There are a few electrolytics on the IIfx logic board contrary to popular belief.
Let's have a look at the schematic, courtesy of Gamba, and discuss the symptoms and possible causes.
From your observation, the button on the Mac and the button on the keyboard act differently. This is a good clue.
As you can see if you look for a bit, the top group of NAND gates are controlled by both power buttons, and it is responsible for turning the Mac ON using Q3. The bottom group of NANDs is NOT controlled by the keyboard. It is controlled only by the button on the Mac and a software shutdown and uses Q2. Something is triggering this section of the circuit too easily.
We must not forget our clue from before. What would make the Mac's power cycle around only from the power button on the back of the Mac?
Resistor R110 is intended to make sure that when both the "ON" nands and the "OFF" nands are going at the same time, so both Q2 and Q3 are turned on, that the "ON"s always win. Somehow the "ON"s are not winning.
I therefore suspect that transistor Q3 is to blame for the power fluttering. But for the power cutting out unless you hold the power button for a length of time, I suspect the bottom group of NANDs. These are possible symptoms of leakage from the small radial electrolytic cap that's in the area. Mine leaked.
-- Recommendations --
Let's have a look at the schematic, courtesy of Gamba, and discuss the symptoms and possible causes.
From your observation, the button on the Mac and the button on the keyboard act differently. This is a good clue.
As you can see if you look for a bit, the top group of NAND gates are controlled by both power buttons, and it is responsible for turning the Mac ON using Q3. The bottom group of NANDs is NOT controlled by the keyboard. It is controlled only by the button on the Mac and a software shutdown and uses Q2. Something is triggering this section of the circuit too easily.
We must not forget our clue from before. What would make the Mac's power cycle around only from the power button on the back of the Mac?
Resistor R110 is intended to make sure that when both the "ON" nands and the "OFF" nands are going at the same time, so both Q2 and Q3 are turned on, that the "ON"s always win. Somehow the "ON"s are not winning.
I therefore suspect that transistor Q3 is to blame for the power fluttering. But for the power cutting out unless you hold the power button for a length of time, I suspect the bottom group of NANDs. These are possible symptoms of leakage from the small radial electrolytic cap that's in the area. Mine leaked.
-- Recommendations --
- Do not hold down the power keys, this drains your batteries
- Focusing especially on Q2, Q3, UI18, and UJ18, clean vigorously with soap and water with a toothbrush and rinse thoroughly. I give this a 75% chance of solving the problem if cleaned very thoroughly.
- This will not damage your Mac if you allow it to completely dry before powering it on.
- Replace the mentioned electrolytic cap if it looks bad to prevent future problems.