Question cpu short circuits

[jalim]

My laptop Toshiba satellite m40 CPU 730 when I put the laptop on it is starting everthig normal but after some minutes becomes very haught on the back side of the laptop than starts troubles in the windows

When I have disassembled the laptop and I have tested components on the motherboard I have found alot off short circuit in many points especially on the CPU side than I have disconnected the CPU out and I did another tests for short circuit on that motherboard with multimetre and I found nothing off short circuits

After I did another tests for short circuit on CPU pins out off the motherboard and i have found all pins off CPU have being shores please anyone's can help me how that can be happen and resolve this problem
Thank you my friends

Than

 

[Achaios]

You need to clean the IHS of CPU and reapply quality TIM such as Arctic Silver 5.

View: https://www.youtube.com/watch?v=aD3AGwHKWp4

View: https://www.youtube.com/watch?v=uZYU3Z2QFVQ

 

[mortemas]

Regarding the "shorts" I wouldn't worry too much about that because many of the pins are connected to each other for the purposes of power delivery and redundancy. See diagram below of a coffee lake cpu pinout:

 

[InvalidError]

Regarding the "shorts" I wouldn't worry too much about that because many of the pins are connected to each other for the purposes of power delivery and redundancy.

What are you calling "redundancy" here? Only about 30% of the pins are something other than power/ground and most of those are absolutely critical (must work for normal operation) with no redundancy: lose any single DRAM data or control pin and you lose an entire DRAM channel, lose a PCIe lane and whatever device it connects to loses half of its bus width. The ton of power/ground pins are there to reduce socket parasitic impedance so high speed signals can survive going through the socket without getting mangled beyond recognition and EMI compliance, not really for any sort of redundancy.

 

[mortemas]

Power pins not there for redundancy? So with coffee lake as an example what you are saying is that you can send 95 watts of power through a single power pin, but they use numerous pins only to reduce parasitic impedance? I would say it is for all three reasons of power delivery, redundancy, and noise reduction, but my original statement still stands. You don't want failing power pins to cascade into catastrophic failure as every pin lost reduces max current carrying capability, so there are more than enough to provide the expected power draw. If a few weren't making contact the CPU would still function. I believe it was the z390 mobo that was said to have added power pins for the i9 CPUs, but the headroom is such that the i9 chips can work on a z370 mobo. Sure you're closer to the "line" with less power pins, but yeah that's the pins you can afford to lose 1 or 2 of (or whatever the difference with i7 vs i9 CPUs) without dire consequence. I said nothing of data pins. The meat of the statement was that the OP was just not going to get anywhere taking a multimeter to the cpu without acutally knowing what all the pin functions are, not to mention what the expected result should be on the multimeter for each pin tested.

 

[InvalidError]

Power pins not there for redundancy? So with coffee lake as an example what you are saying is that you can send 95 watts of power through a single power pin

You don't need anywhere near 400 power pins to carry 100A, LGA775 could do it with ~250 pins for the 136W QX9770. You do need the extra pins to reduce socket parasitics both for improved signal integrity on ever faster IOs and also so the motherboard's VRM can more effectively regulate voltage that actually gets to the CPU substrate.

If a given socket has 1nH of effective inductance at 1GHz and you have 20A worth of ripples at that frequency, that's 125mV of ripples on the substrate about which capacitance on the motherboards can't do anything about. Can't have that on a CPU that operates at ~1.2V, so you have to more than double the power/ground pin count if you want to get this under 50mV. That's the main reason why modern sockets for today's CPUs have so many more power/ground pins than they used to.