8 Critical Checks Before Turning On Your Prototype

Last week, Upverter CEO Zak participated in Hardware Workshop Toronto, where he led a presentation on manufacturing prototypes to local startups and entrepreneurs. Even for people who deal with hardware, we often find that people don’t have a clue, or know where to start when it comes to getting their ideas mass/manufactured. So we put together a slide deck to shine some light on the topic, covering everything from the hardware lifecycle to how to find a manufacturer in China.

But it wouldn’t be fair to keep the goods for just the people who attended the event, would it? Here’s a link to the deck, as well as the checklists we made as handy resources. Feel free to download and share!

The complete slide deck

Hardware life cycle

Download the hardware lifecycle here.

Turn-on checklist

Download the Turn-on checklist here.

6 Things they don’t teach you in school about being a hardware engineer

6 Things they don't teach you in school about being a hardware engineer

1. School gets you half way there. The rest is self-driven.

There’s really no school for hardware design. I’ve gotten way more out of self-directed learning than sitting in a classroom. Make sure to learn from mentors, regularly read articles written by industry experts, attend seminars, read white papers from big semi companies, and participate in forums. It’s good to look beyond what you think you know. Don’t just assume the way you’ve been doing it is the best way. The degree of collaboration and knowledge-sharing in hardware pales in comparison to the software world.

2. Mistakes are valuable.

Even if it takes you 7 days to figure out, causes hair loss, and many sleepless nights, you’ll come out of it way more equipped with hardware knowledge than before.

3. Every problem has a solution. Every single one.

That doesn’t mean you’ll like the solution. But it exists. And if you keep plugging away methodically, trying things, experimenting, trusting your intuition, seeking help when you can, a clue will appear that will ultimately be the turning point in figuring out a problem. Remembering this has helped me get through some despondent times when I was stuck on deep FPGA timing problems, power supply start up issues, and signal integrity mysteries.

4. Power supply problems can be like the chicken-or-the-egg conundrum.

Fixing a power supply problem can be extremely tricky. While debugging, you can be damaging your board and changing the very thing that you’re debugging. And just because a section of the board powered up and started to work, doesn’t mean it will continue to work.

Try to choose power supply devices that are debug-friendly (i.e. a digital controller that you can read status from). Be methodical and power up each supply one by one. Build in debugging test-points for all important circuit nodes and pins on controller chips. Power-up LEDs are super helpful. This way, fixing the problem won’t be like untangling one long piece of string.

5. Never underestimate the value of good hand-soldering skills.

Just like it doesn’t make sense to go to fashion school and not learn how to hand-sew, not knowing how to hand-solder will come with many limitations. You’re going to come across a number of instances where you’ll have to rework your board: parts have to be removed and replaced, passives have to be changed, jumpers have to be added, etc. Knowing how to solder by hand will open up a number of new debugging channels for you, allowing you to pinpoint problems more effectively.

6. There’s great knowledge value in imparting what you know to others.

Nurture a co-op student or intern. Teach them everything you know. You’ll be surprised at how hard it is to support and explain what you’ve always understood to be true. Not only will this be valuable for the listener, but verbally walking through things will deepen your own comprehension on the topic.

Infographic: Choosing the Right Development Board

The hardware revolution is picking up steam. Oculus Rift, Nest, Basis, Pebble. Big new companies selling hundreds of thousands of units, and exiting for billions of dollars. So naturally, you want to start a hardware startup of your own. Good! You’ve picked the perfect time.

But before you get to a real life product, you’re going to have to get your proof of concept working. While development boards are the easiest, most affordable way to test and prototype your idea, we know that it can be overwhelming to choose the right one out of so many options. That’s why we’ve put together this handy infographic to help you find the development board that will jumpstart your project.

Development Board Infographic

How to be a better EE #1: Design Review


At Upverter, we love talking to hardware engineers and learning about how their teams review schematics. It seems that everyone has put together their own unique method:
from buying large-format printers and printing everything out on 11″x17″ paper, to long email threads in Outlook; running WebEx conferences to making PDF bookmarks.

Despite the variations in workflow, most of them agree that 2 of the most important factors in finding errors are:

  • How thorough you are
  • How experienced you are

We know everyone is crunched for time and usually isn’t as meticulous as they’d like to be. We also know that gaining experience often comes from mistakes: making them yourself or learning from the ones your colleagues have made.

If you’re new to reviewing schematic (or have simply been neglecting it), here are some general guidelines to help you become an effective reviewer.

    1. Read the datasheet of each component! It sounds so obvious, right? You’d be surprised at how many reviewers skip this. This is one of the most valuable things you can do. Read the datasheet, understand how the part functions, how each pin should be connected, and then check the schematic. With practice you’ll become an expert at reading datasheets quickly and figuring out which parts are important for schematic review. It’s not as daunting (or time-consuming) as you might think.
    2. Collect all the supplemental documentation on the vendor’s site. Look for reference designs, evaluation boards, application notes, layout guidelines, etc. Then quickly compare your schematic against each of these documents. Any differences might signal a mistake. This is a fast way to check connectivity with confidence.
    3. Leverage the vendor’s field application engineers. FAEs are often very knowledgeable on how their vendor’s components can be used in a design. Send them your schematic and they’ll be happy to review it for you. They often have their own internal checklist for this purpose.
    4. Create a culture of asking questions. Why is this connected this way? Why did you choose this part? How do you plan on dealing with X? When someone verbally walks through something, it often triggers important things that they hadn’t considered before. You’ve probably experienced this phenomenon yourself. There’s a ton of value in not only knowing the answer, but in this associative way of thinking.
    5. Share “lessons learned” with each other. Create a communal list of errors you’ve found on previous boards. Review this list before looking at a schematic (or designing one from scratch!) to remind yourself of things to watch out for. Have your senior engineers get this list started.
    6. Review the schematic in small chunks as it’s being designed. Many engineers don’t have the time to be thorough when faced with a giant schematic. It can be a daunting and a stressful process. Instead, consider sending out a small piece of the schematic every few days during the design process. This way reviewers are only faced with one circuit or component. It’s not only more manageable, but also an enjoyable way for engineers to familiarize themselves with the overall design as it comes together. They’ll be much more prepared for the big review at the end.

While some of these guidelines might seem obvious and basic, it’s important to build on a good foundation of reviewing schematics. There are simply no shortcuts. Once you incorporate these steps into your work and process, it will be loads easier (and faster) to prevent errors from actually becoming a problem.

Stay tuned for a checklist of specific circuit errors to watch out for.

5 Steps to Smooth Price Quoting With a Contract Manufacturer


You have an idea. You make a prototype at home on your bench. Crowdfunding is an astonishing success. Now you need thousands of identical widgets. The scale is beyond your home bench and it is time to manufacture on a larger scale. What now?

In the world of contract manufacturing, we rely on automation. Robots, glorious robots, perform most of our tasks. This is accomplished both with physical machines transferring, placing, and soldering parts, as well as with scripts processing data sets. The goal of automation is to reduce the opportunities for error. This begins with the quoting process, which it is important to understand.

For a board containing surface-mount and through-hole parts, the following items are needed to produce an assembly:

  1. Gerber file
  2. PCB fab file
  3. XY placement file
  4. Bill of material
  5. Requirements (sometimes on an assembly drawing)

Since all of these items are needed to manufacture, try to provide this information up front at the time of quoting. Missing information will result in delay, production errors, or a change in cost. Due to the nature of development, it is understandable that you might not have all of these items available. If so provide as much information as you can.

The world of contract manufacturing currently has no accepted standard for how these files are packaged; an unfortunate state. However, while everyone does it different, in the end, we all need the same things. If possible, wrap it up in a tidy zip file and forward it via email or through your favorite cloud storage solution.

Now you have the list, let’s dive into more detail and look at how to avoid errors:

Gerber File

This artwork is exported out of your CAM/CAD software and should be in RS274x format. All current popular packages have this export feature.

It’s important to realise that the contract manufacturer should not work directly with the raw design files. There are many options when data is exported and it is ultimately up to the customer to provide a universally accepted data set with all the decisions about options made.

Please provide the Gerber file as a 1-up, not panelized. Your contract manufacturer and PCB Fab house will determine the best way to ensure compatibility with their manufacturing setup.

PCB Fab File

Printed circuit boards can be manufactured with a variety of options, and it is important to understand some basics so that you can both support your design, and avoid unnecessary cost. A PCB fab file basically states the fabrication requirements in plain ol’ black and white for easy understanding. There are many websites that provide detail about this subject. For the moment, we will keep this to a quick and simple crash course. This information should be documented within a simple PDF file. At a minimum, it should contain the following:

  • Finished PCB Thickness: A normal, run of the mill circuit board is usually .062 inches thick. High reliability or rugged designs are typically .093 inches thick. Really thin consumer products usually run .031 inches thick. If you are unsure of what you need, stick to the .062 as this will be fine for most applications.
  • Finished Copper Thickness: Specify the final thickness of copper for both the inner layers and outer layers. This is typically stated as “1 oz finished copper”. High reliability, high power products may have a greater copper thickness, such as 2 oz finished copper. Consumer products are usually ½ oz finished copper. A printed circuit board fab house will sometimes start with a ½ oz copper sheet, and plate it up to 1 oz, hence the “finished copper thickness”. If you don’t know, just stick with a blanket statement of “1 oz finished copper for all layers” and you should be fine.
  • Layer Stackup: You can name files whatever you want to. In the end, though, a designer should label the order of the files being stacked up. By providing this information, you will eliminate an opportunity for error. This is especially important if your design requires controlled impedance or differential pairs.smooth-price-quoting-layers
  • Color: Specify the PCB mask color and silkscreen color. Typical colors are green, black, white, yellow, red, and blue. Any color is possible, but you will likely incur additional charges and are less likely to get the exact color you desire. Stick with the basics for greatest success. Note that silk screen colors should be complementary: it is not a good idea to specify black ink on a black board. Typical silkscreen color is white or black.

    A few notes are worth making about white PCBs. If you desire a white finish, the surface treatment is very important! Do not specify immersion gold with white as the gold color will bleed through the masking and your board will end up with an undesired pinkish color. Stick with immersion silver or immersion tin. Also, reflow temperatures for RoHS (lead free) can sometimes discolor white finishes, so it should be specified as “high-temperature white masking” or “double-painted white masking” for the greatest chance of success.

  • Drill File: Some CAD/CAM software exports the drill file in a separate folder. Make sure you include both the drill file and drill chart in the final data.
  • Surface Treatment: All exposed copper where your parts ultimately end up must have a surface treatment as exposed copper all by itself will corrode. If you are unsure of what to specify, stick with immersion silver for good performance and lower cost. The available options are usually:
    • LF-HASL – Hot Air Solder Leveling. “LF” stands for Lead Free. Get used to going lead free! LF-HASL is the lowest cost surface treatment, and should only be used for designs that are completely through-hole or surface-mount assemblies which do not have any fine pitch parts. LF-HASL will cause erroneous shorts on fine pitch designs.
    • Immersion Tin – This is the lowest cost solution for high volume and fine pitch designs. The electrical performance of immersion tin is not as good as that of silver. Also, if not used immediately, immersion tin will corrode over time. If you intend to have your boards sitting around waiting for customer orders, do not use this finish. It should only be used if you immediately run your production order upon arrival.
    • Immersion Silver – Slightly lower cost and performance than immersion gold, typically anything using this finish will perform just fine. Silver suffers from the same corrosion as described above under immersion tin.
    • Immersion Gold aka ENIG – this is the gold standard. It has the best electrical performance, looks great, and will not corrode. It is also the most expensive finish. Use this for any assemblies that require lots of fine pitch, ball grid arrays, or other advanced features.
    • * Special Features – Controlled impedance or differential pairs should be specifically identified! If your design contains high-speed flash memory and/or USB connected directly to a microcontroller, you probably need this. Controlled impedance is characterized by a trace width and trace separation note. A printed circuit board fab house will typically ask for minor alterations to these widths to hold your desired impedance. Make a note stating it is okay to alter widths to hold the desired impedance.

XY Placement File

An XY file is a text file that locates all of your surface mount parts. This file is translated into machine code for pick-and-place machines to perform automated surface mount placement.

Remember those robots mentioned in the beginning? These are the instructions that the robots use. As with a Gerber, an XY file can be exported out of your CAD/CAM design suite. Without this file, a programmer would need to put your board on the placement machine and manually enter every position. You don’t want that to happen, since generating the XY yourself costs nothing. Some quoting formulas use this file to automate analysis and calculate labor times. It’s a nice thing to have up front.

Bill Of Material

It is beyond expression how valuable a correct bill of material (BOM) from the customer is to a contract manufacturer. Correctly formatted, this data will lower your costs, cut down how long it takes to get your quote, avoid additional charges later, and ensure correct production the first time.

However your CAD/CAM software exports data, it is likely incorrect. Here what you want it to look like: this is your master template forevermore. Free. Here. Now. The best format for a bill of material will look like this:


Oh, one more thing: .XLS format is a must. For all of you Open Source types, I respect your decisions. Just know that in the contract manufacturing world this file format is king. As you are able to export into the .XLS format, save everyone time by providing it up front. Little tangent here: All machines used to make Apple products run on Windows. All of them.

You are of course welcome to add additional information. This is the bare minimum that is needed. Absolutely nothing less. Spend time on your bill of material. Google part numbers, make sure they are easily found. If you are using something not readily available out of a standard vendor or catalog company, include the contact people needed to buy. Make sure the quantity listed actually matches the number of reference designators.

Feel free to refer to the bill of material examples sheet we uploaded to see many examples of problem BOMs and the opportunities for error that are introduced.

Remember the beginning, when we talked about robots and automation? When quoting your project, a BOM is sent to various vendors for pricing. They use scripts to automate matching the listed part numbers to their internal data, which then gives a price per line item. If a number cannot be matched through automation, it is done by a real live human being, which takes more time. Most customers demand that a quote be finished yesterday, so don’t shoot yourself in the foot on this one.

You may have designed a product exclusively using products from a particular popular vendor, and have that vendor’s internal part number on the BOM. This is a bad practice! The same part can be sourced from my set of vendors at a lower price. Take advantage of my lower price by listing the actual manufacturer and actual real-world part number. Consider this: you are Vendor A and need to price a bill of material. All of the part numbers are specific to Vendor B only. This requires someone to Google each part number on Vendor B’s site, and translate it to a real-world part number. Not only does this eat up time, but there is an opportunity for someone other than you to make a parts decision. Keep control over the design by providing real-world part numbers that do not require translation! Historic note – Part numbers ending with “-ND” do not have anything to do with Nu Disco. It used to mean “No Discount”.

Find numbers are useful for everyone. When issues come up, it is very expedient to refer to bill of material items by the find number. Think about it: “I would like to discuss Find #6” versus “Hey, the transistor on the ladder filter”. This helps expedite questions, and thus helps you get a faster and more accurate quote.

Do not put assembly notes on the bill of material (things like, “Place U2 upside down” or “Remove J1 Pin 4”). These should be communicated through a vehicle such as an assembly drawing or other PDF specification. Most companies make use of a database system called MRP (Material Resource Planning) or ERP (Enterprise Resource Planning) software. This is a big fancy database where every relevant number is stored. BOMs are often uploaded and converted to a format that is specific to a particular MRP system. Thus, it is possible that critical information can get stripped out.


Now we are in miscellaneous item territory. Remember contract manufacturers are not a one-person show. There are departments, such as quoting, engineering, manufacturing, and materials. All of these departments talk to each other but sometimes details can be missed. Given this, it’s important that you make certain that your requirements are easily available to everyone that needs to know them.

Assembly drawings are a good vehicle for all the miscellaneous requirements. If you can provide an assembly drawing, please provide, at a minimum, top and bottom side rendering with a full set of reference designators. Through-hole parts don’t always get mounted on the board’s top-side. If any through-hole parts are facing down, it is a good idea to note this. Special features such as manual wire modifications, heat sink assembly and so on should be noted on a drawing.

Examples of common requirements are lead free status, first article, pre-programming of ICs, packaging, shipping address information, etc. A simple PDF document is usually sufficient.


Can you get by with less? Yes, it happens all the time. About as often as a purchase order is cut for rework. Understanding the process from a contract manufacturer’s perspective will certainly help you get manufacturing completed faster, error free, and on budget.

Author BIO:

Jason Duerr is director of engineering for Aimtron Corporation, a contract manufacturer in Chicago.

The Hardware Product Canvas

Hardware Product Canvas

This is the Hardware Product Canvas. It’s a tool designed to help you get through the earliest stages of a hardware design. It’s open-source, creative commons, and totally hackable. We encourage you to take it, hack it and use it for your next hardware project or at your next hardware event.

How It Works



All hardware is made up of building blocks. Engineers call these blocks electrical-mechanical components, or simply parts. Almost every part in existence is either a sensor, a communicator, a storage device, a user interface, or a processor. And almost every hardware device is made up of just a handful of blocks. For example, a fitbit is just a motion sensor, flash memory storage, and a microcontroller processor (for more examples see further down).

How To Use It?

To use the canvas you just need to fill in the blanks. What kind of forces does your product need to sense? What kind of communication do you need? Do you need to store anything on the device? How much data do you need to process, and what kind of processor do you need? Do you need a user interface? LEDs or LCD or something else?

Building Block: Sensors

Sensors are components that measure forces in the physical world. They are good for measuring or detecting light (optical light sensor), or movement (accelerometer), or sound (microphone). Sensors are very popular recently in internet-of-things, and quantified-self devices for their ability to take previously offline data and bring it online. Almost all devices have at least one sensor, and many devices are little more than sensors hooked up to processing.

Some common sensor parts:

Building Block: Communication

Communication are components that send and receive signals between devices. They are good for getting data onto and off of your product. Bluetooth (in your phone) and Infra-Red (most remote controls) are two of the more popular ways of communicating between devices. As devices get smaller and more distributed, and with trends like machine-to-machine computing, communication is getting more and more important. Some type of communication is essential for almost all devices.

Some common communication parts:

Building Block: Storage

Storage are components that hold data. They are good for remembering the things your device senses or the things that are communicated to it. The most common storage these days is small flash memory ICs connected directly to the PCB in a device. Most devices, like your alarm clock, get by without any storage at all (which is why they “forget” things when they lose power).

Some common storage parts:

Building Block: User Interface & Experience

User Interfaces are components that allow the user to enter information, or components that display information back to the user. They are great for turning things on and off (switch), changing settings (push button), displaying what mode a device is in (LCD), showing power status (LED), etc. Almost all devices have at least a power-on light, if not a full LCD and button user interface.

Some common user interface parts:

Building Block: Processing

Processing are the central components in most devices. They are the brain of your product. They get data from sensors, send and receive through the communication components, write data to the storage components, receive user input and display back user information. There are many, many different processors but there are only a couple of styles, microcontrollers and FPGAs being the most common. Every product needs a processor, and almost every other component in your product will connect back to it.

Some common processing parts:

Building Block: What About Everything Else?

There are some components, like those that connect to mechanical elements, that don’t fit perfectly into the above 5 roles. Motors, relays and servos are good examples of this – they are similar to user interfaces, but they often manipulate the physical world because of a decision the processor has made. These kind of elements don’t have a great place in the canvas (yet!) so for now we recommend putting them in the User Interface or Storage sections.

Some common other parts:


At Upverter we help out with a ton of hardware hackathons, and one of the biggest problems we’ve run into at every event we’ve thrown or helped with is the very first one: idea generation. In the ideation phase of a new hardware product you’re trying to answer questions like: “what are we going to build?”, “which chips are we going to use?”, “which accelerometer does everyone else use?”.

Upverter and each of our partners has gone through a couple of different hacks on the way to the current version. Like most hackathons, we started with whiteboards, but they very quickly became too unstructured and our projects got too vast for the scope of an event. We tried moving to forums, and adding a bit more offline collaboration, but that fell apart pretty quickly too – its just way too hard to describe hardware that doesn’t exist yet in a paragraph of text. After one of our last events I was grabbing a coffee with a couple of the hackers and talking through the problems that they had during the event, and ideation came up again and again as a problem that needed to be solved. After a ton of brainstorming, and a bit of inspiration from the Business Model Canvas, we realized that we were over complicating things.

All hardware is made up of parts, and those parts fit into a very small group of roles. There are sensors, communication, storage, user interfaces, and processing. We just need to make it easier for people to talk about what’s in their product. We just needed to break it up, and give people a way to write it down.

Its taken a couple of tries to get here, and its still not perfect, but it looks like we’re getting pretty close. Please give it a try on your next project, and send us all your feedback and suggestions!


More examples

This is a Hardware Product Canvas for a device like the FitBit Activity Tracker

Fitbit Canvas

This is a Hardware Product Canvas for a device like the Pebble Smart Watch

Pebble Canvas

This is a Hardware Product Canvas for a device like the Thalmic MYO

Thalmic Canvas

This is a Hardware Product Canvas for a device like the Nest Learning Thermostat

Nest Canvas

This is a Hardware Product Canvas for a device like the Scanadu

Scanadu Canvas

This is a Hardware Product Canvas for a device like Lockitron Smart Lock

Lockitron Canvas

This is a Hardware Product Canvas for a device like the InteraXon Muse

InteraXon Canvas

What is the Hardware Revolution?

What is the Hardware Revolution?

hard·ware rev·o·lu·tion
/ˈhärdˌwe(ə)r/ /ˌrevəˈlo͞oSHən/

A resurgence in the consumer popularity of independently -and startup- designed hardware devices, for example: smartphones, fitness trackers, wearable devices, internet-connected devices, aerial drones and 3D printers.

Hardware Renaissance – Internet of Things – Maker Movement – The New Industrial Revolution

We are at the very beginning of an amazing change in the way we build, buy, consume, and experience devices. It’s been called a whole bunch of different names over the past decade, but now that things are (finally) heating up the name that feels like it’s going to stick is “hardware revolution”.

I want to dive into how this all came to pass. Where we are in the resurgence of hardware. What the triggers were and what needs to continue to happen. For starters: why. What has actually happened?”

What changed?

“The evolution in hardware development in some ways parallels what the software industry saw ten years ago.”

Matt Witheiler

I believe the biggest change has to do with the way hardware comes to life. And to echo Matt, I think what is happening in hardware right now is a bit like what happened to software a decade ago.

The hardware design process used to look like this:


And if you know anything about the “lean startup” principles, or Agile software development, you’re going to see a couple of very big problems with this. Personally I see a few really big ones, first the time between idea and customer. Second the time and cost between iterations. Third the upfront capital outlay for manufacturing and prototyping. And lastly the supply chain and manufacturing pressures at the end of the product cycle and their effect on growth and distribution. Long story short these things made hardware harder to do.

And this is what it looks like in indie shops and startups post-revolution:

The three really obvious differences are: the move away from old school specifications, the decoupling of final manufacturing from prototype manufacturing, and the reversal of sales and manufacturing. All three of these share some DNA with the developments happening in other industries following trends like collaboration and the move to the cloud. They also share a lot of software development and lean startup principles.

For starters, the early development stages of a hardware product historically were very waterfall-style, specification driven, design-by-committee projects. The innovation here was the introduction of both Open-Source, allowing the reuse of existing trusted hardware IP, and the introduction of Development and Breakout Boards, allowing much faster iteration in the earliest ideation stages of the project.

Following the development of a specification, hardware companies would then begin negotiating supply relationships in parallel with the design of their product. These relationships were necessary, as small and prototype-focused manufacturers didn’t yet exist and the hardware companies would have to fight for time on a manufacturing line or suffer ridiculously high prototyping costs. In the last few years the introduction of specialized prototyping equipment combined with the emergence of small-lot-size, prototype-focused manufacturers has led hardware revolution companies to decouple final manufacturing from rapid prototyping, allowing them to move substantially faster through the design phase.

Finally, the reversal of mass-manufacturing and sales, through pre-sales and crowdfunding, have allowed hardware revolution companies to both fund the manufacturing of – as well as market test – their product before building and inventorying thousands of units.

First movers

“One question I can answer is why hardware is suddenly cool. It always was cool. Physical things are great… Hackers love to build hardware, and customers love to buy it. So if the ease of shipping hardware even approached the ease of shipping software, we’d see a lot more hardware startups.”

Paul Graham

There have been very clear signals that hardware is back for about the last 3 years. Before that it was just Wired, Sketching and Maker Faire showing us the hacked up projects of a few burners. There were lots of cool IOT and M2M devices, hackerspaces, and neat writeups on Instructables, but not much commercial and little of substance. But then it started getting real. First with FitBit and MakerBot, then it grew into Pebble and Kickstarter, and in October 2012, PG wrote his famous “Hardware Renaissance“ piece.

These days we’re seeing a rate of approximately one new crowd-funded hardware startup created per week. Over the past year the press has taken to both loving and hating the hardware revolution. Here’s a snapshot:

* The hardware revolution will be crowdfunded / Venture Beat

* The hardware revolution is upon us and why it matters / Pandodaily

* New crowdfunding platform for the hardware revolution / Dragon Innovation on Betakit

* How a Garage-Based Incubator Is Fueling the Hardware Revolution / Wired

* A Hardware Renaissance in Silicon Valley / NY Times

* Silicon Valley’s hardware renaissance is stalling / CNN

* Why we’re not quite ready for the hardware startup renaissance / Venture Burn


“It’s the peace dividend of the smartphone war. Cheap processors, cheaper memory, and even cheaper sensors means it’s a great time for people who like to tinker with hardware.”

Chris Anderson

I think we are about 8 years into the hardware revolution. It all started back in 2005 with the Arduino. That led to the Raspberry Pi, the iPhone, and startups like Meraki. In 2008 FitBit would kick-off the quantified self movement with their super-smart pedometer. Continuing this trend Kickstarter, Dropcam, and Makerbot were founded in 2009, followed by crowdfunded successes like Pebble, Ouya, and Lockitron. By 2013 there have been hundreds of new hardware startups founded, with a growth rate of new startups of more than 50% year-over-year. Consumers have purchased many tens of millions of dollars of these startups’ products, and many hundreds of millions have been invested in these same companies.

If the hardware revolution as a whole were to be treated as a technology adoption lifecycle, I predict we are mid-way through the second stage. We are about 2 years away from consumers in the early majority participating in pre-sales and buying the products these startups are producing. By 2020 Gartner estimates there will be 80 billion internet connected devices, and I would predict upwards of 50% of these devices will be the products of hardware revolution companies. And finally by 2025, after 20 years, I believe we will be mid-way through the laggard market, and that there will be approximately 100 new, major, and influential consumer electronics companies in existence. I believe we still need to create a workshop, a factory, and a store to sustain the hardware revolution and I think these will get resolved in 2014, 2016 and 2019 respectively. More on this below.


“Before we can see a true “Hardware Renaissance” [we need] a company to do for hardware manufacturing what Amazon Web Services did for SaaS hosting and delivery. A commoditisation of manufacturing. Hardware-(production)-as-a-service.“

Richard Oakley

The first stage in the product ecosystem is where both designers and ideas are born. It is the most developed of the three stages, and the least in need of future innovation. The first part of the education stage is the school, a collection of resources, articles, mentors, and institutions like universities that take beginners and turn them into engineers. The second part of the education stage is the lab, a collection of hardware building blocks for testing out ideas. This is where ideas get hacked into parts, development boards get cobbled together, and the first lines of code may be written. I predict the continued dominance of both community curated content like Make as the school and open hardware merchants like Sparkfun as the lab.

The second stage in the product ecosystem is where ideas and development boards get deconstructed and refined into products. The first part of the design stage are tools, software application built to help engineers turn ideas into CAD-style manufacturing files. The second part of the design stage is the workshop, sophisticated overnight product prototyping. This is where designs get rapidly turned into testable prototypes. It will be a service offered at first by specialty manufacturers, and later addressed by the evolution of 3D printing. I (obviously) predict the success of web-based, collaborative, multi-player communities like Upverter as dominating both parts of the design stage.

The third stage in the product ecosystem is also the most in need of innovation. The first part of the production stage is the store, a mixed physical and virtual marketplace. This is where customers discover and purchase products based on a pre-sales model. The second part of the production stage is the factory, a scalable on-demand API for manufacturing. This will be the Amazon Web Services of hardware. This step still requires huge investment and innovation in logistics, smaller batch sizes, manufacturing equipment, and process control. I predict a consolidation in the store stage and the emergence of a dominant marketplace explicitly pre-selling products. Likewise I also predict manufacturers will differentiate through their APIs with the most accessible and elastic of these dominating.

What’s next?

“We believe this is just the very beginning of the hardware revolution. The world is eagerly awaiting new devices and new device platforms. Look around you and it is hard not to see opportunities.”

Jon Callaghan

Over the coming years I think we will both see a massive increase in hardware startups – Indiegogo has seen more than a 100% increase in the number of technology campaigns launched over the last year-, but also an increase in the companies that exist to make these startups successful. I think startups will be created to solve the ecosystem voids, I think massive consumer hardware companies will grow out of the startups being founded today, and I think we will be successful in putting 70+ billion more devices onto the internet over the next 7 years.

I think the wellness craze will fade out over the next 5 years, eventually being replaced by actual medical-health tech. I think drones and logistics will get have 8 or so more interesting years, that energy tech and transportation is just getting started, and that by 2020 agricultural tech will pick-up where both of these markets leave off.

I expect to see continued double or triple digit market growth, massive investment, massive innovation, and some very real consumer behaviour changes as they adopt pre-sales and purchase more and more niche devices. It’s going to be a very exciting two decades, stay tuned!

7 Hardware Ideas by Stephen Hamer


We started working together on the Open Activity Tracker design a month ago. With the help of a lot of people (Chris Gammell, Sam Grove, Bryan Thomas, Eric Evenchick, George Hahn…) we managed to design the schematic and PCB layout in slightly more than 5 hours of real-time collaboration work. We will receive prototypes shortly and will now work on the software side and the enclosure.

When we picked that specific project (#3 in the list), Stephen Hamer – co-founder here at Upverter – had drafted a list of devices he would be interested to build. We obviously won’t have time to work on all of them and we thought making it public may inspire some of you. We encourage hardware hackers to fork projects and build derivatives of the original design everyday, so it made sense to share our thoughts – well, Steve’s thoughts.

Here are 7 concepts waiting for someone to bring them to life!

1) Sleep environmental sensor suite:

There are lots of things that can affect the quality of the sleep that we get.

There are devices that record data about us while sleeping.

This device would provide the other half of that picture, what happens around us while we sleep.

This device would periodically record environmental data like:

  • ambient sound level
  • light level
  • temperature
  • humidity

2) Cloud offload GPS data collector:

Location data is fantastic but is relatively power-expensive to collect.

A team of researchers at Microsoft developed a clever method of reducing GPS power
requirements while retaining accuracy. (http://research.microsoft.com/apps/pubs/default.aspx?id=172624)

  • GPS antenna, filters
  • real time clock
  • storage

3) Activity trackers are all the rage right now.

Knowing how active you were in a day is fun data to have.

There’s limited analysis that you can personally do because of the high level of information that’s presented to you from most commercial products.

Having a device that recorded basic physical information like acceleration,
compass heading, altitude… would give an opportunity to be able to write and
share algorithms to analyse this data personally.

  • accelerometer
  • gyroscope
  • magnetic compass
  • pressure/altitude
  • storage

There are lots of things I’d like to track and record. How many times I open
the fridge in a day?

How cups of coffee do I drink a day and when?

I’ve tried lots of different things, paper, smart phone apps, text editor…

All of these have been too high friction and have only lasted for a week or two.

A push button that records the time of the press is really what I want.

I can then glue these to all of the things that I want to track stats associated

4) “Event Counter” buttons:

  • count events like: opening the fridge, cup of coffee, glass of beer from the keg…
  • push button
  • real time clock
  • storage or RF to transmit event

I’ve never had a phone whose vibration mode has been noticeable while walking.
As such I miss lots of calls and notifications while walking.

People have suggested using a Pebble smart watch. It solves the problem, I like wearing a
watch though.

What I’d really like is a Bluetooth pager. Something that I can wear on my belt
that would vibrate instead of the phone.

5) Wearable BT4 phone notification (BT pager):

  • vibration motor
  • BT4 + microcontroller
  • silence switch

Ambient display lighting can be lots easier on the eyes and provide a more
immersive experience in low light viewing.
Projects like (http://learn.adafruit.com/adalight-diy-ambient-tv-lighting) are awesome! They’re limited to computer driven content (so that Processing can
calculate the edge colors).

To be able to use this on any screen a pass-through device would have to be constructed.

6) Hardware ambient display lighting:

  • DVI-D/HDMI pass through
  • RGB LED control for back light

If your TV setup looks anything like ours you’ve got more remotes than you know
what do with. The remote also have a tendency to find the deepest, darkest
nooks and crannies of the couch to hide in.

What I’d love to solve this problem is a network accessible “IR bridge”.

It could be programmed with the different button presses from the collection of
remotes from all of your devices.

These signals could then be replayed through a local web interface.

This would let you replace all of these remotes with your smartphone.

7) Network accessible IR bridge for controlling IR controlled devices:

  • Microcontroller with network access
  • IR LED
  • IR sensor for “learning”

28 Basic PCB Design rules


These 28 Basic PCB design guidelines set out best practice to reduce the cost of your boards and to minimize the risk of errors arising during manufacture.High power boards have different rules especially in terms of spacing, traces size and power isolation. Manufacturers have different requirements; make sure you read their own guidelines before sending your design. Naming and file formats also vary depending on the manufacturers.

PCB Layout

1. Create your board frame on a 0.05″ grid. Make the lower left corner start at 0,0.

2. Usually the board frame is rectangular. For specific reasons you could also do other types of shapes such as polygons.

3. Stick parts on a 0.05″ grid. You should not break this rule unless you have a very good reason.

4. Any LED should be labeled with its purpose (power, status, D4, Lock, etc).

5. Idem for connectors: e.g Vin, Port1, Batt, 5V, etc.

6. Idem for pins where applicable: e.g TX, Power, +, -, Charge, etc.

7. Idem for switches and switch states: eg. On, Off, USB etc.

8. When applicable, it is better to avoid having vias go through the silkscreen when adding labels.

9. Group components together. For example the resistors surrounding a transistor in your schematic will also be grouped together on the PCB.

10. Minimum drill size should be 15 mil.

11. Minimum annular ring size should be 7 mil.

12. 7 mil is the minimum size for traces. 8mil is acceptable. When possible try to keep the traces size to 10mil.

13. Use thicker traces for power lines. 12mil=100mA max, 16mil=500mA max etc.

14. 7mil between traces and space is reasonable.

15. Avoid 90 degree corners. Straight lines with 45 degree corners are preferable.

16. Where applicable use a ground pour on top/bottom layers.

17. To prevent pours from shorting to traces make sure you use a 10mil isolation setting on any of the ground pour.

Schematic Layout

18. Use a GND symbol for all the GND connections.

19. Use appropriate power symbols for All VCC, 5V, 3.3V etc.

20. Add color notes to separate a complex design into various smaller bits (for example,charge circuit, accelerometer, etc).


21. All footprints need a reference designator {{refdes}}. If you come across a part on a board that doesn’t have this, you should change it and save the library. For parts requiring it a pin one marker should be defined.

22. All footprints need silkscreen indicators showing mechanical sizes, dimensions, or anything wired about the part.

23. To prevent it from flaking off easily silkscreen within a footprint or board should not go over pads or metal that will be exposed.

24. Top component layer should be marked by a red centre cross.

25. Package outline layers should outline the actual package size.

26. The Top Courtyard layer should include all of the pins.

27. When adding a footprint make sure you add a solder mask.

28. Every new footprint and part will have a human readable description.


Startup Reading List

Read? What are you crazy? We’ve got work to do!!!

That sounds a little like me harassing the troops, but don’t worry I’m smiling! Haha. As a bit of lite and fluffy Christmas reading I thought I would diverge from my hardware stories and my OSHW rant to share a bit of startup wisdom. This is the as-official-as-it-gets Upverter Startup Reading List. Back before we started Upverter there was a fairly enormous Google Wave that for a couple months seemed to be growing exponentially. It was basically a dumping ground where we all put our personal reading lists and then any new finds we stumbled across. Maybe because there was a lifetime worth of reading on it, or maybe because we had exhausted the ISBN index, I’m not quite sure, but the list eventually stopped growing.

Personally I’m maybe 1/10 through the list, Steve is probably close to half way, and Mike is a lost cause, haha. But either way, I wanted to share our favorites from the monster list. The stuff we think you absolutely need to read if you ever decide to do the startup thing. The list is a little software oriented, and that’s really just because we are building a startup that builds software, and so that’s a focus for us; but you could absolutely trim the software titles and kick-ass with your revolutionary pet rock business!

Upverter Must Read List:

Basically you’re off the team if you haven’t read these.

  • Code Complete, Second Edition This book is on more programmer book lists than should be fair. Its a must.
  • Rapid Development Right up there with Code Complete, this one normally seems to rank second on the book lists. Again a software book. Also a must.
  • The Pragmatic Programmer Don’t fix something that isn’t broken, this book often scores around 3rd. Again a software book. Also a must.
  • Peopleware This book is probably the most important management book of all time. Ever. If you ever have to work with another human being, ever, you should read this book.
  • The Mythical Man Month This book goes hand in hand with Peopleware. If you ever, in your working life, will have to make a schedule – read this book.
  • Don’t Make Me Think This book is down to earth usability. If anyone ever has to use something you build – this will explain why they hate it or get lost inside.
  • The Art of the Start This book is pitched as for anyone starting anything. And thats a pretty good take on it!
  • The Tipping Point Epidemics, viruses and human behavior. Another great read!

Upverter Reading in Progress:

The books on our kindles and iPads at this very moment.

Upverter Next in the Queue:

If I ever make it there, I have to conquer these next…