here is to get them in front of your users and get their opinions on what they love and what could be better,

The Works-Like

Augmenting things that already exist

If you are an experienced hardware hacker, or even sometimes when you aren’t, it’s easiest to take things that already exist and reprogram them. I often encourage people that are new at hardware to begin by taking things apart just to find out what goes on inside of common electronics.

If what you are making is similar to something that already exists you can learn a lot by taking that thing apart, adding what you’d like and putting it back together.

this is a “bass fuzz” which is an electric guitar synth mod

It’s also a great way to learn all about your competition. We learned a ton from a deconstructed Fitbit we kept on our desk at Highway1. You can learn before taking things apart by searching for “teardown” + the name of the item. Examples:

deconstructed game controllers

deconstructed Nike fuelband

deconstructed beats by dre

Littlebits

A Littlebits prototype

At Highway1 we had a wall of Littlebits full of all the components you could ask for. They are simple and handy for rapid prototypes and fast feedback. They have everything, wifi, bluetooth, servos, buttons…etc. They even have a platform where you can create and vote for your favorite bits to be created called the BitLab.

There have been a lot of cool projects made with Littlebits. They snap together and are the quickest way to do rapid prototyping. At $249 for their home automation kit it isn’t the cheapest way to build a prototype, but it’s certainly the most convenient.

Arduino

This is certainly the most popular way to build electronic prototypes. In fact, the invention of the Arduino is definitely what has lead to the hardware revolution that we are seeing today.

An early Jewliebots prototype made with the RFDuino

The Femtoduino, RFDuino, and Sparkcore are my favorite microprocessor modules to start with.

The Femtoduino is great if you don’t need connectivity for your prototype. It’s super small, and really easy to work with. I prefer it over the Arduino Uno as it is much much smaller, however, the Uno can still fit your needs if size isn’t a concern. They have also recently come out with a Bluetooth module that’s super tiny.

The RFDuino is what we ended up using for our “works like” Jewliebots prototype. It has BLE capability and is fairly easy to work with.

The Sparkcore is an awesome Wifi module and their support is second to none. I love this team and I know that this is only the beginning for them.

You can find all the LEDs, batteries, servos, resistors, wires, and such at Sparkfun or Adafruit for super reasonable prices.

The Looks-Like

Making a looks like prototype is a lot of fun and there are many resources out there to help you. The secret to a good “looks-like” prototype is to be constantly iterating and incorporating user feedback.

There are a ton of different mediums and many resources, I will try to go through everything I have encountered.

Mediums

Product design is hard, if you haven’t yet watch this video of how IDEO approaches product design. It’s super inspiring, there is also a great book called The Design of Everyday Things that can help you if you don’t have a background in design.

In the beginning things like polymer clay, foam, and moldable plastic are great ways to quickly crank out your ideas. From there you may want to graduate to 3d printing, improving the resolution the closer you get to your final product. Ultimately you can have a factory do a few manufacturing ready prototypes for a relatively low cost when you are confident in the design.

a Jewliebots looks like prototype

Resources

When you don’t have a background in product design you will be looking many places for help. I’ve had great experience searching Etsy for folks whose design aesthetic I like and hiring them for smallish design jobs.

This is a 3d printed early Jewliebots prototype. I was able to get the design done by someone I met on Etsy, and a 3d print shop here in NYC called CuboNYC.

For later prototypes I found high-res 3d printing shops on 3dhubs.com. You can choose your resolution and the color of the print. I found it to be super easy to use.

For more refined prototypes we relied on a factory called Solid Concepts located in San Jose. I’ve also heard great things about Fathom that is also in that area as well as a place called Strong D that we visited in Shenzhen when we went there to visit. These places will use materials like brushed aluminum, polyurethane, and all different kinds of plastics polymers to get the look and feel you need. They will do super low volume production runs.

All these places will want CAD designs from you in order to manufacture your prototypes. If you aren’t super great at autoCAD I’ve used oDesk as a great resource to find help with 3d design. I don’t go for the folks with crazy low hourly rates, as their work can be sloppy, and it’s important to pay people what they are worth for a good job. If you establish a relationship with someone good you cn use them repeatedly.

Okay! That was my brain dump on prototyping. Stay tuned for Part 2: Debugging.

Show your support

Clapping shows how much you appreciated Sara Chipps’s story.