Optiplex 3050 Vcore Mosfet short

[pavithra_uk]

it is 3+2 vrm
vcore high side mosfet (AON6380) shorted when turned on instatly killed the processor (i5 7400)
gate drive chip (NCP81166) also gone.

any suggestions before replace mosfet and gate drive chip? what if I replace 3 high side mosfets with more current capable mosfets?

odd question:
what if I put two silicone diodes series in forward way? assume each diode 0.7v drop and total of 1.4V. vcore not exceed this but can protect cpu from direct 12V hit from shorted mosfet ?

[DynaxSC]

The question is what was the reason for the mosfet short. Exchanging mosfets and driver should be enough, but check also if no lowside mosfet has a short. When exchanging mosfet more important is Rdson parameter, than max. current. Rdson defines how much heat is generated inside conducting mosfet, the lower Rdson, the better. Current can be same or higher, provided Rdson in lower. Also efficiency of VRM will rise with lower Rdson. Last is the frequency range of the mosfet, must be not lower than original.
Adding the diodes is very risky, as CPU voltage can achieve also 1,4V under stress, and then you have a short, vrm broken, diodes melted. You would have to take 3 diodes in series, but this will no more have any sense, as Vcore of 2,1V will probably kill any CPU anyhow. Also diodes begin to conduct with lower voltages than 0,7V, so you would have anyway an significant additional parasitic load, even at 1,0V, not good for CPU power stability.
VRM controllers do have overvoltage protection, and if they work properly, this is completely enough to protect CPU. What can lead to VRM failure are electrostatic discharges from mains (over the PSU) or LAN cable, so better protect PC PSU and LAN input with a good local surcharge protection. Also the whole house/flat electric installation should have protection devices against surcharges/lightnings. So such a 2 step surcharge protection is optimal.

[DynaxSC]

The question is what was the reason for the mosfet short. Exchanging mosfets and driver should be enough, but check also if no lowside mosfet has a short. When exchanging mosfet more important is Rdson parameter, than max. current. Rdson defines how much heat is generated inside conducting mosfet, the lower Rdson, the better. Current can be same or higher, provided Rdson in lower. Also efficiency of VRM will rise with lower Rdson. Last is the frequency range of the mosfet, must be not lower than original.
Adding the diodes is very risky, as CPU voltage can achieve also 1,4V under stress, and then you have a short, vrm broken, diodes melted. You would have to take 3 diodes in series, but this will no more have any sense, as Vcore of 2,1V will probably kill any CPU anyhow. Also diodes begin to conduct with lower voltages than 0,7V, so you would have anyway an significant additional parasitic load, even at 1,0V, not good for CPU power stability.
VRM controllers do have overvoltage protection, and if they work properly, this is completely enough to protect CPU. What can lead to VRM failure are electrostatic discharges from mains (over the PSU) or LAN cable, so better protect PC PSU and LAN input with a good local surcharge protection. Also the whole house/flat electric installation should have protection devices against surcharges/lightnings. So such a 2 step surcharge protection is optimal.