How pervasive is the OSHW movement into the professional EE feild?

[george graves]

Allow me to assume, that if you're on this forum, you already know the names like Arduino, Adafruit, Sparkfun, and Hackaday.  Throw in Dangerous Prototypes, Bunny Hung, DIYDrones, and a few others (I'm missing a dozen)

The question is:  If you walked into an EE department at a large company, would the engineers there know those names?

[klr5205]

They certainly would at my company, although you can decide for yourself if a 15 person engineering department is large. 

Cheap ready-made boards are useful for making one-off test fixtures, and the dept is littered with Sparkfun FTDI breakout boards.

[EEVblog]

Likely not.
In my experience the majority of engineering colleagues have known or cared little about the hobby market and scene, which effectively is were OSHW is positioned.

[tszaboo]

Not really. Even if the engineer knows it, just because playing with it at home. Once I designed a calibration equipment, mad it "arduino compatible" meaning that it used a mega328 and an FTDI, and programmed it with the Arduino IDE. But I will never do that again, because all the ridicule I got for it from colleagues. Truth to be told, it is a one off board, and I wrote the code in two days, instead of spending a week on "software architecture planning  meeting pre-meeting" alone.
Now, I have a different job, small company, and my boss wants to integrate the C.H.I.P into our product . And you know what? It is a very stupid idea and I'm against it.
So, there must be a middle ground. One thing for sure, I will never rely my product on a hardware startup, kickstarter, Open source 3D printed community penguin, which cannot reliably supply components, or would force me to redesign everything every year.

[T3sl4co1l]

No one in industry has the slightest thought about OSHW.

OSS is pretty common among pure software companies, but the economics and practicalities are very different.  Namely, there is zero incremental cost of production, and having a thousand developers look over your code means free bugfixes.  Respectively, it really only works well for huge, popular projects, like web browsers, and tends to flounder with small, useful but uncommon projects.

Tim

[dannyf]

What OSHW?

If your livelihood depends on it, you are going to use quality tools that are well supported.

That's why you don't see what you don't see among the pros.

[rsjsouza]

IME mostly the younger folks that have an interest in electronics or are in technical marketing will quickly recognize these names.

[rstofer]

I would think that the hardware engineers are in the business of creating purpose-built hardware.  They certainly aren't in the business of creating the next great Arduino Shield.  That said, boards like the Arduino are great for proof-of-concept or even developing code while the product board is still in development.  Here I'm talking about driver code to, say, SPI peripherals.  The code will port easily to the product.

[zapta]

No one in industry has the slightest thought about OSHW.

OSS is pretty common among pure software companies, ...

It may be a generational thing. OSH is relatively new so it will take time for the generation that grew on sparkfun and adafruit to take control of the industry.

[mikeselectricstuff]

It's starting to very occasionally appear in some of the trade press,but only very recently.

[tszaboo]

Well there is also stuff, like Linduino, from linear tech, which is also backed that a bunch of their development board has libraries and sample code, which runs on their Arduino clone. Someone realized that there is a way to make the engineers work easier if you include OSHW in your toolset.

It may be a generational thing. OSH is relatively new so it will take time for the generation that grew on sparkfun and adafruit to take control of the industry.

I hope that will not happen. Same for the kickstarter generation. I kinda like finished products that work out of the box.

[nctnico]

Allow me to assume, that if you're on this forum, you already know the names like Arduino, Adafruit, Sparkfun, and Hackaday.  Throw in Dangerous Prototypes, Bunny Hung, DIYDrones, and a few others (I'm missing a dozen)

The question is:  If you walked into an EE department at a large company, would the engineers there know those names?

IMHO the names you mention are more targeted at people who want to whip up a working piece of hardware quickly (either for hobby or prototype/demonstrator). However there are also very high tech open hardware projects. CERN drives a couple of these on http://www.ohwr.org/ and they are definitely not for the faint of heart! I'm pretty sure there are more high-tech projects out there to share resources.

[b_force]

It all depends on the type of colleagues you have and how the company is running.
If you have workers who just do their jobs, but ones they are home they don't bother at all about it and wonder why the hell they actually have an EE degree, than you probably don't see much interest.

My personal experience, is that you see these people mainly in big companies (no offense).
In general big companies also have a pretty conservative point of view (even if they bring out new products) and mainly just one or two goals to focus on.

My other experience is that if you do meet people who are also interested on an hobby kind of level (like myself), they seem to be more involved and open.

That being said, I don't think you can really use Arduinos, Raspberry Pies, BeagleBone boards etc etc as professional finished products.
They are fine as a prototype solutions to find out if a concept is working.
As a finished product they simply lack certain (safety) features, robustness or are very bulky and expensive.

That doesn't mean you can still use their applications and typical ucontrollers/ic's
 

[Neilm]

Generally, I will  put OSHW stuff in test gear for boards that are too small or too simple to go on the full test equipment (for instance, a board that only has a few I2C devices).

I might use it for prototypes or proof of concept stuff but not into a product - it would be cheaper to design everything onto a single board than have to worry about connecting to a sub-board. Also, large companies might use a particular range of chips so these would be of preference as they benefit from economies of scale.

[AndyC_772]

It's always useful to be able to lash together a one-off of something for lab use. For that application, I don't doubt that if there's someone in the office who's already familiar with whatever this week's fashionable kind of 'duino happens to be, then it might serve a purpose. Same goes for RPi and every other similar product.

In terms of actual commercial products, though, the impact is negligible... as in, 'why are we even having this conversation?'.

I spend my professional life designing electronics, and for the most part, the end product is a board with a microcontroller, maybe an FPGA, and some kinds of transducers or other I/O. The fact that some sort of OSHW could possibly do some of the same things, could not possibly be less relevant.

No off-the-shelf board is going to be the right physical shape and size, or have the right interfaces, or be the most cost-effective option in quantities of hundreds or thousands. There are no 'duinoes that would survive 150 deg C or aerospace-grade vibration tests.

Bespoke hardware isn't just the incumbent way things are done today; it's a requirement that's not showing even the first sign of going away.

After all, you might enjoy making stuff out of Meccano - but you've never actually bought a thing (other than a Meccano set) that was actually made using it!

[magetoo]

It may be a generational thing. OSH is relatively new so it will take time for the generation that grew on sparkfun and adafruit to take control of the industry.

I hope that will not happen. Same for the kickstarter generation. I kinda like finished products that work out of the box.

There's nothing stopping people from opensourcing a polished product, is there?  What we're seeing now is just the result of having the first few successes be devboards and DIY 3D printers.

I might use it for prototypes or proof of concept stuff but not into a product - it would be cheaper to design everything onto a single board than have to worry about connecting to a sub-board.

I might misunderstand what you say here, but that sounds more like "the Arduino ecosystem" or building from existing parts and modules more generally.  What open source is supposed to do is give you everything that was used in the design, so you can modify it; maybe import the schematic, get rid of what you don't need, and start working from there.

[Smokey]

When you are making a product with a team and you aren't working by yourself out of your basement, you typically don't spend 80+ hours making an unsupported, poorly documented, half implemented, and stupidly "licensed" solution for something kinda work when you can spend $5k on the pro solution and be up and running in 1 hour.
If it's going to be something you are going to sell as a company, buying open source hardware doesn't make sense.  For just about any hardware solution you can buy from Sparkfun or whoever, a smart junior EE could make it from scratch for half the price or less.  It's mostly breakout boards and reference circuits.

[b_force]

When you are making a product with a team and you aren't working by yourself out of your basement, you typically don't spend 80+ hours making an unsupported, poorly documented, half implemented, and stupidly "licensed" solution for something kinda work when you can spend $5k on the pro solution and be up and running in 1 hour.
If it's going to be something you are going to sell as a company, buying open source hardware doesn't make sense.  For just about any hardware solution you can buy from Sparkfun or whoever, a smart junior EE could make it from scratch for half the price or less.  It's mostly breakout boards and reference circuits.

Completely agree with that!
Not to mention the fact that you will get a very sub-optimal solution because of all the limitation of these breakout boards and other things.

Although, you could still use a Atmega chip with the Arduino IDE, which is practically the same thing, you just don't use the standard Arduino board.

[zapta]

Although, you could still use a Atmega chip with the Arduino IDE, which is practically the same thing, you just don't use the standard Arduino board.

That's qualifies as a 'mixing' of an OSH project.

[kc8apf]

In my time at both Apple and Google, hobbyist dev boards have been quite prevalent for one-off designs and quick prototypes. With Google, I now work on server designs that will be given to the Open Compute Project as open designs (the schematics, gerbers, BOMs, firmware) will be publicly available under a fairly liberal license. It's not quite OSHWA level but it's a step in the right direction for big companies.

Sent from my Nexus 5X using Tapatalk

[b_force]

Although, you could still use a Atmega chip with the Arduino IDE, which is practically the same thing, you just don't use the standard Arduino board.

That's qualifies as a 'mixing' of an OSH project.

Well, it depends.
You could open source the entire project.
Or you could 'qualify' as a certain level of open source, like we talked about in this topic: https://www.eevblog.com/forum/oshw/oshwa-open-source-hardware-certification-version-1/25/
Correct me if i am wrong, but as far as I know, you could still use the Arduino IDE for non open source projects?

[exmadscientist]

There is also the issue that a lot of OSHW is just, well, kind of crappy.

At work right now I'm taking over a prototype system that was designed around an Arduino Due by another team. (I approved this approach because I didn't have any hours to spare at the time, and this let the bunch of non-EEs Get Stuff Done without any of our EEs diverting time from current projects. I now regret that signoff.) I have to get their benchtop prototype portable (as a portable system really does need a portable prototype), and it's been a major headache. I've decided to take the approach of designing a custom shield for several reasons. This is my first serious amount of time spent with an Arduino, and I'm really coming to hate the bloody thing. Just a few reasons why:

• The documentation is crap. There's nothing official out there that I could find on how to design a shield -- no mechanical drawings, no connector specs, no 3D models. I've got to rely on unofficial models, and what happens when they're weird? One of the headers is 5 mils off the grid of the others. Is that because someone did something stupid on the original design, or is it a mistake in the unofficial file supplied? Who knows! Schematics are provided, but they're poorly drawn (OK, that's hardly unique to OSHW) and partially in Italian! (Really, guys? Really?)
• They left reference designators off the board, which is always fun when the parts that are installed don't match the part numbers listed on the schematic.
• The board design is silly. A ring of female headers around the outside, plus male headers of two different pitches in the interior -- that's going to be fun to mate up with. And of course I can ignore some but not all of the headers, because power and signal lines are splayed out all over the place.
• How can I power this thing off a Li-Ion battery? I can't just supply it with 5V or 3.3V, since if I do and someone plugs in the USB cable, there will be a direct connection between two separate 5V supplies, and that never ends well. There's a FET in there which could isolate the two, but is tied off so that it does absolutely nothing a piece of wire couldn't do better. I can inject 9-12V on the VIN pin, but that's a single pin to supply the entire device, and suffers a diode drop. Still, it's the best option for this project, since I need to have a 12V supply anyway.
• Input protection is really not up to par. Dev boards get abused and should be protected accordingly; this one has some protection but not nearly at the level it should be. (Again, this is hardly unique to OSHW -- we've toasted thousands of dollars in commercial dev boards too, all because of stupid shit like missing diodes on trivially mismateable connectors.)

Yeah, sure, not all projects are this way, and I'm just a cranky engineer with too much to do and not enough time to do it. Join the club, right? But no way in hell am I signing off on any more projects involving Arduinos that have an expected time commitment of more than an afternoon. At the end of the day there are just flat-out better options.

[julian1]

Speaking as a amateur in hardware but professional in software, I love the open-source fpga toolchain IceStorm (Yosys, Arachne-pnr, and IceStorm) tools as alternatives to vendor toolchain's IDE's like ISE/vivado.

The reason is that I can make fit it into my workflow - in terms of build-system, tests, git code management, review process etc, rather than have vendor's product dominate these decisions.

Likewise for cpus - if I can apt-get install the cross gcc-compiler in Debian or emerge a Docker dev instance I am far more interested.

[NiHaoMike]

I have worked at a few companies that use Arduinos or Arduino derivatives in test setups. Arduino was chosen since it's easy to develop for. Why use something that would take a new engineer half a day to figure out if it wouldn't give any real advantage?

[Smokey]

...Why use something that would take a new engineer half a day to figure out if it wouldn't give any real advantage?

Because projects don't end after that half day Arduino project hack.  The further you get through a project, the harder it is to change directions.  You end up like exmadscientist, stuck with some Arduino thing and having to work around the hardware choices and limitations instead of getting exactly what you need because you designed it.