Intel Skylake CPUs bent and broken by some third-party coolers

[Kenrou]

http://arstechnica.com/gadgets/2015...-bent-and-broken-by-some-third-party-coolers/

http://www.tomshardware.com/news/skylake-cpus-damaged-by-coolers,30690.html


Bad news for Scythe...

 

[wagex]

uhoh

 

[Johan45]

That's good to know Kenrou thanks. Better be careful with water block and pots I guess.

 

[ATMINSIDE]

And this is why it's best to buy heatsinks and blocks that use springs as the final clamping mechanism.
Parties like Noctua and Swiftech strictly adhere to Intel's 50lb clamping force for a reason.

 

[JrClocker]

The "possible damage during shipping" is an interesting take noted in the articles.

Hey ATM - have you ever looked any sort of center of gravity measurements on the large fin-type coolers? Basically, if the COG is too high off the chip, I could easily envision more than 50 lb worth of pressure applied asymmetrically during normal shipping "bumping" for some of the larger fin coolers.

 

[EarthDog]

The question to me isn't so much that, but zoom out and look at the big picture...

... why was this not reported with previous generation CPUs that used the same coolers? This seems to be an issue with the substrate the CPU die is on? I mean, those coolers have been around for quite a while and I do not recall hearing about this issue.

 

[ATMINSIDE]

The "possible damage during shipping" is an interesting take noted in the articles.

Hey ATM - have you ever looked any sort of center of gravity measurements on the large fin-type coolers? Basically, if the COG is too high off the chip, I could easily envision more than 50 lb worth of pressure applied asymmetrically during normal shipping "bumping" for some of the larger fin coolers.

This is why good coolers use springs to mount, it keeps force both symmetrical and consistent.
By using the springs you effectively negate the issue of hanging weight.

On a solid bolted cooler, it would be an issue.

The question to me isn't so much that, but zoom out and look at the big picture...

... why was this not reported with previous generation CPUs that used the same coolers? This seems to be an issue with the substrate the CPU die is on? I mean, those coolers have been around for years.

Skylake PCB's are thinner.

 

[EarthDog]

Right. I read that.. but they also said the 50lb of force is still the same. So, in theory, it shouldn't do that.

 

[ATMINSIDE]

Right. I read that.. but they also said the 50lb of force is still the same. So, in theory, it shouldn't do that.

Right, but a lot of coolers, especially the ones not using spring mounting, have higher force than 50lb.

 

[EarthDog]

Surely. But why didn't they bend previous gen CPUs and this did even though it was rated for the same amount of force?

50lbs of bricks is the same as 50lbs of feathers, right?

 

[ATMINSIDE]

Surely. But why didn't they bend previous gen CPUs and this did even though it was rated for the same amount of force?

50lbs of bricks is the same as 50lbs of feathers, right?

Because the previous gen CPU's had a thicker PCB to handle this same force.

The PCB I'm referring to is the actual PCB that goes into the CPU socket that holds the die and pads.

 

[EarthDog]

Yep, me too. I guess with the thinner substrate, it is allowing (a lot) more deviation...

... but you would like they would have found this in their testing (an underlying point of my statements) and this wouldn't/shouldn't happen...

 

[ATMINSIDE]

Yep, me too. I guess with the thinner substrate, it is allowing (a lot) more deviation...

... but you would like they would have found this in their testing (an underlying point of my statements) and this wouldn't/shouldn't happen...

My take is that coolers adhering to the 50lb force aren't causing issues, which Intel wouldn't have released anything that caused an issue within their specs.

 

[Robert17]

It's much easier to design a spring mount pre-configured not to exceed 'X' amount of post-tensioning pressure than it is for a hobbyist to guess if they're acheiving proper load; numerous threads on this forum question how to torque down screws and not mess something up. The calculations for both methods, springs and thread tensioning, are Knowns. Application in the real world by folks that don't perform the actual feat very often makes it very hard for designers to idiot-proof.

 

[Alaric]

Nothing is idiot proof. Mother Nature will always invent a better idiot. LOL

 

[RollingThunder]

Are all Skylake boards thinner or is this a vendor-by-vendor issue?
I ask because I have an Asrock 170 Extreme4 here waiting for a 6600K.
However I won't be mounting any cinder block coolers on it!

 

[EarthDog]

This is about the Skylake CPU and its PCB/substrate, not the motherboards.

 

[ATMINSIDE]

Are all Skylake boards thinner or is this a vendor-by-vendor issue?
I ask because I have an Asrock 170 Extreme4 here waiting for a 6600K.
However I won't be mounting any cinder block coolers on it!

Not motherboards, but the actual PCB on the CPU itself.
All Skylake CPU's are thinner than Haswell, yes.

As long as you use a cooler that adheres to Intel's 50lb clamping force you'll have zero issues.

 

[RollingThunder]

Not motherboards, but the actual PCB on the CPU itself.
All Skylake CPU's are thinner than Haswell, yes.

As long as you use a cooler that adheres to Intel's 50lb clamping force you'll have zero issues.

Ah, understood!

 

[Joe88]

In the first link there is a nice picture comping the size of the substrate of skylake and haswell
The skylake looks to be about 25% thinner, this was probably done to save money more than anything else.