Help with building Android tablet

Note: This was originally posted on 15th July 2012 at http://forums.arm.com

Hi there!

I have a school project to do, and I am really considering constructing my own Android tablet for that project. I'm aware that there are many difficulties with this. I'm planning to build something on the high-end for tablets today. This is what I'm planning to do:

I want it to be a 10.1 inch tablet, and I'll use Corning Gorilla Glass 2. I'd probably want to use an 1280 x 800 IPS display, since IPS is the only type of screen I know that is actually good. I'd also like it to be LED-backlit.
Optimally, I'd be looking for a thin, light profile, I think it should be 9.4mm thick (like the New iPad), if not thinner, and no heavier than 650g.
I'm not really giving much importance to cameras, but I think I'd like to have a 1.2MP front camera, and an 8MP rear camera, with an LED flash.
Under the hood is where most of my doubts lie. If I'd have to license an entire SoC, I'd probably be inclined towards NVIDIA Tegra 3, just because of its quad-core capabilities. But honestly, I was wondering whether I couldn't have a custom SoC built. In that case, I'd be willing to have an SoC with a dual-core ARM Cortex A-15, running at 1.5GHz. I'd also want dual-channel LPDDR2 or DDR3L memory controllers, although I'd prefer the DDR3L. For a GPU, I'd really like to have Imagination Technologies' PowerVR SGX543MP2 or MP3. I still have no idea about ISPs (Image Signal Processors) and video encoders/decoders.

I'm very skeptical about how can I take all of the internals and put them all in a chip. I also don't know how all these components go together to function. Any help is much appreciated. All I need to know is: is it actually possible? (by the way, I only have 9 months for this project)

Thanks in advance
  • Note: This was originally posted on 16th July 2012 at http://forums.arm.com

    Hello,
    Would you have a budget?
    9 months to get a custom SoC designed, tested, and produced seems kind of short to me.
    Cheers,
    Alban
  • Note: This was originally posted on 16th July 2012 at http://forums.arm.com


    Would you have a budget?
    9 months to get a custom SoC designed, tested, and produced seems kind of short to me.


    No, I don't have a specific budget, although my preference is as high-end as possible.
    If 9 months is too little for a custom SoC, do you think that I should buy an existing SoC? If so, I don't know whether I should get NVIDIA's Tegra 3, or the Qualcomm Snapdragon S4, although I do prefer Tegra 3.
    By the way, does someone know whether the Samsung Exynos is licensable? So far, I've only seen it in Samsung devices. I like the Exynos because of it's really powerful GPU, the Mali-400MP.
    Usually, I'd really prefer to have a custom SoC, but if 9 months isn't enough... an existing one will have to do
  • Note: This was originally posted on 16th July 2012 at http://forums.arm.com

    Hi again,

    In order to understand what is achievable, I feel I still need a rough idea of the amount of money you have available for this project.
    Is is $10, $100, $1000, $1 million...?

    Cheers, Alban
  • Note: This was originally posted on 17th July 2012 at http://forums.arm.com


    In order to understand what is achievable, I feel I still need a rough idea of the amount of money you have available for this project.
    Is is $10, $100, $1000, $1 million...?


    I've about $1000. Maybe a bit more, maybe a bit less
  • Note: This was originally posted on 17th July 2012 at http://forums.arm.com

    Creating a SoC requires to license, then integrate, test and manufacture. You would need more in the millions range than the thousand.
    This means you would have to take existing SoC.
    It might prove difficult or impossible to buy "one unit" of a brand new SoC.
    Today, you can buy some Computer on Modules which would simplify your design stage. You would still have to design in your display, all interfaces and the case.
    I think it is fair to say that it's a big job as once the electronics hardware is done, you will have to deal with software and the product case.
    Good luck and keep us posted!
    Kind regards,
    Alban
  • Note: This was originally posted on 23rd July 2012 at http://forums.arm.com

    Hmm - even building a custom multi-layer PCB for an off-the shelf SoC is likely to cost around $1000 if you are only getting a few made. In terms of what is achievable, remember that companies which make tablets from scratch employ hundreds of engineers, and even they buy in components from other companies, and it still takes years to develop each device.


    I agree with Alban, for a school project I'd pick one or two interesting aspects of a tablet to develop, rather than trying the whole thing. Failing that you can start looking at some pre-assembled mother boards - they are in the $100-200 range - and it means you have something which can run code as a starting point and then can worry about integrating the peripherals.

    HTH,
    Iso
  • Note: This was originally posted on 7th September 2012 at http://forums.arm.com

    lets say...
    You spend a small part of your fortune in designing hardware, get a board, and work on a part of it for 2-3 months. Then one day when you start working on the other part of the board, you find it not working, scratch your brains and after a lot of troubleshooting you realize a mistake in your hardware design. Bad luck!!! That will be the time you'll be very disappointed and feel frustrated. You'll have to start rectification process, waste of time and effort. You spend another part of your fortune in rectification. now this is waste of money too.

    I agree with Isogen74. Building everything from scratch is a tough job consumes a lot of time and effort. why not start working with what is readily available in the market and save a lot of developing effort, money and time. This will also mean that we have a assurity of the a working hardware.

    Unless you are planning to launch your phone in the market, is the whole process of self-building everything form scratch needed?

    Don't be too ambitious dude. Just start with what is readily and easily available. Develop something (write a generalized software for all the peripherals you use such that you don't have to worry even if the board is changed), and when successful in building what you desire (ready with the final product that you want), get your own everything. Finally LAUNCH IT IN THE MARKET (and sent one for me too ;) )
    Remember STARTING a project is VERY VERY important. Wasting time in getting started is not a wise move dear. All the best. I know that with the enthusiasm and energy you have, you will definitely design something worth and LEARN A LOT.
    Share your experience and keep us posted. Good luck.
More questions in this forum