Hardware for Android Device

that's an ambitious project, I've just finished 2 degrees in EE and in the long term i'm looking to do similiar projects, but right now it is beyond my capabilities. But what i have done is buy a very cheap dev kit from STmicroelectronics with their ARM m4 chip onboard. (STM32F4)
this chip should be powerful enough to get started on and all the pins are broken out, plus the device includes a programmer and is powered over usb.

It was less than €20 but is still sat in its box as I've a lot to learn before cracking it open.

Have you any experience with RTOS for ARM, Keil offer a free trial version of their well respected uVision MDK software, it supports the above board directly and removes the need to configure a tool chain etc. Personally i'm trying to get eclipse on ubuntu to program it bit Keil uVision will allow me to blink LED's etc so long as my program is under 4Kb.

I too am only starting down this project but i hope the little i know has been of some help.

As for custom devices, well thats a whole other ball game, you will need to make out a schematic, then a board layout, then gerber files. After that you need a small run on a pick and place / reflow line. It's very rare these work out first time round, attention has to be paid to details like noise sinking, pull up resistors, matching logic levels and optically isolating external devices etc.

It's great that you are looking beyond your course material, I've learned much more from personal geekery rather than just taking notes from a lecturer. Anything you do outside the course will benefit you in a better degree at the end.

I've never been designing device from scratch, and I'm also just first grade student. Anyway I could imagine how this might look for small company or single person:
1) Decide what do you want to build-up. Easiest todo is custom dev-board, it can be always redesigned and packed into tablet case. The hardest to-do is mobile phone, and it's nearly impossible to create such thing due to high level of embedding everything, and need to sign pretty serious agreements with RF CPU (and other things like transceivers, antennas, duplexers) supplier like Infineon or Qualcomm.
2) Think what main components you'll need, like LPDDR, SoC (CPU), PMIC (SoC manufacturer usually recommend PMICs to be used and provide reference board schemas for using both), battery fuel gauge, charging controller (both might be built into PMIC, depends on model), screen+touchscreen (there are dozens of such, one might want to decide its size already, but in case of dev-board like build it usually can be replaced by some smaller/bigger with small HW modifications or without modifications at all), sensors like gyro, compass, pressure, light, whatever.
3) Search through suppliers websites and decide what models of ICs you want to use (I'd pick only open hardware), order engineering samples and get reference schemas, rather start from SoC(OMAP4460 for eg.)+PMIC pair, then decide about the rest.
4) Don't forget about extension slots like USB ports, DC supply, serial converters, whatsoever.
5) Start designing PCB board. IMO it's impossible for begginer to project any usable PCB for embedded system, I'm begginer and I'm failing with simplest boost HF DC/DC converters (like 10-20 parts on board), while such board would have thousands of elements on it, and multi layer board to fit it everything in some rational size.
6) Find company that will make prototype for you - they should make board + solder all the components you provide them - one with no professional (and very, very expensive) soldering stations is not able to solder BGA components at home.
7) Test it out.

Relatively, assuming that main components are free engineering samples, this might be not so money-expensive way to create some useful stuff. But for sure it's very, very time expensive, and begginer alone will nearly for sure fail.

//edit:
I just re-read my post and figured it might be pretty demotivating. I'm not saying it's impossible, but I'd suggest you to start from something easier - ARM dev board is the thing you need. As Quiggers stated above.

Just noticed these - cheap and powerful dev boards:
http://wiki.xbmc.org/index.php?title=Allwinner_A10#Other

Custom design

I'm looking to do the same, has this worked for you? I'm looking to build a custom android based mobile device as the original poster. I haven't had any look finding the correct electrical or device engineer to provide me any assistance. Are you available to assist?

Quiggers said:

that's an ambitious project, I've just finished 2 degrees in EE and in the long term i'm looking to do similiar projects, but right now it is beyond my capabilities. But what i have done is buy a very cheap dev kit from STmicroelectronics with their ARM m4 chip onboard. (STM32F4)
this chip should be powerful enough to get started on and all the pins are broken out, plus the device includes a programmer and is powered over usb.

It was less than €20 but is still sat in its box as I've a lot to learn before cracking it open.

Have you any experience with RTOS for ARM, Keil offer a free trial version of their well respected uVision MDK software, it supports the above board directly and removes the need to configure a tool chain etc. Personally i'm trying to get eclipse on ubuntu to program it bit Keil uVision will allow me to blink LED's etc so long as my program is under 4Kb.

I too am only starting down this project but i hope the little i know has been of some help.

As for custom devices, well thats a whole other ball game, you will need to make out a schematic, then a board layout, then gerber files. After that you need a small run on a pick and place / reflow line. It's very rare these work out first time round, attention has to be paid to details like noise sinking, pull up resistors, matching logic levels and optically isolating external devices etc.

It's great that you are looking beyond your course material, I've learned much more from personal geekery rather than just taking notes from a lecturer. Anything you do outside the course will benefit you in a better degree at the end.

Click to expand...

Click to collapse

Technexion

I have used OMAP3530 CPU. The TAO3530 is a good starting point and you can get a Tsunami board.

s8500 board with tablet touchscreen

hi dudes,

i have an old wave s8500 but the screen is broken. and i have an old tablet screen 7" from herotab8/dropad8.
can i use the tablet screen with the s8500 board? is not drivers necessary for the touchscreen? and where will i get the drivers?
and do i not need the datasheets of the pins to connect?
)

What we REALLY need is for someone to make a SoC that's basically like the one in the Raspberry Pi, but substitutes a FPGA for the GPU that's big enough to re-implement GPU functionality... long after the chip has left the fab & gotten soldered onto an open-ended generic ARM stick with no specific purpose, and thus manages to officially avoid getting infected by DRM-mandated licensing terms (ie, anything *officially* licensed to support h.264 or HDMI) that keep making totally open drivers nearly impossible. After all, if the drivers were 100% open source, there's no way they can stop you from commenting out the part responsible for implementing Cinavia, or lying to endpoint devices (like your home theater amp) about HDCP compliance

To deflect infringement claims, a company that made Android boards from the FPGA-equipped SoCs could make it with a soldered-on DVI port instead of HDMI (HDMI connectors are encumbered by viral licensing, DVI isn't), and put a reference design on their website for a wacky octopus cable that used the DVI-A pins to output unbuffered 3-bit pseudo-VGA, and used the remaining pins as a high-density breakout connector for a bunch of half-duplex RS-485 ports and GPIO lines that just *happened* to use DVI/HDMI logic levels

Of course, you'd never be able to legally sell a product based upon that board to end users in the US with the taboo technologies supported "out of the box", but other companies outside the US not subject to our self-inflicted wackiness could, and hopefully WOULD, buy enough of those boards to drive the price down enough to make them cheap for American hobbyists to buy on eBay and use for our own guerrilla Android-powered hardware projects.

In theory, the Xilinx Zynq 7000 series sort of does this... but at the moment, they're so ungodly expensive, you could almost buy a half-dozen Nexus 7 tablets for the price of their Android-capable dev board.

sounds great dude

Nice

Nice post

Hardware for Android D

Its not even turning on now...guess i will have to take it to a computer shop now, are you sure it has to be major things like "dead hard drive to a burned up chip to a bad motherboard."?