Phi Go

So let's see what's under the shield! I had to remove the shield, the SIM and SD connectors with my hot air station:

And here's the PCB in all its glory:

The Mediatek chipset requires a very small amount of passive components to work, so the PCB design is quite simple, as it's just a matter of connecting the chipset to the data bus and take care of the antenna connector coupling. Here's the basic public block diagram for the chipset:

Chinese phones are clearly designed by very practical engineers that value function over design. When was the last time you opened an iPhone to find a clearly labeled SPI port?

But wait, there's more!

A nice labeled serial port on the back of the PCB! Let's solder some wires and see what's there...

Some interesting serial data at 921600 bps! Boot logs here: http://pastebin.com/MMXR4y4S

It is extremely important to forecast all possible problems you may encounter and their solutions if you want to make a successful plan. Here are the two biggest problems we will have: 

- Mediatek has a bad reputation on the developer community due to its continuous violations of the GPL license. Even a reputable manufacturer like Asus did not release the correct kernel sources for their Mediatek-based tablet a while ago. At the last MWC through, they announced a developer program and said that they will release SDKs and HDKs for their products, so things should change soon.

- FCC (and more) certification. Certifying a mobile phone is quite hard and expensive, but I should be able to get around this by releasing the board as a component and not as a finished product. You will have to get the certification yourself if you plan to sell it. I will consider getting the boards certified if I get enough support from the crowdfounding campaign through.

These are not huge problems, and they should be easly solvable with some communication and a good design to start with.

It’s been almost ten years from the enormous success of the Linksys WRT54G hacking, bringing an affordable embedded Linux system to the masses. Since then, a huge amount of incredible hacks came to light thanks to the popularity of cheap Linux routers, like the TP-link TL-WR703N, my favorite.

Then the Cambridge University came along and designed the Raspberry Pi, and this was the first cheap mainstream single board computer to hit the market, followed closely by a plethora of other options like the BeagleBone, the pcDuino, the Cubieboard and many other.

All these platforms offer good processing power and a bunch of GPIOs and peripherals, but I want more.

40 dollars at the dodgy cell phone market in Shenzhen can buy you this really nice phone, in a retail box with included headphones, charger and two batteries. This phones core is a Spreadtrum SoC, with a dual core Cortex A5 running at 1.2GHz, GSM, GPRS and WiFi connectivity, a big screen with capacitive touch and two cameras that may not excite a photographer but seem good enough for some serious OpenCV trickery.

So, my idea is to make a development platform out of cheap Chinese SOCs! Mediatek produces a nice range of chips with interesting features, with CPUs that can range from a dual core Cortex A5 to an octacore with four Cortex A17 and four Cortex A7. These chips are quite cool, because they’re designed to include a lot of functionality in a single package, so my board will have a lot of features that are not available on your average raspberry:

• An ample set of connectivity standards: WiFi, GSM GPRS and 3G (maybe LTE), Bluetooth, GPS, ANT+
• Very low power consumption, even while being connected to the network waiting for notifications
• Lithium-ion battery charger and manager
• Expansion slots for SDIO peripherals, and a bunch of SPI, UART, I2C ports and GPIOs
• USB host and device
• Native dual camera support for up to 20 megapixel sensor and up to full HD video
• Support for cheap LCD screens and capacitive touch, plus HDMI output

I will also design a cheaper, less powerful board for smaller applications.

All this goodness will enable a lot of interesting projects, for example the first thing that comes to my mind is a completely standalone home automation and alarm system, interfacing with e.g. an ANT sensor network and using the GSM connection when the WiFi is not available to alert the user.

But the first project that I personally would like to build with this platform would be a port of the Open Vehicle Monitoring System (http://www.openvehicles.com/), a small device that you can connect to your electric car’s diagnostic port and allows you to remotely check the battery state of charge and do all sorts of cool stuff like turning the AC on before you get to the car or track it with the embedded GPS.

Hold on for the next article where I'll talk about the problems that I will have to deal with and more!