Things you think are opensource hardware, but aren't

[charlespax]

Can you help me this? I'm writing an article about things people think
are open source hardware, but aren't. So far I have MakerBot, Red
Pitaya, and the Raspberry Pi. What else can you think of?

I basically want to make a list of things that are almost opensource
hardware or that people think is opensource hardware when it's maybe
just the software that's opensource or nothing at all. I'd also like
to provide opensource alternatitives.

Here's what I have so far

Closed: Raspberry Pi. Closed hardware, open software
Open: Beagle Bone Black. Open hardware and software. On Github.

Closed: MakerBot. Was open through the original Replicator. Closed
now. Many Replicator derivatives
Open: Crap ton of derivatives. RepRap.

Closed: Red Pitaty. They really rub me with this one, "Red Pitaya is
an open-source measurement and control tool replacing many expensive
laboratory instruments!" They are NOT open hardware.
Open: Unknown

Closed: TC4C coffee roasting controller. The TC4 shield is open source
and the author provides schematic for the TC4C device, but there Eagle
files are not available.
Open: What a coincidence, my T400 can be hacked to do the same stuff,
but there's much better alternatives.

[donotdespisethesnake]

I know some people argue about what constitutes "Open Source Hardware", in my view it is must have an officially recognised license.

The list of things that are genuinely Open Source is quite short... I've seen dozens of projects licensed under CC-BY-NC (not an Open Source license). The RepRap forums have decided on their own very weak version of "open source", which no Open Source organisation would recognise.

The term Open Source Hardware is so widely abused, it's hardly worth it - look at this list https://medium.com/@gendor/10-commercially-successful-open-source-hardware-projects-in-2013-515a799daa89 - probably half are not open.

Take a close look at the Reprap derivatives - most of the commercial "open" 3D printers are licensed as CC-BY-NC, famous ones like printrbot. They were very unhappy when someone tried to sell products based on their designs, threatening legal action. LulzBot are the only people I have seen that have gone as far as getting their design OSI approved.

Not-open: OpenROV http://www.openrov.com/ sounds open doesn't it? Software is, hardware isn't. I gave up on OSHW organisation after finding that OpenROV is a member, and their guy was even invited to be a guest speaker.

Not-open : http://www.cmucam.org/ It even has the OSHW logo on the board, but their page has this claim

To use or sell any CMUcam model commercially you must contact us for a royalty-based license.

. So not only not open, they are claiming legal rights where they have none! Hardware designs do not legally carry Copyright - so what are they charging for?? I queried this with them, they gave me some BS answer.

Open: OpenMV https://hackaday.io/project/1313-openmv

[donotdespisethesnake]

And not all Arduino's are open ... http://hackaday.com/2015/02/24/is-the-arduino-yun-open-hardware/

Also the new Arduino Zero has the Atmel EDBG chip, which crosses the line. It's a standard Atmel chip - programmed with proprietary debug firmware. People might argue it is a single source, commodity chip, e.g like FTDI. But it is clearly a way for Atmel to skirt the Open Source definition and hide source code in a proprietary form.

[Prime73]

Raspberry Pi is on github: https://github.com/raspberrypi/documentation/tree/master/hardware/raspberrypi
Why you consider it as a closed hardware project ?

[zapta]

Teensy 3.1.

It is promoted for example by OSH PARK on checkout  (the teensy proprietor is a local buddy of them) but is not open source hardware and you need to buy a licensing chip if you want to actually build one.


http://store.oshpark.com/products/teensy-3-1

[savril]

Raspberry Pi is on github: https://github.com/raspberrypi/documentation/tree/master/hardware/raspberrypi
Why you consider it as a closed hardware project ?

There's no clear indication of the license of the source. Exposing your code doesn't mean it is in the public domain.
And there's another problem with the video driver. There's no open source driver for the video chip and you have to pay a fee to use MPEG4 acceleration. So even if the hardware was open source, the whole platform isn't open-source.
Another problem is that you can't buy the CPU by yourself.

More info here : https://wiki.debian.org/RaspberryPi#line-21

[janoc]

But it is clearly a way for Atmel to skirt the Open Source definition and hide source code in a proprietary form.

Hold your horses!

Not everything is an anti-open-source conspiracy. Atmel's debugging protocols, including their JTAG implementation, were always proprietary and undocumented. That predates Arduino and even the entire Open Source Hardware concept by years, if not a decade.

All that the EDBG chip does is that it implements their proprietary debugging protocol(s) and lets you interface with them using a published interface. The same as e.g. the AVR chip in the AVR Dragon does (which also predates Arduino and all that by years).

I am not happy about this solution neither, but:
a) I don't think Atmel cares about some open source hardware or not (honestly, Arduino market is a drop in a bucket for them, pretty much all their tools are completely proprietary and always have been, etc.
b) They certainly aren't "skirting" anything - all that proprietary debugging stuff has always been there.
c) Atmel != Arduino - they aren't responsible for whatever some 3rd-party does with their hardware.

If you want to blame someone for the thing being closed and proprietary, blame the Arduino designers for choosing a platform that requires proprietary debuggers.

[janoc]

And there's another problem with the video driver. There's no open source driver for the video chip and you have to pay a fee to use MPEG4 acceleration. So even if the hardware was open source, the whole platform isn't open-source.
Another problem is that you can't buy the CPU by yourself.

The SoC (it is not just a CPU) can certainly be bought if you want to buy a reasonable quantity. Now Broadcom's definition of "reasonable" likely isn't the same as yours and mine, unfortunately.

The fee for MPEG4 encoding is actually because of the patent licensing from the MPEG consortium. That is the same regardless of platform and nothing much the Raspberry Pi Foundation or Broadcom can do about.

And the deal with the video drivers - yeah, but big deal ... What you are describing is the absolutist, the strongest possible definition of "open source" that pretty much nothing will ever satisfy, because any hardware you use is full of binary blobs with firmware, microcode, configuration, etc.

The question is not that "everything must be open" (that is not realistic to achieve) but how much should/needs to be open for the device to be actually useful. That is where a more meaningful discussion could be had.

I do wonder whether people condemning the Pi for not being open source (which wasn't even the design goal, btw) can actually do something meaningful with that data, should it (theoretically) become open - how many people can write an open graphic driver for an obscure platform? Sure, there are some such people around, but hey, I think most have better things to do with their time. And for non-graphic uses (like video encoding and such) - why bother, one can probably find a different, less locked down SoC for that than the old Broadcom one used in the RasPi.

[eas]

Raspberry Pi is on github: https://github.com/raspberrypi/documentation/tree/master/hardware/raspberrypi
Why you consider it as a closed hardware project ?

There's no clear indication of the license of the source. Exposing your code doesn't mean it is in the public domain.

And Open Source != public domain in many jurisdictions. Open Source licenses often depend on copyright law. And often, public domain are distinguished as such because they are not under copyright.

The question is now that "everything must be open" (that is not realistic to achieve) but how much should/needs to be open for the device to be actually useful. That is where a more meaningful discussion could be had.

I do wonder whether people condemning the Pi for not being open source (which wasn't even the design goal, btw) can actually do something meaningful with that data, should it (theoretically) become open - how many people can write an open graphic driver for an obscure platform? Sure, there are some such people around, but hey, I think most have better things to do with their time. And for non-graphic uses (like video encoding and such) - why bother, one can probably find a different, less locked down SoC for that than the old Broadcom one used in the RasPi.

I tend towards pragmatism, but from that perspective, I realize that, while on the one hand, RiPi didn't have complete openness as a design goal, and that they seem to have opened up some things that weren't open at the start, that may be in part due to the pressure placed on them by absolutists. Moreover, as a pragmatist, I have to acknowledge that absolutism may be a pragmatic tactic in pursuit of pragmatic goals.

As for things that aren't as open as people think they are, how about the lovely AVR Transistor Tester project? Based on foolishly accepting the opinions of various forum dwellers, I believed that the software, if no the hardware, was covered by some sort of copyleft license like the GPL. When I looked more closely though, I couldn't find a clear license on the m & k versions. The original creator seems to have released the code and circuits as free for non-commercial use.

[charlespax]

The RepRap forums have decided on their own very weak version of "open source", which no Open Source organisation would recognise.

I thought the RepRap was GPL. Are you talking about just the content of the forum or about the hardware?

The term Open Source Hardware is so widely abused, it's hardly worth it - look at this list https://medium.com/@gendor/10-commercially-successful-open-source-hardware-projects-in-2013-515a799daa89 - probably half are not open.

Awesome! Thanks for that link. I hope to bring some awareness to the problem and suggest alternatives for the closed stuff.

With all these projects I might have to make make more than one post. Otherwise it'll be "53 Things you think are opensource hardware, but aren't"

[charlespax]

The question is now that "everything must be open" (that is not realistic to achieve) but how much should/needs to be open for the device to be actually useful. That is where a more meaningful discussion could be had.

Exactly. It's not like people are releasing silicon wafer files or manufacturing process documentation.

[charlespax]

As for things that aren't as open as people think they are, how about the lovely AVR Transistor Tester project? Based on foolishly accepting the opinions of various forum dwellers, I believed that the software, if no the hardware, was covered by some sort of copyleft license like the GPL. When I looked more closely though, I couldn't find a clear license on the m & k versions. The original creator seems to have released the code and circuits as free for non-commercial use.

Are you talking about the AVR TransistorTester by Karl-Heinz Kubbeler referred to here and here

[senso]

Not-open: OpenROV http://www.openrov.com/ sounds open doesn't it? Software is, hardware isn't. I gave up on OSHW organisation after finding that OpenROV is a member, and their guy was even invited to be a guest speaker.

Never had any interest in the OpenROV thing, but isn't schematics, layouts and gerbers enough for the Hardware part?
https://github.com/OpenROV/openrov-electronics

[m100]

Nothing to stop anyone anywhere in the world producing a Raspberry Pi, and ignoring the binary blobs everything is published , every component freely available as far as I know if you meet the minimum order quantities. 

Just what is the point of it not being fully open souce hardware / firmware in this case though?    It's cheap, readily available off the shelf in many versions, in two distinctly different form factors.   The fact it has sold in its millions and is used other than on a desk in teaching environment says a lot.

It's just another building block, the BC107, 741, 4001, Z80, 6502, 8085 of its day.  With the exception of the Broadcom chip it's multiple sourced too.  Something you can't always say about the 'alternatives'

[jeremy]

Which authentication chip are you talking about?

There is a small micro with a usb programmer, but you can program a teensy using gcc and some sort of jtag adapter. Ultimately the boot loader is not on the main micro so that you don't lose code space, and can't brick your teensy.

[hamster_nz]

After a long time being passionate about Open Source Hardware, now all that comes to mind is this comic:

https://xkcd.com/1095/

I now let myself decide if any given product is "Open" enough for me rather than any community. Can I do what I am going to do with it? If so it is open enough, and any additional Open Source features are a (big) bonus to factor into my decision.

For me, there is nothing worse than an Open Source/Open Hardware Zealot who sits typing on their laptop about how the product of others are not open enough for them. Being so close to all that closed firmware, they must feel so dirty....

[charlespax]

After a long time being passionate about Open Source Hardware, now all that comes to mind is this comic:

https://xkcd.com/1095/

Now I want to Kickstart a DIY bendy straw kit; open source of course. Now I just need to figure out how to make it Bluetooth enabled and Arduino compatible.

[f4eru]

Now I want to Kickstart a DIY bendy straw kit; open source of course. Now I just need to figure out how to make it Bluetooth enabled and Arduino compatible.

http://dilbert.com/strip/2015-04-21

http://dilbert.com/series/70

[charlespax]

For the piece I'm writing I'll consider something to be open source if it meets the following criteria.

All necessary design files and embedded or other software required to make or operate the device are available in the public domain or under a license that permits modification and redistribution without discrimination against any persons, groups, or fields of endeavor.

I think the part not terribly well defined is what constitutes "the device." Maybe "the device" should just be the top level aggregate or parts. I make what I consider open source hardware, but I use microcontrollers that are not themselves open source. Even if I had all the instructions to build the microcontroller, would it be "open source" if I need proprietary equipment? And would I need instructions for making the chemicals, mining the raw components for the chemicals, how to build and operate the mining equipment, how to make the materials to build the equipment., etc.... it's turtles all the way down I'm afraid.

For the article I want to focus on how deeply open source a device is believed to be compared to how deeply open source it actually is.

[charlespax]

I think "public domain" has a specific legal interpretation that was not required to be open source.

You are correct; public domain is sufficient, but not necessary for something to be open source.

[helius]

You are correct; public domain is sufficient, but not necessary for something to be open source.

I don't think it's sufficient, either. You could say "I relinquish all interest in the source code" without publishing it to anyone.

[unitedatoms]

Strict Open Hardware terms are impossible for most FPGA boards - bitstreams "languages" are closed by chip manufacturers.

Graphic chips drivers for linux require closed blobs from chip manufacturers like NVidia. Anything with linux/android and fast screen is not true Open Source. Google for "PowerVR SGX Source Code" and see that 90% of computational power of linux (GPU > CPU) is closed. Example: Beaglebone based on TI Sitara has PowerVR GPU. The most entertaining part of linux (3D gaming capabilities) is closed.

If Open Hardware product contains USB, then most likely there is black boxed microcode being injected from PC driver into board every time it is connected. FTDI is an example. I do not expect that FTDI and other USB chip makers have anything to do with Open Hardware, I only pointing at any Open Hardware projects with USB.

Intel Quark: everything is Opened, documented and published, including ... some ~4K bytes binary blob in the middle of ROM. Who knows what is its purpose ? Strictly speaking there is no such thing as Intel based Open Hardware / Open Source products.

I, personally, do not feel that Open Hardware terms should be that strict. There is some pragmatic limit for openness. But the problem is that users buy, say, Beaglebone or Galileo / Edison board and think that they support Open Source movement with their money, when in reality it is just another closed source products.

For same PowerVR GPU reason, nearly all Android phones have closed source blob between the App and the user every time when user sees any 3D graphics.

With the arrival of SecureBoot on most product worthy chips (basically any contemporary phone, PC or tablet chip) every designer is automatically, by design, is required to sign the firmware. This step involves chip manufacturers to use their private key to allow firmware to run. It is very good feature and serves for end users protection. But also means that all SecureBoot products are in no way Open Source.

So if someone (who is not chip manufacturer) announces some Open product based on any complex chip, then many of claims about openness of code should be interpreted as "Open APIs with Source examples", but not "Open Source Code of every programmable part, which you actually can build from scratch".

TLDR: All FPGAs, all high-end CPUs, all GPUs, most USBs are closed on bitstream level, firmware level and at drivers level. They are may be opened only as APIs.

[EEVblog]

What about custom enclosure designs?
Take for example and Open Source multimeter where you got the schematics, PCB, firmware etc, but not the design files for the case. Is that open source enough to claim the title?
I think it is, but purists will almost certainly say no.

[unitedatoms]

What about custom enclosure designs?
Take for example and Open Source multimeter where you got the schematics, PCB, firmware etc, but not the design files for the case. Is that open source enough to claim the title?
I think it is, but purists will almost certainly say no.

I am not purist, so I will say it is Opened, even when enclosure is obscure. But one day, there will be a noble maker who will start carving a steel matrix for his 500 ton press to produce enclosures for people. Then community will change its mind and start demanding all CAD data for home made plastic extruded parts to be opened. Or else their 500 ton presses running on $10K molds is a waste.

[charlespax]

What about custom enclosure designs?
Take for example and Open Source multimeter where you got the schematics, PCB, firmware etc, but not the design files for the case. Is that open source enough to claim the title?
I think it is, but purists will almost certainly say no.

I wouldn't draw a line in the sand, but I'm happy as hell to raise the bar: https://github.com/PaxInstruments/t400-enclosure. See the current prototype in action.

[unitedatoms]

Grats on the milestone! That enclosure is fantastic. If you will make extruded parts done and mass produced, this will be a nice product by itself.

[zapta]

And the deal with the video drivers - yeah, but big deal ... What you are describing is the absolutist, the strongest possible definition of "open source" that pretty much nothing will ever satisfy, because any hardware you use is full of binary blobs with firmware, microcode, configuration, etc.

It does matters to me why some design aspects of blobs are opaque. Is it because the author just chose to have them closed (e.g. the Teensy case) or were there external constraints that prevented the author from sharing them (e.g. third party trade secrets). In the later case the author did a good faith attempt to open as much as he could which I appreciate.

[charlespax]

Grats on the milestone! That enclosure is fantastic. If you will make extruded parts done and mass produced, this will be a nice product by itself.

Thank you :-) My plan is to sell the LCD, keypad, enclosure, and a development board, so people can make their own project in a beautiful case. I will make a bare bones board with microcontroller, MicroSD slot, buttons, LCD, power system, battery charging, and all. There will be an empty section of the PCB where people can solder in one or more modules they purchase or design. It'll all fit in the case and be beautiful. Every day I wake up excited to make this happen.

[unitedatoms]

Is the width of plastic enclosure less or equal 103.5mm ? If yes, I have immediate use for it.

[ElektroQuark]

Wrong reading...

[charlespax]

Is the width of plastic enclosure less or equal 103.5mm ? If yes, I have immediate use for it.

Lucky you...

[unitedatoms]

That is great. I am lucky. I have steel enclosure 104x104mm. So the front of plastic instrument enclosure may work as panel mount bezel + screen.

[charlespax]

That is great. I am lucky. I have steel enclosure 104x104mm

Nice. I'm shipping probably in September. If you're okay with hacking the hardware a little you can buy one of my prototypes. I've just been putting them in the Mystery Bags. If you pick one up, just email me your order number and I'll make sure you get one that has a prototype, with LCD.

[unitedatoms]

Great. PM-ed you