What is Open Source Hardware?

ptorrone writes "In their piece 'What is open source hardware?', MAKE magazine divides up electronic hardware into layers, each of which has different document types and licensing concerns: Hardware (mechanical) diagrams, schematics & circuit diagrams, layout diagrams, core/firmware, software/API — each layer has an example provided and links to many of the open source hardware projects currently being worked on."

Wrong Preferred Document Formats

[Anonymous Coward]

The preferred document formats don't make sense. Distributing a schematic diagram as an image (pdf, png, etc.) is like distributing a program as an object file. In both cases modifying the item is nearly impossible. Schematic diagrams should be distributed in a format which a schematic editor, such as geda, can read. This is the electronic equivalent of source code. The same thing goes for printed circuit board layouts: just the image is not sufficient.

In theory ...

[Anonymous Coward]

there's no limit to what you can open source. At some point, it ceases to be worthwhile. For instance, I could build the occasional part for my hot rod. In theory there's nothing to stop me from digging and smelting my own ore and really building a car from scratch. Obviously it isn't going to happen though. There are practical limits on what it's worthwhile to open source. Yes you can open source any design but if nobody builds that design then, what's the point?

Re:In theory ...

[amolapacificapaloma]

I think the point is that maybe someday you might only be able to
buy sophisticated electronic products which are full of DRM or any
other crap (rootkits, etc) and if by that day there is no open
source alternative (that meaning open documentation and the likes)
with at least the most important features and no patent problems,
nobody is going to be able to start from the scratch such a big project.
And that would suck big time.

Re:My only problem is that...

[hackerjoe]

They have no geek passion. They are irrelevant to the discipline. And they are exactly what Corporate America wants them to be.

Whatever, I have geek passion, I just don't have all-consuming geek passion. My job is a perfect outlet for the geek passion -- I get to be a geek all day. Then in my off-time I can hang out with friends, listen to music, dance, cook, whatever other hobbies I'm currently pursuing.

I'm pretty sure that if I worked in sales or management I'd have to work on electronics or write code when I got home. I sure did when I was going to school.

How much hardware variety is truly needed?

[bmajik]

Rank the following tasks in order of complexity:

- gate-level design of a modern CPU
- gate-level design of a modern GPU
- gate-level design of a modern northbridge
- gate-level design of a modern southbridge
- gate-level design of a modern audio controller
- gate-level design of a modern ethernet controller
- gate-level design of a modern wifi chip
- gate-level design of a modern usb controller
- the linux kernel

my understanding is that there is a lot of really, really badly made hardware out there. the software people are clever enough to reverse engineer the hardware and write drivers. Why not put a few of them to work forward engineering the hardware?

Which peices of a modern computing system cannot run acceptably off of re-flashable firmware, or better yet, re-flashable FPGAs?

At this point, are (some) resources better spent trying to create F/OSS reference designs for every essential component to build a fully open computer platform?

I like the idea of being able to have a 100% open computer, where each of the components is well understood and discussed out in the open, and people aren't wasting a lot of time supporting badly made hardware. Some de-facto standardization around reference open source implementations of the hardware could be a pretty good thing.

It's actually not stuff like the CPU that i care about.. its more like.. all of the other things that make it onto a motherboard. There's no reason to put up with noisy audio, non-functional s/pdif outputs, buggy "hardware" raid, crappy bios, etc. The only value-add in these components is when they manage to live up to their as-advertised specs reliably.

Re:Wrong Preferred Document Formats

[AndreyFilippov]

Unfortunately it is yet difficult to design hardware using only the GPL-ed software. Sometimes - even running of GPL-ed OS - until recently fro FPGA design I had to use Xilinx software on other OS - Linux version was much worse (I had frequently ssh to my computer after it stopped to respond to any keypresses). Now it is better, and I do not need that OS for this job anymore. But for PCB design I still need a combination of the proprietary OS with proprietary EDA software - even if I would pay at least twice more for the free software if it could do the job with the same convenience.

So we too provide hardware documentation as pdf files. Actually, each of our hardware products has tarball with all the source code (software and FPGA) sufficient (combined with other free software available) to regenerate the bootable firmware (original or modified) inside the flash image itself.

I would not agree that pdf is like binary - it is more like printed source code that you still can use to build the code if you type good enough :-)

Re:How much hardware variety is truly needed?

[bmajik]

What was the point of your message? Do you actually think I don't know about pci cards?

I don't think there is an issue of "cheap" or "not cheap" here.. irrespective of how much or how little you pay for a peice of PC hardware, it will tend to have some fault when used in combination with some other peice of hardware.. or it will have some quirk that makes it irritating for your particular scenario. Want your machine to S3 sleep? Hope all of your expansion cards work properly with S3. The fan on my VGA card doesn't power down in sleep modes.. only in hibernate... so I effectively can't use sleep. Now, if i scour high resolution board photos of any part i buy before buying it, i MIGHT get to learn things like that.. but whenever you do a new machine build there is always some discovery / quirkyness to uncover.. no matter how much time you spend reading reviews of hardware from other people, or how carefully you research components.

One answer to this is "buy a mac", where the whole stack from silicon to software is owned and tested as a cohesive unit. There are some advantages to that model, and I don't see why the same model can't work, or even be better, with a mostly or completely open system.

My only problem is that...obsession.

[Anonymous Coward]

"It's not that there is a lack of talent. Rather, apathy is fatal to open source. And we need to come to terms with the fact that the overwhelming majority of those with the knowledge to do something disruptive, to use their skill to change their world for the better, choose just to go home at night and watch tv."

Well I have the skill and talent however my education wasn't "open-source" and therefore I spend a lot of time at work paying for it, and when I get home the last thing I want to be bothered with is "my job".

Re:How much hardware variety is truly needed?

[evilviper]

there is a lot of really, really badly made hardware out there. the software people are clever enough to reverse engineer the hardware and write drivers. Why not put a few of them to work forward engineering the hardware?

I don't believe I can reasonably even count all the reasons why not, let alone explain them all here...

First, I'd say economies of scale... The fewer people buy it, the more you'll have to charge, and the more you charge, the fewer people will buy one...

Another is the pace of technology... Every time hardware changes, you have to update the design, and start building new hardware... eg. DDR to DDR2 RAM, Socket 939 to AM2, etc., etc.

To be a real option, you're going to have to have different form factors for hardware. With motherboards that means ATX, microATX, nanoATX, and whatever else. For graphics that means PCI, AGP 2/4/8X, etc., as well as PCIe, and integrated chipsets for the purpose as well.

Additionally, while creating drivers for undocumented hardware is quite difficult, it's still at least an order of magnitude easier to send bytes to a device and see what they do, than it is designing an efficient chip, even for something simple like sound.

But the point that I think cuts directly to the heart of the issue is: If people were willing to standardize on a single reference platform, as dictated by an open source guru, you could just start doing that tomorrow... Name the CPU, name the motherboard, name the sound card, graphics, etc., etc. Then everyone's efforts are focused on a single set of hardware, with working drivers for that small set of hardware, etc.

That would be using normal economic forces to your advantage, instead of trying to fight market forces, and enter the market yourself. It could make open source a valuable bloc of customers for any company who can offer reliable and documented products. The problem is, of course, that nobody is going to accept those terms. People want to use the hardware they have, and don't want to be restricted to the lowest common denominator hardware, lacking the features, specs, or the form factor they want.

As has already been said by others, a hardware review site, which extensively tested equipment for 100% correctness, all-around quality, and open source compatibility, would be extremely valuable, and much more helpful than an over-priced reference platform.

Re:In theory ...

[shaitand]

'Open source means free. Everything else is just a motivation to get people to work without being paid while the work itself is systematically devalued.'

Open source does not mean free as in beer. Its about freedom to use, modify, and distribute.

Re:My only problem is that...

[iplayfast]

The lack of people who do this stuff for fun is frustrating. I program for a living and for fun, http://code.google.com/p/crylib/wiki/CryLib [google.com] The problem is that in order for people to be excited about working after hours as a hobby, they have to have a motive or project that really excites them. Programming for most people is a means to an end. If the end is exciting or liberating in some way (an OS kernel is a good example) then more people will join in.

In my case I'm interested in A.I. and decided that I needed a framework to build some of my ideas around, so I started working on my library. Several years later, I'm still working on my library and only have a couple of A.I. things in it. But the library is getting kinda cool, so I guess it's taking over.