These who follow our blog know that in November we learned for the very new A10S processor from Allwinner http://olimex.wordpress.com/2012/11/26/new-exciting-processors-from-allwinner-a10s-cortex-a8-with-hdmi-and-a4x-cortex-a7-4-cores/ and start working on board with it.
A10S is like A13 but have also HDMI and 100 Mbit Ethernet, which makes it perfect for small Linux computer.
In December we routed the boards, but as we had already problems with A10 prototypes this time we didn’t want to give chance for errors so we routed two revisions – Revision A with all components on top and Revision B with some decoupling capacitors on bottom for better de-coupling.
The blank PCBs were produced in January and we assembled them. There are some silly mistakes we did and we will fix in Revision.C but generally boards functions and works as expected.
Dimitar Gamishev had to fight with several issues on the software side. Needless to say Allwinner A10S SDK didn’t generate working image, when I asked Allwinner about it they said they sent me version 1.0 which have bugs and now they have version 1.4 which have no bugs :) we are waiting their new SDK to arrive, but meantime Dimitar had his hands itching to play with the board so he start to implement the linux support with what he had.
1. the PMU was changed to AXP152 instead of AXP209 this is reduced version with less voltage outputs but generally (with attempts and errors we found on which address it responds and fortunately most of the commands for AXP209 worked for AXP152 too, just on differnt address and with one additional register)
2. Ethernet is not supported in original Allwinner SDK as they target Android TV dongles which have WIFI, First our contact was uncertain if the Ethernet is implement at all, but I told them that it’s in the datasheet, so they asked the right person at Allwinner and confirmed that there is Ethernet but they never used it :). Dimitar took A10 Ethernet driver and after playing a while the Ethernet got alive.
3. HDMI and Audio were also implemented.
Dimitar did this for few days with the the first prototype. Sure the Linux support is far away from perfect but at least we could validate the design.
Here is the console log http://pastie.org/5735893
What’s next? We have to fix all silly mistakes we saw on the first prototype – like console UART shared on same ports as the second SD card :)
We wanted to make rock bottom low cost boards but now we re-consider that having NAND on board is good if you want to run Android 4.1 (yes Allwinner have Jelly Bean SDK for A10S which we wait to arrive)
Then we will run initial small batch form of Developer Edition and we will spread among the developers for additional testing before we proceed with mass production.
The Revision A and B CAD files are on GitHub, the Linux software support is to be updated to linux-sunxi repository.