- Joined
- Aug 30, 2004
- Location
- Santa Barbara, CA
Why Mod?
The motivation is to prepare this board for a quad-core CPU. There have been reports of caps blowing up with the P5K series of boards when pushing a quad-core CPU to the max. In most cases the caps that blow up are either on the CPU VRM side or on the source ATX 12V connector side.
There can be two possible reasons booth which have a cascading effect.
What to mod?
The motivation is to prepare this board for a quad-core CPU. There have been reports of caps blowing up with the P5K series of boards when pushing a quad-core CPU to the max. In most cases the caps that blow up are either on the CPU VRM side or on the source ATX 12V connector side.
There can be two possible reasons booth which have a cascading effect.
- Since the P5K sports only a 3-Phase VRM, the ripple current is going to be larger. This means when large currents are drawn by a highly OC'ed quad-core, the ripple current will increase as well. The problem here is not just the current draw, but the slew rate, i.e the time-rate of change of current. When the Slew-rate beginss to increase as the draw increases, so will the ripple current. At some point this will exceed the ripple current rating of the capacitor leading to a kaboom!
- Another way of looking at this would be to understand how a multi-phase VRM topology works. In an n-phase power distribution scheme, power is divided into n-phases that are offset by a fixed angle. To understand this, we have to consider only the AC part of the current from the voltage regulators (i.e ripple current). The purpose of the n-phase offset is to ensure that the average AC current is close to zero due to destructive intereference between the AC components. These phases are controlled by the PLL IC and the duty-cycle regulator ensures that the desired current amplitude is maintained. When highly overclocked, there can be a phase mismatch because everything is operating far out of spec, so this in effect contributes to a large ripple current.
- As the current draw from the PSU increases, the teeny 16V 270uF Solid Cap turns out to be insufficient. You can see on the silkscreen that there is space for more. If the ripple current rating of the ATX12V cap is exceeded it blows as well.
- Having no filter caps near the 24 Pin connector also worries me a bit. I'm going to use a 16V 1500uF Chemicon LXZ in that location.
What to mod?
- Add 16V 1500uF or 16V 680uF Chemicon LXY near the ATX 12V connector.
- Add 4V OSCON SEPC near the phase inductors.
- Add 6.3V 680 uF Nichion Conductive Polymer near the FETs.
- Add 6.3V 680uF Nichion conductive Polymers near the chipset.