"Skylake safe" cooler could damage Sky-, Kaby and Coffee Lake CPUs

[bowrilla]

Two years after the initial issues regarding mounting pressures of big air coolers with the then new Sky Lake CPUs the issues celebrate a revival as it seems.

 

German tech magazine "PC Games Hardware" reported a serious issue with Skylake system that's just over two years old and was assembled by German blogger Bjoern Olausson. Equipped with an i7 6700k and the Be Quiet Dark Rock Pro 3 died a mysterious death – mysterious until the "autopsy" revealed serious deformations on the CPU's substrate. Olausson reported that his system has never been moved around and was therefore never exposed to sudden movements

 

PC Games Hardware published further tests unrelated to that specific incident in their July and September issues and published a mounting force table (see original article at PC Games Hardware). Test results revealed that most coolers exceeded Intel's specs for static mounting forces (222 N) by quite a margin (maxium value of still produced cooler was 540 N) though Intel's limit for dynamic forces as in sudden shocks and other momentary spikes was not exceeded (712N). The Be Quiet Dark Rock Pro 3 was measured with a mounting force of 380 N.

The question remains if the excessive mounting forces of large air cooler are or at least can still be an issue. Interesting enough: even some AiO might exceed Intel's spec of static mounting forces (i.e. Cooler Master Master Liquid 240 with ~240 N). Maybe the issue is not fully resolved and large air coolers might still damage CPUs in the long run which could be an issue regarding RMA requests as well.

 

For copyright reasons I'll only link to Bjoern Olausson's page. The damage is obvious.

 

Sources:
PC Games Hardware (german)

Bjoern Olausson (english)


 

[Eduard the weeb]

Well hopefully it is just a "fluk" but there is a chance this could be a problem again and if lots of Intel cpus start dying after two years pretty sure that would hurt there credibility. 

[DocSwag]

Praying right now that my dark rock pro 3 won't end up killing my 6700k...

[bcredeur97]

what about my NH-D15? what mounting pressure does that use?

[PCMasterDebater]

Just now, bcredeur97 said:

what about my NH-D15? what mounting pressure does that use?

I would like to believe Noctua spent more on R&D then Dark Rock developing their NH-D15 mitigating such issues. *crosses fingers* 

 

Those with their motherboards set flat in a case likely don't have this issue at all.

[bowrilla]

according to the table at PC Games Hardware it exceeds Intel's specs by ~50 N at a total of 270 N. That's about 12% more.

[bcredeur97]

4 minutes ago, bowrilla said:

according to the table at PC Games Hardware it exceeds Intel's specs by ~50 N at a total of 270 N. That's about 12% more.

at least it isn't much over the rating..

 

[Princess Luna]

Intel's Stock Cooling master race.

[Bananasplit_00]

doubt this will impact more then a coupple of people to be honest

[NumLock21]

5 minutes ago, Princess Cadence said:

Intel's Stock Cooling master race.

Aio liquid cooling supreme master race.

Exposed cpu cooling yolo master race. 

[Gasolinie]

What about the Corsair H100i V2 on i7 7700k? How much mounting force does it have.

[Tam3n]

Unless the threads bottom out, doesn't mounting pressure depend on how vigorously you torque the screws on the cooler?

[bowrilla]

 

That depends on the construction of the cooler. If the manual tells you to mount it in a certain way (as in: tighten screws until springs are fully compressed) then that results in a specific mounting force. The heavier the cooler the tighter it needs to be in order to not move around (which could result in damages as well). 

Of course, you shouldn't overtighten the screws. Engineers usually have a certain torque in mind. There's a German saying about overtightening screws: "Nach fest kommt ab" that roughly translates to "It's tight until it rips off"
 

32 minutes ago, Tayfun1212 said:

What about the Corsair H100i V2 on i7 7700k? How much mounting force does it have.

There are no values for Corsair coolers yet. Any infos from @Corsair @Corsair Blake  directly?

[Matu20]

This sounds really bad, basically crossing fingers and hoping for best is the only option.

[TidaLWaveZ]

1 hour ago, Tam3n said:

Unless the threads bottom out, doesn't mounting pressure depend on how vigorously you torque the screws on the cooler?

That's what I was thinking too but thought there might be something I wasn't considering, like maybe the numbers are minimums(lowest possible mounting pressure) that could be achieved given the weight of the cooler.

[TVwazhere]

Theoretically you could just reduce the amount of force we as builders use to tighten the screws? This doesnt sound like a air cooler weight issue.

 

I remember someone posting a question asking why we dont torque screws to a specific spec. but the AMD thread ripper torque screw and now this are making me think he might have been onto something....

[bowrilla]

49 minutes ago, TVwazhere said:

Theoretically you could just reduce the amount of force we as builders use to tighten the screws? This doesnt sound like a air cooler weight issue.

 

I remember someone posting a question asking why we dont torque screws to a specific spec. but the AMD thread ripper torque screw and now this are making me think he might have been onto something....

This has never really an issue right until the substrate got thinner. Another one of Intel's moves to save a few cents next to the crappy thermal compound. 

I've just checked with Noctua's installatioan manual for their NH-D15: it says "Caution: Tighten the screws until they stop."

That means there's literally no room for applying less torque since the screws would be loose and the cooler might not have full contact, which could impede cooling performance. That means the cooler will put aforementioned pressure on the CPU and its substrate.

Cryorig isn't really specific in ther installation guide for their C1 cooler.

Be Quiet! changed their manual and from 09/2017 on is now specifically saying in ther installation manual for their Dark Rock Pro  3 cooler (the one mentioned in the initial news): "Place the hexagonal nuts (8) over the spacer nuts (7) and progressively tighten these in a crosswise manner until firm using the hexagonal wrench (14)."

Before that there was no mention on how tight it should have been. The wrench is potentially an issue here and poses a danger to overtighten it compared to a simple screwdriver.

 

Still: there isn't much room for adjusting the mounting pressure. 

Maybe this should be checked systematically? Maybe this could be part of a video since it does pose potentially fatal danger to 6th-8th gen Intel Core CPUs. At least mounting pressures should be checked when reviewing new coolers. @LinusTech @nicklmg @AlexTheGreatish
 

[AlexTheGreatish]

1 hour ago, bowrilla said:

~snip~

 

I wonder if it's more an issue of differences in pressure as opposed to pure pressure.  I wonder what kind of a moment a tower cooler is putting on the CPU and if having it flat would make things better.  

[Dabombinable]

God I love how tough/non skimped on my i7 4790K and LGA775 collection are. The substrate really is too thin.

[flibberdipper]

5 hours ago, Princess Cadence said:

Intel's Stock Cooling master race.

Nah, Dell stock cooler master race.

[bowrilla]

3 minutes ago, AlexTheGreatish said:

I wonder if it's more an issue of differences in pressure as opposed to pure pressure.  I wonder what kind of a moment a tower cooler is putting on the CPU and if having it flat would make things better.  

Good question. I've read the theory that LGA sockets were more prone due to a smaller contact area on the socket. Back in 2014/15 cooler used to got wild on mounting pressures which led to several damaged CPUs and cooler manufacturers releasing new versions. In comparison the substrate got really thin (abouth half as thin) with Skylake. The pictures in Bjoern Olausson's blog post indicate a weird pattern: the middle and the outer corners are bent upwards.

PC Games Hardware did some tests back in 2014 figuering out by how much thermals are affected by the mounting pressure of the cooler. It seemed as if it peaked at ~250 N of force while not rising significantly once that pressure was applied (and supposedly thermal compound was spread out). 

The question remains: how much do you need to tighten it in order for it to be safe for transport (as in LAN parties or semi mobile ITX systems)? Is it actually safe to have a high end big air cooler in such a system? As of now I'd seriously avoid in high performance systems big air coolers even though they perform just as well for half the price.

Gosh, I'm glad to have a 4790k …

[Curufinwe_wins]

This is why Noctua and a number of high end watercoolers use spring systems. Tightening the screw down to maximum doesn't directly apply the force, rather the distance compresses the spring and thus applies a specific force which is easier to control against people being dumb. Now with that said, such mechanisms have their own issues in theory.

[NumLock21]

I messed up on the meaning from my previous post. What i meant to say was,

 

No cpu cooler supreme yolo master race.

 

[RyanEsau]

Well.. that's an interesting development.

 

I'm using a Hyper 212X which in their list is 200N of force so I'm under (albeit close) the 222N mark. Surprised the 212 EVO isn't there. (In their listed coolers w/ their respective forces)

 

I also have a Hyper T4 but I couldn't get that thing mounted for nothing and ended up buying the 212X. (Darn seesaw design...)

 

I wonder how much it'd add if I had 2 fans on the cooler, if any. I actually bought the 212X Dual Fan on Newegg at the time because they only wanted like $5 more over the standard which seemed like a deal to me. I actually had trouble getting the fan on the cooler though... so I just stuck with the single fan.

[bowrilla]

Well, the weight of 1 kg equals about 10 N on earth (9.8… to be more precise) so 2 fans won't add much. I estimate 250 g per fan, so 0.5 kg in total equaling a force of 5 N. A NH-D15 weighs 980g without fans and 1.320g incl. fans. That's even less than I've estimated for general fans. That NH-D15 with two fans without being screwed in place would equal to a force of ~13 N. 

Comparison of Devil's Canyon vs Skylake with a good picture of the thinner PCB with fewer layers by Anadtech from 2015. So Intel probably just optimzed the hell out of their chip to get away with fewer layers saving a few cents and not wasting a thought about weakening the chip.