Overview: We have roughly 100 of these Via Embedded manufactured boards, and 60 of them (give or take a few) exhibited a rather odd power up problem.
Symptoms: after the board is powered off for roughly 10 seconds or more,
*and* +5SB is still applied through the ATX connector, (a very common condition) the power on button connected to the front panel connector would power up the board, but there are no POST codes to port 80. No #CS or #OE occurs to the flash BIOS which would signal a code fetch after reset. In other words, the board is completely and utterly dead. Removing power completely from the power supply would allow the board to recover, *or* a power up, power down and then a quick power up would also recover the CPU board. Additionally, when the board is dead, pressing the reset button 10 or 20 times would also recover the CPU.Using a 'scope, I noticed that the "high" signal levels on the two
14.318MHz Oscillators on U19 pins 10 and 12 were of rather marginal voltage levels, barely making it above 2 volts. (and even then, only for about 6ns or so...)The pullup termination is currently 100 ohms to 3.3VDC. Changing the two resistors to 75 ohms has improved the signal levels and has fixed the problem on 3 boards that I've tried so far. I just found the fix today after struggling for several days. Now to repair the rest... The two resistors are located right next to U19 (a 14 pin chip marked LVC07 and very close to the DIMM and J13), and the resistors connect to pins 12 and
10 of the chip. The device is an open collector non-inverting buffer similar to a 7407. The resistors are right next to the chip pins as well. Doing a little tracing, I believe that one of the oscillators drives the southbridge, but I was unable to trace the other. Perhaps the northbridge?Anyways, get out your magnifying glass! If you don't like the idea of replacing the parts, you could also solder 300 ohm SMD resistors right on top of the existing ones for a parallel resistance of 75. Good luck!
[Please remove mybrain to reply by email]