SAN JOSE, Calif. – There's no doubt that engineers like the idea of open-source hardware. There are an increasing number of open-source hardware board designs – Arduino, Raspberry Pi, Beagleboard and many others – that enable hobbyist projects and the reuse of board designs in commercial products. And many engineers are putting a lot of time into enabling these movements via collaborative work online and through the creation of vibrant online communities.
What is less clear is whether such movements will scale into the commercial world. There is a lack of clear business model and dependence, in some cases, on the donation of engineers' time by commercial organizations. That was one of the conclusions from a panel discussion moderated by EE Times editor-in-chief Alex Wolfe.
Over its short life the Raspberry Pi low-cost single-board computer, based on an ARM11-based system-chip from Broadcom, has been a phenomenal success in terms of shipments. But what remains unclear is how widely the board is fulfilling its original brief of teaching young people how to program or is being adopted as a building block in commercial embedded equipment designs.
Gert Van Loo, senior principal engineer with Broadcom Corp. in Cambridge, England, and architect of the prototype of the Raspberry-Pi computer board, said that commercial uptake has been gated by considerations of whether the Raspberry Pi Foundation can guarantee to be able to supply boards in five or ten years time. The Raspberry Pi Foundation is doing its best to make those assurances, he said.
Another road-block is the case that writing the license terms for open-source hardware is complex compared with open-source software, the panelists offered. Jason Kridner, software architecture manager with Texas Instruments and co-founder of BeagleBoard.org, commented: "It is cheap to replicate bits but it is expensive to replicate atoms."
Chris Taylor, engineering project manager with SparkFun Electronics (Boulder, Colorado), said that there are issues that make convergence on a unified open-source license for hardware much harder than it has been for the open-source software movement. SparkFun is an online vendor of components for the "maker" community. "The Open-Source Hardware Association is still trying to figure out how that [license terms] will play. A hardware license is nebulous but there is work with Creative Commons [organization]," Taylor said. One camp thinks a universal license can be created; another camp doesn't, he added.
Taylor made his position clear when he said the creation of a single licensing environment for OSH is neither likely nor necessary and that many types of license would be tried. "OSH is built on that chaos. No one is ever going to agree on one license," he said.
Click on image to enlarge.
From left to right: Alex Wolfe, editor-in-chief of EE Times, moderates panelists Jason Kridner, software architecture manager at Texas Instruments; Chris Taylor, engineering project manager at SparkFun Electronics; Pierre Michael, co-founder of Party Robotics; and Gert Van Loo, senior principal engineer with Broadcom Corp.
Arduino and Raspberry Pi are neat little boards which are fine for hobby projects and the one-off test fixture or whatever.
However, they're nothing more special than an MCU or FPGA vendor's eval board or starter kit. Their main attraction is that there's a whole lot of firmware modules and plug-ins available for them.
But as for "What is less clear is whether such movements will scale into the commercial world," the answer is pretty obvious: they won't. If you have to add on other hardware, or you want to use reasonable enclosures, or any of the hundred other things that go into a product design, you won't use a Raspberry Pi as part of your product. You'll spin a PCB that fits into your enclosure and uses a proper power supply and rolls up all of that extra hardware.
So, no, "open source hardware" in that sense doesn't scale. Maybe you'll take the same processor as the Raspberry Pi and integrated it into your design, but you'll do that to take advantage of the various firmware models others have written. But when you do that, you'll look at whether the processor is optimal for the design. Is it too big and you can get away with a smaller (fewer pins, less memory) device that's also cheaper? Are you pushing the limits of the micro and want to use something bigger to allow for future expansion?
I think the real power of the open source boards is the ability for hobbyists and professionals alike to quickly prototype a design using readily available modules. The switch may come when the OS Hardware model is better understood, but in the meantime could there be an agreement between the creators of the OS hardware and a potential commercial producer? Just wondering if it is possible to pay a small fee (either per board or a larger one time) to use it in production.
Back in the late 1970's and 1980's, the barriers to entry for new hardware companies were pretty low. Schematic capture could be done with paper and pencil. Layout was generally black tape on clear plastic. Parts were big and easy to and solder.
That helped spur an unprecedented flood of innovation and start-up companies. Open source software helped do the same with the Internet and open source hardware, as well as inexpensive or free development tools, is now doing the same again with the hardware world.
Certainly a great number of those start-ups are destined to fail, but the strong ones will drive our economy as did Apple, Microsoft, Dell, Google and others. Open source has brought the barriers to entry back down and opened up possibility to uncountable numbers of people that otherwise would be stuck toiling away in a cube for someone else.
It really doesn't matter if open source, specifically, has a business model in it. Maybe it does, maybe it doesn't. The greatest value with it comes from greatly increasing the opportunity for people to get into the business of designing, building and manufacturing.
Bob Baddeley gave a great talk yesterday at Design West called "An Engineer's Guide to Braving the Hardware Startup World". The real problems and what soaks up your time and seed money are things like FCC/UL approvals, fixed costs like molds for injection molding, setting up manufacturing, making sure manufacturing is actually doing its job, and patent exposure. I'm personally amazed that Bob was able to stick with it and learn everything he had to learn to do this. He says that half his income now comes from sharing this largely non-technical knowledge as a consultant :-)
I don't know why Arduino and Raspi are unique “open source hardware” projects. Since decades MCU or FPGA evaluation board provider offer the free schematic and layout by default, not really calling it open source.
The Arduino IDE is open source software, but I haven't seen many projects “really” modifying this IDE to add new functionality (not only adapting it for their processor).
The unique point is, that these tools are very cheap and very easy to use, that's it.
There is a nice multicore Arduino project coming up: http://www.cloudx.cc/projects.html .
The business model for Open source hardware is subtle.
With Open source hardware you give away the schematics and source code so that people can buy the main board from you and then tinker with add-ons and the software. This is a good marketing ploy as it increases sales of the actual hardware. Also, because the main board uses SMD components and FCC/UL approvals etc. are so onerous, there is little to no chance that any domestic customer will actually start producing variants of your main board in any competitive quantity.
This is the business model that has enabled the Raspberry-Pi to sell in the millions and make pots of money.
Overall it is win-win. It helps educate the public and encourages engineering innovation by consumers, whilst still allowing the company to make sizeable profits.
(NB: This business model breaks down if you try to sell open source hardware to big corporations who have the resources to copy your design and then manufacture it themselves in quantity)
Open source hardware is also a nice antidote to the modern trend of super-closed systems like modern laptops which make it very difficult (or sometimes actually impossible) for an end-user to change the operating system from windows to Linux.
The difference between a great idea and a great business is huge. From my perspective, that's a large part of the Internet bubble/crash in 2000 and 2001. Whether it's hardware or software, an idea isn't always enough and that fact that one or a few of something can be easily made doesn't indicate that the thing can be mass produced.
Regardless, our economy depends on a lot of people being willing to try to start businesses. I can see open source allowing more people to try with a bit less upfront investment.
Duane, I agree there is a significant difference between a good idea and a great business. In many ways the corrallary holds true as well, there is a significant difference between a hobby design and a production worthly design. I am not sure about the internet bubble / crash being a bad business; I thought it was good business but bad (overly optimistic - carried away) valuation of the companies. Prices soaring without any one THINKing about the value add and the business model. A lesson for all of us to keep in mind. It is all too easy for us to think our hobby product (hand crafted, tuned, babied, etc.) is production worthy without a lot more work both on the product itself and the support, documentation, warraenty etc..
Join our online Radio Show on Friday 11th July starting at 2:00pm Eastern, when EETimes editor of all things fun and interesting, Max Maxfield, and embedded systems expert, Jack Ganssle, will debate as to just what is, and is not, and embedded system.