17 Feb 2017: First revision EOMA68-RK3288 CPU Card PCB assembled
The components have all been placed onto 5 PCBs, with the exception of the RK3288, RAM ICs and eMMC. Two of the PCBs have been fully assembled with an RK3288 and the 4 DDR3 RAM ICs. The first PCB was not successful in being recognised on the USB-OTG bus. Power was all okay, but there was no activity on either the eMMC or MicroSD clock line during power-up, and the USB-OTG was not responding. Turns out that the OTG port was mis-labelled as the 2nd USB (Host) and vice-versa, so tomorrow's task will be to try using the correct port before proceeding further. Also the PCMCIA break-out board arrived so it has been possible to test using that.
Assembling these PCBs has been exciting as it's been done entirely using equipment in Taiwan in a "home office" environment. A small commercial pick-and-place machine was used, which is due to be replaced by an OpenPNP design. The cost of 10 8-layer PCBs here in Taiwan was only $USD 600, which is about 20% less than the cost in China. Amaazingly, using a small IR oven, the 600-pin 0.6mm pitch BGA RK3288 IC was successfully soldered on. Even the tiny 20-pin 0.5mm pitch 1.8mm x 2.0mm 3A PMICs were successful. It is a fascinating process to watch and experience. Photos to be included in this update, later.
19Feb2017: first board, memory isn't initialising at full speed: setting DDR3 clock rate down to 200mhz successfully completes DDR3 initialisation: moving on to mmc initialisation that fails spuriously which generally indicates memory unreliability. attempting the bootup with a second board.