Originally Posted by bonami2
Trying to understand all those part is pretty hard since i have no electronic training. All learned by myself and google.
Don't worry, i 've no electronic trainning either. I just liked physics at school.
Btw seeing Sin Chart of how a vrm is made. Is the vrm send power to the capacitor and after it go to the cpu ??
The capacitors in the diagram above, are on the output, the last stage before the power arrives to the CPU. Those capacitors, mainly serve as power conditioners (filters), to prevent spikes and ripple. Basically, the problem is this. The PSU transforms AC to DC and uses the 12V rail to give power to the motherboard. The CPU though would be fried at 12V, so another circuit must bring down the current to lower values. The motherboard has the PWM chip that "controls" the mosfets, so everything is fine and dandy. Almost... The problem with the CPU, is that it doesn't require always steady power (think of Cool N Quiet, load etc) and the nature of the phase switching, makes it so that the CPU requires rapid power adjustments in the order of msec. That's too fast without something to smooth out the power. This "something", are the capacitors.
The capacitors, have 2 benefits: 1) they retain charge , 2) they can give displacement current every time there is a voltage variation applied to their two extremities (the 2 "legs" that each capacitor has). So, they can both smooth out incoming power and also give power steadily outgoing, because they hold charge themselves, so that the CPU doesn't get power interruptions during the milliseconds that phase switches occur. Of course, being small, they can't give power forever, they simply do so for a tiny amount of time, before the incoming power gets adjusted once more. It's why if you shut down your PC and then your PSU's power switch, and you press the power button in your case, your fans will likely spin for 1 second, even though the PSU is shut down. It's the remaining charge in the capacitors that make the fans spin without AC current from the wall coming to the PSU.
At least this is my mundane understanding of the situation, as an amateur.