$25 PC alpha board successfully runs Linux
geek.com
geek.com
I'm going to automate my new greenhouse with two Ardiunos + some single-board, low power computer like this one.
Arduinos will drive motors (window openers, fans etc), analyze data from temperature/humidity sensors in real time, i.e. "convert" USB port to a bunch of 5-volt analog and digital signaling lines. 8-bit realtime architecture is well-suited for that purpose.
32-bit ARM-based computer will drive Arduinos via USB, keep logs on SD card, act as http server, keep me informed via email, maybe even tweet, and also download weather forecasts from the Internet. New code will be uploaded to it via ftp. 32-bit multitasking architecture is well-suited for that purpose.
I was looking at different ARM-based single-board computers recently, all of them are with ~$150 price tag. This one seems much cheaper. I'm not sure about its aptitude as a cheap computer for third world countries, but it is indeed usable as "next-level Arduino". Because 8-bit realtime and 32-bit multitasking architectures ain't mixable anyway.
8-bit realtime and 32-bit multitasking architectures ain't mixable anyway.
I'm working on an ARM board (http://rascalmicro.com) that plugs directly into an Arduino. If you're serious about building this system for your greenhouse in the next few months, I'd love to have you as a beta tester. My email is in my profile if you're interested. If not, good luck with the vegetation.
One of my beta customers has a demo up at http://angerlights.com that is pretty amusing and demonstrates the Rascal's basic functions. All the code is on Github too: https://github.com/rascalmicro
If you want to try a Rascal in person, and you happen to be in New York, I'll be doing a demo at the Open Hardware Summit in NYC in mid-September.
$35.00
You can grab it for $50.00 off the buy.com website.
Ditch the Ethernet, add Wifi, GPS and a mic jack (and maybe a riser for a GSM/GPRS board?) and I would throw away my Android phone. A little eye-mounted analog HUD and an earpiece with attached mic would be all I need, and speech to text could do simple commands and transcription for me with maybe a micro-querty keyboard I could unclip and type on when necessary. Call me crazy but I think it would be cheaper than a phone and extremely practical for hands-off interaction. Plus, plug it into a monitor and keyboard and use it as a PC.
And the price would no longer be even close to $25
With only two, after you've added a keyboard and mouse, you're already out of expansion options for the desktop pc use-case. (Of course you can add easily a hub, but I just have a feeling that you really shouldn't have to.)
* you'll need to bind the mouse and keyboard to the same receiver on a windows machine before you connect it to your linux box until linux unifying support is available.
"We plan to develop, manufacture and distribute an ultra-low-cost computer, for use in teaching computer programming to children. We expect this computer to have many other applications both in the developed and the developing world.
Our first product is about the size of a credit card, and is designed to plug into a TV or be combined with a touch screen for a low cost tablet. The expected price is $25 for a fully-configured system."
Old Android versions on old phone hardware used to have a slowish/imprecise keyboard, but more recent versions on a fast CPU are much better.
I'm a gadget lover, wouldn't want to give up my Android (and mourn the braindead WebOS and MeeGo kill-offs), but a feature phone Nokia beats a touchpad in every way if we're talking text messages.
You can type blindly (you might need some time to know that a certain combination always suggests something wrong first, the real word is N alternatives away). You have good tactile feedback. And - the prediction is far better than what I've got on WebOS and Android as well..
For something fancier you want a Panda Board, Beagle Board, or Gumstix which are priced from $120 - $250.
Also, most people can achieve a higher WPM on Swype than T9. I know it's a little more frustrating and trickier to text blind, but it's also way faster.
The picture in this article looks to be a Broadcom evaulation board for the processor they're using (look at the silkscreen). That ethernet jack with integrated magnetics is $4 alone. There's no way this board is $25.
The actual board looks like this: http://www.geek.com/wp-content/uploads/2011/07/raspberry2.jp...
Looking through the media section says that the picture you have is a prototype board, not the manufactured alpha board they're talking about here.
the $25 price point is out of the question unless: A) They sell at or VERY near cost B) They sell advertising on the PCB C) They are subsidized by a tech vendor selling them parts at a discount D) They are subsidized by a charity or an angel E) They are planning on making 100,000 units in one shot.
At least SEVERAL of the above have to be true, or the $25 is just made-up nonsense.
Add in the passives and other junk, and leave out the ethernet module and we are talking about a bill of materials of close to $20.
Now add 1.75 for the PCB (qty 5000), 2.50$ for assembly, another $2.00 for testing, packaging and assorted costs (like shipping raw materials and finished goods to and from china, since US assembly and PCB costs are at least 3 times what I noted) and the bare per-unit cost is $25.
Dont forget to factor in costs like - uhh - keeping the lights on in their officepartment - and even if their dev time is free they just cant sell that product for $25 without outside cash subsidizing the whole thing.
In this business, distributors pretty much demand the product at 25-40% discount, so if they want to be in shops and catalogs, you have to factor that in as well.
Consider the raw cost per arduino is probably around 10.50$, they sell 100,000 units a year, and their onboard processor is about 1$ and they have less overall plastic, passives, and other overhead parts than this thing.
Given that Eben Upton is currently a technical director at Broadcom while simultaneously working on this Raspberry Pi project...maybe I'm not so shocked anymore. It's actually refreshing to see chipmakers put out cheaper EVKs to play with instead of $4000 development systems. It's good marketing.
I'd single out TI as the largest culprit of making EVKs hard to get, but I'm also inclined to believe they've been giving Beagleboard/Pandaboard a helping hand (like where can you find loose OMAP4s on the market? I sure can't.)
[1] https://estore.ti.com/MSP-EXP430G2-MSP430-LaunchPad-Value-Li...
Note that the OMAPs are trickier as they expose their memory bus above the chip - the SDRAM and NAND are mounted on top of the main die as a stacked BGA. This isn't easy to work with.
Really? They are 40 cents, in quantities of 1. From Jameco, who are not known for their low prices. [1]
Your $4 for headers is bunk of a similar magnitude.
> I cant find an ARM11 chip for less than 10$ on octopart
This is true, but 500MHz Arm11 chips are only $11 in reasonably small numbers. [2]
I wish I was your supplier - you are putting their kids through college.
[1] http://www.jameco.com/webapp/wcs/stores/servlet/Product_1000...
[2] http://search.digikey.com/scripts/dksearch/dksus.dll?Detail&...
Without breaking down everything piece by piece, we spend about 3.00$ in plastic and other headers on a medium qty design that looks about comparable to what theyre doing (i/o wise). Also, female 1/10" headers are quite a bit more expensive than male, which I think skewed my intuition a bit.
Package all those materials together, and you end up with something along the lines of a database.
I spend around that much every time I stock up on local produce, bison steaks, the latest micro-brew, etc. - why not hardware too?
In fact, I would appreciate being able to head down to some future kinkos and print out a replacement cell phone case, even if I could have one for pennies on the dollar if it was shipped over from Malaysia. I'm not (just) being an eco-hippie either.
but I will just ignore the bit about mining.
For a computer like this, you simply don't gain a lot. It's mostly prefab components, with a bit of PCB and soldering. No big need for made-to-order, built from scratch items. You'd get almost nothing out of this, apart from a "support your local automated shop" deal. The eco impact of transport for electronics is negligible.
Now, I'm not saying that there isn't a spot for "local electronics", but that would involve other devices, ones that are highly experimental, customizable and/or short run. If you'd get things like this (or an embedded systems) with oodles of different connectors, memory options, different processor subtypes, or open source hardware (probably FPGA based, to make it more reproducible). I've been dreaming about stuff like that since I read about "nano-facs" in the Cyberpunk RPG…
But again, I don't see the big deal for the Raspberry. It makes sense to print a sole copy of "Best fly fishing sites for Llarregub Creek in late autumn" in your Kinko's, not to get a copy of the Bible or Harry Potter.
It is just a guess on my part, but I assume a custom fabricator might appreciate being able to produce a moderate, steady volume of commodity parts, based on open hardware specs (if they command a premium by virtue of being 'local'). And that same fab would then be more likely to be able to offer a custom model with, for instance, an integrated FPGA.
I wouldn't expect many people to pay $6K when they could pay $3K, but $150 vs. $60? I think there's room for that.
And whoever owns all that big expensive fabrication hardware has to be able to keep the lights on. (Yeah, a lot of tools have come down in price, but most of them don't manage the same level of quality of output and now you're asking them to pay multiples of what they could get elsewhere, for lesser-quality stuff.)
What you'd need for truly "local" production is some kind of cheap shortcut. 3D printing uses this "bits of molten plastic" process, which isn't exactly industrial injection moulding, but workable enough. Book print-on-demand doesn't use offset, it uses relatively cheap and crappy laser and inkjet printing (and e.g. no hardcover as of yet). I can't think of any universally acceptable equivalent for electronics (common processor or FCPGA only work for a certain subset), and the plethora of different connections are beyond any sane automatic manufacturing concept. In addition, this would to be highly automated to be profitable, laden with patents and still quite huge. So it's quite likely that this will be run by a international chain, which kinda makes the "support your local electro-plutocrat" argument moot. Haven't seen someone calling for support to their local McDonalds…
This needs some really big manufacturing breakthroughs to properly work, and given those, the whole world would look quite different.
And we haven't even talked about regulations…
Interesting idea, but impossible for most people today.
Bringing some manufacturing capacity back to the regional level is exactly my objective. Specifically, product runs on the order of 5000 units, with a modest bill of materials, using open specs that lend themselves to lower cost redesign and re-purposing.
General purpose, trivially priced computing devices based on open source software and hardware... There is no reason to rely exclusively on established logistical chains at this point, and I think it would be beneficial to establish new (regional) ones.
Seriously though, throw WiFi in there and lock the output down to 720p, and you've got a $99 AppleTV sans software.
My Arduino's cool and all, but the extra power you'd get from this for the same price... Wow.
http://www.sparkfun.com/products/9521
but then you're back again at arduino, with the SerIO having a higher price tag than the device itself.
For twice the price of an arduino, the $25pc+SerIO could still be a really nice way to create some entertainment.
In fact, since it has audio out and SD card, it would already be cheaper than another project I was working on. lol
The Arduino would probably boot faster, though, so that's a consideration as well.
There are approximately 16 spare GPIOs, which are brought out to 1.27mm pin-strip. Voltage levels are 3v3. The connector choice is deliberately annoying to connect to directly; there is no over-voltage protection on the board so the intention is that people interested in serious interfacing will use an external board with buffers, level conversion and analog I/O rather than soldering directly onto the main board.
We also bring 2x I2C (3v3), I2S and an SPI (3v3) interface out to the same connector. We support one slave interface for I2C and one for SPI.
Edit: I'm not sure how easy it'd be to wire up sensors and such to USB. Don't some of Apple's ear buds send commands over the headphone jack? Does that require a special jack, or could it be done easily with normal hardware?
Edit 2: Would the GPIOs work for this? [3]
1. http://www.raspberrypi.org/?page_id=8
2. http://www.raspberrypi.org/?age_id=43&mingleforumaction=...
3. http://elinux.org/RaspberryPiBoard#General_Purpose_Input.2FO...
Many people have tried to make ARM versions of Arduino, usually with Cortex-M3 microcontrollers. None seem to have been very popular. Getting more power usually means more cost and always means more complexity; there's no way around it.
There are some other products out there running .NET runtimes and basic real time OSs as well, but these have not really caught on.
Add an operating system into the mix, and as you said its a whole different ball game. That said, 700MHz is a lot of room to try and add a real-time Arduino style layer with access to GPIOs and such. You probably wont get 1MHz real time, but you can probably get 100KHz real time, which is fine for MOST things. Plus you get a network stack, and a display driver, etc. etc. to boot. Of course it cant be done for $25 (that is silly unless the project is nonprofit AND subsidized with charity or advertising dollars), but I would believe $70.
The software for such a device would not be anywhere near trivial to get right, but there is a neat product in there somewhere that a team of 3 or 4 could probably get going rather swiftly.
Of course, on a Maple or an FPGA you can really get crazy - with output timings accurate into the 36MHz and 150MHz range (respectively).