Problem on a Buck Converter for CPU_Core

[caspian]

Dear forum members,

I have a motherboard MS-14851 belonging to a MSI CX41 laptop. Its CPU_core circuit is depicted in the attached picture.
When this circuit gets on, the CPU_core voltage becomes 1.1v. voltage on Q6's gate reaches 1.3v.
In addition, the CPU and Q6 and PEC7 get hotter than normal. When the motherboard gets on, the CPU gets hot after 4 seconds whereas Q6 and PEC7 get hot after 3 minutes.
after turning off, I measured resistance between the gate and the source of Q6. It stays about 1.5 Ohm after turning the motherboard off. then it reaches 1 KOhms. that means when the circuit is on, there is a very low resistance between the gate and the source of Q6.
Is this resistance normal in Mosfets?

the resistance between CPU_core and GND is 18 Ohms. no voltage seems shorted on this circuit.
why do you think CPU and Q6 and PEC7 get hot?


thanks

[caspian]

Re: Problem on a Buck Converter for CPU_Core

the laptop displays successfully on the screen but I have to to turn it off when CPU gets too hot.
even fan and heat sink cannot stop the CPU from overheating but it overheats after a longer time interval.
If the CPU_core circuit gets on, after two seconds, the CPU_core voltage drops from 1.1v to 0.9v and voltage on Q6's gate drops from 1.3v to 1.1v.
what do you think is faulty in this circuit?

[caspian]

Re: Problem on a Buck Converter for CPU_Core

the CPU model is: Intel Dual-core B960 (sandy bridge)
there is no schematics available for this mainboard.
ISL95831 does not get hot and seems to work fine.

is it possible that the CPU is faulty that makes Q6 overheat?
or
is the low gate-source resistance on Q6 the problem?

[dellxps15]

Re: Problem on a Buck Converter for CPU_Core

hello the intel b590 is a 35 watt tpd cpu , so its similar to a core it 3rd generation mobile cpu. it is hot ofcourse if u touch it, but we need to focus on 2 things:

1° apply thermal pase on cpu and make sure it get good contact with heatsink. then power on laptop and one on windows use hwinfo64 to check cpu temperature.

if temperature in idle or normal use are around 50-60° censius then i think all is fine, while if they arent, it ould be:

1° bad cpu .... cpu or integrated vga on cpu is defective.
2° when it power on, it has intel only vga or also a dedicated amd or nvidia card in it ?
3° bad q6, q15 q16 or bad isl95831

i had an intel i7 3610qm that get very very hot just after pressing power botton, and problem is that cpu ha died without reason, with another cpu pc is working just fine.
4° if cpu is not soldered, try another one.

1.1 volt for that cpu is ok it it get too hot as soon as u start the notebook cpould be one of that. . if it give artifact or so, then vga is dead on cpu.

hope youll understand steps to do to find the problem.

[caspian]

Re: Problem on a Buck Converter for CPU_Core

thanks
the mainboard has a dis graphics chip with normal temperature and normal GPU-core voltage of 1.02v. the cpu-core voltage drops from 1.1v to 0.9v two seconds after turning on.
Once I kept the CPU temperature under 60'C using fan/heatsink. but then after 4 minutes, I observed Q6 got hot to about 70'C.
I do not know whether the low gate-source resistance in Q6 is normal or not in ON state.
I do not know which one of Q6 or CPU is defective.

[dellxps15]

Re: Problem on a Buck Converter for CPU_Core

if 1.3volt from q6 gate is pin 14 of isl95831 then isl is ok, try replace the 3 mosfet q6 q15 q16

[piernov]

Re: Problem on a Buck Converter for CPU_Core

PEC7 getting hot is not very normal but it could be from the heat of Q6.

Q6 getting hot is normal. It shouldn't get insanely hot, but may still burn your finger under load (can get above 50°C, shouldn't get too close to 100°C though).

Resistance between gate and source while there is still some power left in the board is irrelevant, it will conflict with the multimeter. (multimeter will inject a small voltage to measure the resistance)
Resistance must be measured will all power drained from the board, so you should wait for several seconds after turning off before taking measurements.

18 ohms on CPU VCore is fine when the CPU is on the socket (it's even rather high for a Sandy Bridge CPU). With CPU not inside the socket it should be higher than that.

CPU VCore dropping from 1.1V to 0.9V is perfectly normal while POSTing.


All in all, I think this laptop is working properly, even though the VCore VRM get pretty hot it may be normal. What you could check is if the other phases work properly, but you need an oscilloscope. It should be at least a 2 phases design, maybe 3 phases if it supports higher TDP CPU. With the oscilloscope, you should check that all the waveform *before* the inductors are the same. (pin 1 of PCHOKE3 on the diagram attached to 1st post)


PS: info are disorganized but I help it's understandable and that it'll help you.

[caspian]

Re: Problem on a Buck Converter for CPU_Core

thank you so much,
the CPU heat sink is mounted on Q6 too. when the laptop was on for 4 minutes, I touched the heatsink. the section of the heatsink on CPU was a bit hot whereas the section of the heatsink on Q6 was pretty hot. I think Q6 temperature with heatsink was about 70'C.