8 ms·

The stealthy industrial revolution already underway in open source hardware

69 points by crashdev 13y ago

neltnerb 13y ago
As someone who is doing an awful lot of work in the supply chain, I completely agree. Open and free to copy reference designs are the only reason I can get anything done without having impractically broad and deep knowledge. Simple example -- I make LED lighting for fun and profit. I adopted the arduino platform years ago because while I could spend the time to make sure I'm laying out all the components properly (I've done boards for lots of other microcontrollers), I hadn't done USB programming before. So instead of needing to find a partner who could write me an entire bootloader system for doing USB in circuit programming, I can just copy the arduino design, add my secret sauce, and redistribute it. Or, more recently I've been migrating to ARM. I know zilch about ARM processors, but with the mbed platform I could easily make my light into a USB MIDI device, a DMX platform, and so on. If I had to learn myself how to do all that, it would be impossibly time consuming and I'd never get anywhere.
- anigbrowl 13y ago
  mbed Oh my goodness, this is fantastic. I had been exploring another platform for this but there wasn't nearly as many examples/tutorials etc., plus it was closed source.
  - spin 13y ago
    mbed + ethernet is wonderful. If you need some firmware for such a thing, may I suggest my own github: https://github.com/mfassler/eFirmata The code is still a bit messy, but I'm still actively working on it. (I've got a bunch of changes that I haven't checked in yet.)
    - anigbrowl 13y ago
      Thanks! I'm extremely interested in the possibilites for embedded MIDI processing, as my music-making is hardware-based and I prefer to minimize the computer use.
      - neltnerb 13y ago
        Sounds like a great project! I hope you share it when you get done =) I'm a big fan of electronic music, although I don't care for MIDI due to some quirks. Are you thinking of the traditional route of making new instruments to do off-board audio generation to your mixing board? One of my favorite projects so far was to make our LED lights MIDI devices so you could sequence them using something like Ableton directly. http://blog.saikoled.com/post/49255468343/myki-show-tell-episode-3-usb-midi http://blog.saikoled.com/post/49255468343/myki-show-tell-epi... The second application is my favorite.
        anigbrowl 13y ago
        No, doing instruments is hard and at most I would build from a kit there. I already have a bunch of good quality instruments and get the sounds I like from programming those. What I'm interested in is hardware sequencing and in particular combinatorial methods for sequence generation, eg where you combine a rhythmic sequence of one length with a note sequence of another to generate changing patterns. At present I do this sort of thing with a Nord Modular but it gets unmanageable for more complex structures. Also, I don't love the sound as much as I used to but I can't give up the control over MIDI in hardware XD I approve of the audio clips on your website BTW, we share some similar musical tastes. I'll keep an eye on the project.
        neltnerb 13y ago
        Thanks! I can't say I fully understand what you mean; I was mostly taught to do electronic music using Max/MSP, so most of the stuff I make is all about generating sounds without much direct conception of things like notes or patterns. It's a little unusual of an approach, myself I do things like build simulations of throat singing based on research papers. Here's a video I did with my throat singing simulator accompanying an LED art piece I did a long time ago. http://web.mit.edu/neltnerb/www/artwork/ultraluminous-2008-03-08.avi http://web.mit.edu/neltnerb/www/artwork/ultraluminous-2008-0... I guess I don't understand how a rythmic sequence would get so complicated as to overwhelm a computer =) Is this like Conlon Nancarrow stuff? His work is incredibly bizarre. https://en.wikipedia.org/wiki/Conlon_Nancarrow https://en.wikipedia.org/wiki/Conlon_Nancarrow
  - rcxdude 13y ago
    I've a feeling MBED would have been much bigger if they hadn't started off with the pure online compiler approach. That really scared off a lot of hackers. Thankfully they've fixed most of those issues (though I don't they've opened up the 'magic interface chip' firmware).
    - neltnerb 13y ago
      I think that the NXP LPC11U2x based boards with on-board USB don't require the second interface chip. I haven't implemented it yet, but they allow direct USB programming via the same kind of system. http://mbed.org/platforms/mbed-LPC11U24/ http://mbed.org/platforms/mbed-LPC11U24/ The chip itself is under $5, so if you're implementing it yourself I think it comes in pretty cheap. You still need two USB ports though, which for me seems pretty irritating to explain to a customer I'd end up integrating it for. The best I can think of is to have a USB port for the interface, and then have a hidden internal USB port that you have to take off the case to access for reprogramming the firmware. That's ugly though, compared to the Arduino style where you can use the same port for both.
ballard 13y ago
OSH is inevitable economics; IBM saw the trend, and Dell is in play right now. Also, it would be interesting to see OSH go as far as processor on up, in the vein of OpenSparc but actual layout not just vhdl/verilog for FPGAs. Requires EDA and layout skills obviously, but it's not impossible.
6d0debc071 13y ago
If OSH really becomes a significant paradigm then we can finally start enjoying reasonably designed CPUs, with a better base abstraction layer. Maybe not everyone's cup of tea, but with the war on general purpose computing well under way that's something that I'd be very interested in.
kqr2 13y ago
Actually these kinds of practices have been used by the Chinese "shanzhai" for a while to quickly churn out products. They appear to have their own internal "GPL" / code of sharing that they self-enforce. From http://www.bunniestudios.com/blog/?p=284 http://www.bunniestudios.com/blog/?p=284 : Interestingly, the shanzhai employ a concept called the “open BOM” — they share their bill of materials and other design materials with each other, and they share any improvements made; these rules are policed by community word-of-mouth, to the extent that if someone is found cheating they are ostracized by the shanzhai ecosystem.
krapht 13y ago
Nobody who has ever made a commercial product for a living would be trumpeting a new industrial revolution brought on by open reference designs. Fact is: that stuff is there already. Component manufacturers literally have field application engineers who are just waiting to send you reference designs that use their products, if you're working at the PCB & discrete parts level. Hardware designs are already open in a way that can't be matched by software - you can look at a product, take it apart, and :voila: - you have the design in a way that is far easier to grok than disassembling a closed binary. Many high end products have full schematic drawings included with the technician's repair manual. Anything novel enough to be patented has mechanical drawings and a full description listed in the patent. The only things you can't take apart are ASICs and other computer chips. But BFD!? Any player in that space has to have hundreds of thousands just to prototype one, let alone rent time in an IC fab to make them. If you have that kind of money, there's plenty of fab-specific IP floating around for your use at nominal prices. Open source software is revolutionary because the incremental cost of duplicating bits is nearly zero. The same can't be said for hardware. In fact - it's worse. The smaller a player you are, the more expensive it is to duplicate a design. And it is a pain in the rear to duplicate designs if they're at all pushing the envelope, since manufacturing tolerances and capabilities differ so much from outfit to outfit. Can you imagine the pain and suffering you'd go through porting software that relies heavily on compiler-specific directives? Well, it's like that when you move from one factory to another. Some Chinese outfits have met with success with open / copycat hardware. I'd argue that's less a result of open reference designs and 3d printing and more of drastically reduced labor and part costs, all situated in a global hub of electronics manufacturing.
- zwieback 13y ago
  Couldn't agree more. I've worked for small companies and now work for a very large company with lots of leverage. Conservatively I would guess that we have 1-2 years of advantage over a startup when competing for the attention of large component suppliers. Just getting the lawyers on both sides to agree on a NDA can be a big production. On the other hand, when a mega company joins up with another mega company, e.g. phone manufacturer and processor supplier, there's a lot of risk aversion on both sides due to the large investments. There's a crack in the system somewhere that small innovative companies can take advantage of but I don't think open reference designs do much more than documenting the state of the art. That's great for hobbyists but makes little difference in the high-volume commercial space.
olefoo 13y ago
The real moneymaking tools in this space are going to be a class of utilities that can broadly be described as "assisted design"; everything from object editors that know about pleasing proportions to linters that will tell you that you laid out the plumbing wrong and you should fix it before it gets printed. The printers themselves will be commodified quickly; although multi-material printers will buck that trend for some time. Feedstock is already commodified and I can't really see anyone making a play on a printer/feedstock combo that required you buy only genuine branded feedstock. The basic design tools ( Meshlab, Blender; etc.) are already free, but require real expertise to understand and use. So the main market opportunities are going to be: 1. finding talented designers/architects/developers and matching them with the demand for design services. 2. parametric design adapters that can take a common set of parameters and produce an individualized design ( from clothing to buildings ) 3. Assistive design tools; tools that help you realize your designs while ensuring that they remain within the bounds of manufacturability and are pleasing to the eye and hand; without requiring that you become an expert in whatever field you are dabbling in. ( Think of a jewelry design application that knows about both physics and proportion, and suggests alterations to you. Or a habitat builder that can show you a map of where mildew will happen in your design...) The basic premise of this class of application is that you could hand it to a bright fourteen year old and she could design a successful and functional instance of the target class of object. If you're hungry for good ideas; there's a fairly rich vein of common items that people will want a customized version of, but won't necessarily be wanting a bespoke or personal design. That would be point 2. above.
sown 13y ago
I recently mentioned in passing that a hardware renaissance is near or underway. This sort of news gives me hope that it happens.
stilldavid 13y ago
Nathan Seidle, founder of SparkFun (and my boss!) recently talked at a hearing of the House Committee on IP with regards specifically to open source hardware. The EFF quoted him in this article: https://www.eff.org/deeplinks/2013/08/tech-sector-does-its-part-promote-reality-based-copyright-policy-will-congress https://www.eff.org/deeplinks/2013/08/tech-sector-does-its-p... And it echoes how he feels and how we as a company operate. 145 people strong, $30MM in revenue annually, and not a single patent. All of the products we make are open source. It's awesome.

69 points by crashdev 13y ago

As someone who is doing an awful lot of work in the supply chain, I completely agree. Open and free to copy reference designs are the only reason I can get anything done without having impractically broad and deep knowledge. Simple example -- I make LED lighting for fun and profit. I adopted the arduino platform years ago because while I could spend the time to make sure I'm laying out all the components properly (I've done boards for lots of other microcontrollers), I hadn't done USB programming before. So instead of needing to find a partner who could write me an entire bootloader system for doing USB in circuit programming, I can just copy the arduino design, add my secret sauce, and redistribute it. Or, more recently I've been migrating to ARM. I know zilch about ARM processors, but with the mbed platform I could easily make my light into a USB MIDI device, a DMX platform, and so on. If I had to learn myself how to do all that, it would be impossibly time consuming and I'd never get anywhere.

anigbrowl 13y ago
mbed Oh my goodness, this is fantastic. I had been exploring another platform for this but there wasn't nearly as many examples/tutorials etc., plus it was closed source.
- spin 13y ago
  mbed + ethernet is wonderful. If you need some firmware for such a thing, may I suggest my own github: https://github.com/mfassler/eFirmata The code is still a bit messy, but I'm still actively working on it. (I've got a bunch of changes that I haven't checked in yet.)
  - anigbrowl 13y ago
    Thanks! I'm extremely interested in the possibilites for embedded MIDI processing, as my music-making is hardware-based and I prefer to minimize the computer use.
    - neltnerb 13y ago
      Sounds like a great project! I hope you share it when you get done =) I'm a big fan of electronic music, although I don't care for MIDI due to some quirks. Are you thinking of the traditional route of making new instruments to do off-board audio generation to your mixing board? One of my favorite projects so far was to make our LED lights MIDI devices so you could sequence them using something like Ableton directly. http://blog.saikoled.com/post/49255468343/myki-show-tell-episode-3-usb-midi http://blog.saikoled.com/post/49255468343/myki-show-tell-epi... The second application is my favorite.
      - anigbrowl 13y ago
        No, doing instruments is hard and at most I would build from a kit there. I already have a bunch of good quality instruments and get the sounds I like from programming those. What I'm interested in is hardware sequencing and in particular combinatorial methods for sequence generation, eg where you combine a rhythmic sequence of one length with a note sequence of another to generate changing patterns. At present I do this sort of thing with a Nord Modular but it gets unmanageable for more complex structures. Also, I don't love the sound as much as I used to but I can't give up the control over MIDI in hardware XD I approve of the audio clips on your website BTW, we share some similar musical tastes. I'll keep an eye on the project.
        neltnerb 13y ago
        Thanks! I can't say I fully understand what you mean; I was mostly taught to do electronic music using Max/MSP, so most of the stuff I make is all about generating sounds without much direct conception of things like notes or patterns. It's a little unusual of an approach, myself I do things like build simulations of throat singing based on research papers. Here's a video I did with my throat singing simulator accompanying an LED art piece I did a long time ago. http://web.mit.edu/neltnerb/www/artwork/ultraluminous-2008-03-08.avi http://web.mit.edu/neltnerb/www/artwork/ultraluminous-2008-0... I guess I don't understand how a rythmic sequence would get so complicated as to overwhelm a computer =) Is this like Conlon Nancarrow stuff? His work is incredibly bizarre. https://en.wikipedia.org/wiki/Conlon_Nancarrow https://en.wikipedia.org/wiki/Conlon_Nancarrow
- rcxdude 13y ago
  I've a feeling MBED would have been much bigger if they hadn't started off with the pure online compiler approach. That really scared off a lot of hackers. Thankfully they've fixed most of those issues (though I don't they've opened up the 'magic interface chip' firmware).
  - neltnerb 13y ago
    I think that the NXP LPC11U2x based boards with on-board USB don't require the second interface chip. I haven't implemented it yet, but they allow direct USB programming via the same kind of system. http://mbed.org/platforms/mbed-LPC11U24/ http://mbed.org/platforms/mbed-LPC11U24/ The chip itself is under $5, so if you're implementing it yourself I think it comes in pretty cheap. You still need two USB ports though, which for me seems pretty irritating to explain to a customer I'd end up integrating it for. The best I can think of is to have a USB port for the interface, and then have a hidden internal USB port that you have to take off the case to access for reprogramming the firmware. That's ugly though, compared to the Arduino style where you can use the same port for both.

ballard 13y ago

OSH is inevitable economics; IBM saw the trend, and Dell is in play right now. Also, it would be interesting to see OSH go as far as processor on up, in the vein of OpenSparc but actual layout not just vhdl/verilog for FPGAs. Requires EDA and layout skills obviously, but it's not impossible.

6d0debc071 13y ago

If OSH really becomes a significant paradigm then we can finally start enjoying reasonably designed CPUs, with a better base abstraction layer. Maybe not everyone's cup of tea, but with the war on general purpose computing well under way that's something that I'd be very interested in.

kqr2 13y ago

Actually these kinds of practices have been used by the Chinese "shanzhai" for a while to quickly churn out products. They appear to have their own internal "GPL" / code of sharing that they self-enforce. From http://www.bunniestudios.com/blog/?p=284 http://www.bunniestudios.com/blog/?p=284 : Interestingly, the shanzhai employ a concept called the “open BOM” — they share their bill of materials and other design materials with each other, and they share any improvements made; these rules are policed by community word-of-mouth, to the extent that if someone is found cheating they are ostracized by the shanzhai ecosystem.

krapht 13y ago

Nobody who has ever made a commercial product for a living would be trumpeting a new industrial revolution brought on by open reference designs. Fact is: that stuff is there already. Component manufacturers literally have field application engineers who are just waiting to send you reference designs that use their products, if you're working at the PCB & discrete parts level. Hardware designs are already open in a way that can't be matched by software - you can look at a product, take it apart, and :voila: - you have the design in a way that is far easier to grok than disassembling a closed binary. Many high end products have full schematic drawings included with the technician's repair manual. Anything novel enough to be patented has mechanical drawings and a full description listed in the patent. The only things you can't take apart are ASICs and other computer chips. But BFD!? Any player in that space has to have hundreds of thousands just to prototype one, let alone rent time in an IC fab to make them. If you have that kind of money, there's plenty of fab-specific IP floating around for your use at nominal prices. Open source software is revolutionary because the incremental cost of duplicating bits is nearly zero. The same can't be said for hardware. In fact - it's worse. The smaller a player you are, the more expensive it is to duplicate a design. And it is a pain in the rear to duplicate designs if they're at all pushing the envelope, since manufacturing tolerances and capabilities differ so much from outfit to outfit. Can you imagine the pain and suffering you'd go through porting software that relies heavily on compiler-specific directives? Well, it's like that when you move from one factory to another. Some Chinese outfits have met with success with open / copycat hardware. I'd argue that's less a result of open reference designs and 3d printing and more of drastically reduced labor and part costs, all situated in a global hub of electronics manufacturing.

zwieback 13y ago
Couldn't agree more. I've worked for small companies and now work for a very large company with lots of leverage. Conservatively I would guess that we have 1-2 years of advantage over a startup when competing for the attention of large component suppliers. Just getting the lawyers on both sides to agree on a NDA can be a big production. On the other hand, when a mega company joins up with another mega company, e.g. phone manufacturer and processor supplier, there's a lot of risk aversion on both sides due to the large investments. There's a crack in the system somewhere that small innovative companies can take advantage of but I don't think open reference designs do much more than documenting the state of the art. That's great for hobbyists but makes little difference in the high-volume commercial space.

olefoo 13y ago

The real moneymaking tools in this space are going to be a class of utilities that can broadly be described as "assisted design"; everything from object editors that know about pleasing proportions to linters that will tell you that you laid out the plumbing wrong and you should fix it before it gets printed. The printers themselves will be commodified quickly; although multi-material printers will buck that trend for some time. Feedstock is already commodified and I can't really see anyone making a play on a printer/feedstock combo that required you buy only genuine branded feedstock. The basic design tools ( Meshlab, Blender; etc.) are already free, but require real expertise to understand and use. So the main market opportunities are going to be: 1. finding talented designers/architects/developers and matching them with the demand for design services. 2. parametric design adapters that can take a common set of parameters and produce an individualized design ( from clothing to buildings ) 3. Assistive design tools; tools that help you realize your designs while ensuring that they remain within the bounds of manufacturability and are pleasing to the eye and hand; without requiring that you become an expert in whatever field you are dabbling in. ( Think of a jewelry design application that knows about both physics and proportion, and suggests alterations to you. Or a habitat builder that can show you a map of where mildew will happen in your design...) The basic premise of this class of application is that you could hand it to a bright fourteen year old and she could design a successful and functional instance of the target class of object. If you're hungry for good ideas; there's a fairly rich vein of common items that people will want a customized version of, but won't necessarily be wanting a bespoke or personal design. That would be point 2. above.

sown 13y ago

I recently mentioned in passing that a hardware renaissance is near or underway. This sort of news gives me hope that it happens.

stilldavid 13y ago

Nathan Seidle, founder of SparkFun (and my boss!) recently talked at a hearing of the House Committee on IP with regards specifically to open source hardware. The EFF quoted him in this article: https://www.eff.org/deeplinks/2013/08/tech-sector-does-its-part-promote-reality-based-copyright-policy-will-congress https://www.eff.org/deeplinks/2013/08/tech-sector-does-its-p... And it echoes how he feels and how we as a company operate. 145 people strong, $30MM in revenue annually, and not a single patent. All of the products we make are open source. It's awesome.