[continued from previous message]   
      
   Let's say an OS boots. In the process of handoff, the BIOS says "we need   
   to engage the four second filter now, on the front panel button, to   
   prevent inadvertent brushing against the switch". If you press the   
   front button, the button must be held down for the four second interval   
   (because the OS or something, set the hardware to start using the logic   
   filter).   
      
   In an OS, you can "bind" the front button input to a response. You can   
   make the system "Sleep" by pushing the front button, or "Hibernate". But   
   most people just leave the ACPI interface set to "When I press the button,   
   shut off". So after the four seconds, the hardware logic shuts off.   
   That should be the default as far as I know.   
      
   If the pad driver that asserts "PS_ON#" on the 24 pin was able   
   to switch the system ON, it should be able to switch the system OFF   
   by "letting go" of the pin. The 8mA open collector signal, stops   
   pulling down on PS_ON# and the pullup resistor on the PSU end,   
   pulls the signal to the deasserted state (the five volt level)   
   and the fans go off.   
      
   The +5VSB is still running at shutdown, but the main rails are   
   off and the fans should have stopped (because the fans depend on   
   the main rail +12V for power).   
      
   Off for your RMA I guess. I can't see how it would fail that way. It   
   would have needed a logic clock to run the state machine (something   
   it can use to time a 4 second event in hardware). The RTC (Real Time Clock)   
   has a 1PPS output (pulse per second) and counting four of those ticks   
   should suffice as a logic conditioner for the momentary power button   
   in the front of the PC.   
      
      Paul   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)