Recently my wife’s EeePC 1000H broke. It wouldn’t turn on properly. The LEDs lit up but there was no other activity whatsoever – no backlight, no fan, no harddrive. You couldn’t even switch if off again, not even by holding down the power button.
The obvious first aid attempts – removing AC power + battery, resetting CMOS RAM, pushing the hard reset button – were to no avail. So the netbook had to be cracked open.
Once inside I noticed there was no power to most of the subsystems – only the standby voltages were present to a number of components. One of the main power supply ICs was completely without power. Backtracing the leads I discovered a MOSFET switch, controlled by the system’s EC (embedded controller), that was responsible for activating the main power supply. The EC just wouldn’t turn it on for some reason.
The embedded controller of the 1000H is an ENE KB3310. It is based on an 8051 core and runs directly from a connected SPI flash that contains the BIOS (+EC firmware). Hooking up a logic analyzer and comparing the transferred addresses+data with a downloaded BIOS image, it looked like the controller was running fine, albeit in a loop waiting for something.
After some more backtracing and studying flow charts from a different EeePC datasheet I learned that the EC would first boot up and then wait for the (physical) power button to be pressed. After that, it would send an electrical power button signal to the southbridge and wait for the southbridge to request power state S0 (power on) from the EC by pulling its S4 and S5 lines low. Which never happened. The EC was waiting forever.
Turns out the southbridge was fried. At that point I hit a dead end and replaced the main board. (Replace the southbridge? I’m crazy but not THAT crazy.) Took me three attempts on ebay to get a working main board too!
So the little white netbook works again and my wife was more than happy about that. From a tech mojo and economical point of view I consider this experiment a failure though. Oh well. :)