Investigating Reports Of Intel Skylake CPUs Damaged By CPU Coolers

[Darkbreeze]

I guess you can add Noctua's "official statement" to the list of updates as well if you don't mind.


Statement from Noctua:

Our SecuFirm2 mounting systems are subjected to prior to the release of new platforms an extensive compatibility testing. It could be determined with reference to the Intel LGA1151 platform (“Skylake”) no problems. Also on the part of our customers and our specialist resellers and system integration partners we have no reports of any problems. Our SecuFirm2 mounting systems access (with the exception of some more compact models of the L-series) for generating the necessary contact pressure on coil springs back, which allow a certain degree of flexibility both in terms of tolerances in the height as well as the case of vibrations or other forces. Compared with conventional spring-less installation systems where pressure is produced exclusively by the deformation of the mounting brackets, so can reduce the mechanical load on the CPU, and motherboard socket and any damage can be prevented by excessive force. As the transportation of a system (eg by shipping service providers) acting forces but not reliably calculate or blank check, we recommend for security in general, cooler with a total weight of about 700g (incl. Fan) take before transporting.


http://www.overclock3d.net/articles/cpu_mainboard/skylake_cpu_can_get_damaged_from_cooler_mounting_pressure/1

 

[corei7xrocks]

Dynamic Load damage? This is why I try to stay away from forums.

 

[Darkbreeze]

Dynamic Load damage? This is why I try to stay away from forums.

What is that supposed to mean/indicate? Are you trying to somehow indicate that dynamic loads do not exist? That there is no fulcrum related stress using aftermarket coolers and that this stress can dynamically increase in proportion to the force of shocks and movement? Or something else?

 

[InvalidError]

Dynamic Load damage? This is why I try to stay away from forums.

Static loads = the forces at play when things are at rest.
Dynamic loads = the forces at play during acceleration, motion, impacts, etc., such as shipping - Newton's laws of motion.

If you have a 2lbs heatsink with a center of mass 3" over the CPU, 1" of leverage from the top bolts and subject it to a 10g impact from a vertical landing, you get an extra 60lbF added to the static mounting force. If you already mounted the heatsink with 50lbF, you have now exceeded the CPU's static load rating by 120% during shipping.

 

[synphul]

I'm more simple, not as handy with math as InvalidError. Another way to look at it, standing on the bed = fine. Static load.
Jumping on the bed = bed breaks. Dynamic load showing an increase of force by the same body weight/mass.

 

[surphninja]

You are making it a issue when it is not. That is the problem with the media and people, they make a big deal out of things that are not a big deal and puts every company on edge.

Most gaming OEMs that offered a after market tower cooler have already been having people install them themselves to prevent these kind of issues due to liability. It is better to allow the customer to install it than to risk causing massive damage to the system and having to replace more components.

If the issue presents itself when not shipping then it is probably due to over tightening which should be handled by the heatsink manufacture. My H100i has a certain limit to which it can be tightened but still fits properly on a Intel setup. Being able to tighten it beyond the specified limits of the CPU and socket is a design flaw on the heatsink manufacture problem not Intel because Intel provides these specifications to the manufactures for every socket and CPU. Just because they are not specified to the end user does not mean they do not exist.

Playing the part of corporate apologist, as usual.

Intel cuts corners, it causes issues that did not exist before the change, but "media and people" are to blame for pointing it out.

 

[corei7xrocks]

Dynamic Load damage? This is why I try to stay away from forums.

What is that supposed to mean/indicate? Are you trying to somehow indicate that dynamic loads do not exist? That there is no fulcrum related stress using aftermarket coolers and that this stress can dynamically increase in proportion to the force of shocks and movement? Or something else?

No just found it amusing to think of someone doing the physics calculations necessary to determine the g-forces required to damage a substrate based on acceleration-deceleration, doing the same with static load or impact/shock to and reaching a determination that the former could cause more damage, which I took the comment as implying. I read it again and admit I likely misread what was meant, if so I apologize.

 

[turkey3_scratch]

Dynamic Load damage? This is why I try to stay away from forums.

Static loads = the forces at play when things are at rest.
Dynamic loads = the forces at play during acceleration, motion, impacts, etc., such as shipping - Newton's laws of motion.

If you have a 2lbs heatsink with a center of mass 3" over the CPU, 1" of leverage from the top bolts and subject it to a 10g impact from a vertical landing, you get an extra 60lbF added to the static mounting force. If you already mounted the heatsink with 50lbF, you have now exceeded the CPU's static load rating by 120% during shipping.

Why are you describing an impact in mass rather than in force? What is a 10g impact?

 

[InvalidError]

Why are you describing an impact in mass rather than in force? What is a 10g impact?

Basic physics: F = ma

Force and and acceleration are directly related to each other. What's a 10g impact? Dropping a box to the floor from a few feet high. Acceleration toward the ground might be only 1g but deceleration when the box hits the ground can easily exceed 10g.

 

[corei7xrocks]

I'm more simple, not as handy with math as InvalidError. Another way to look at it, standing on the bed = fine. Static load.
Jumping on the bed = bed breaks. Dynamic load showing an increase of force by the same body weight/mass.

Just gets complicated as to calculate a dynamic load you need, to put it simply, a "speed" (which can take many forms) component of some type to factor or you will be missing a necessary variable. I don't like getting complex and am not going to, just explaining my reaction to something I misread. People should not really be arguing about this as without actual variables to input into calculation we are dealing with speculation and best guess. Everyone's just trying to think it out which does not mean anyone is wrong. Intel has not even had the time to collect all variables to do any meaningful calculations would be my best guess. I have not meant to offend anyone and apologize if I did.

 

[turkey3_scratch]

Why are you describing an impact in mass rather than in force? What is a 10g impact?

Basic physics: F = ma

Force and and acceleration are directly related to each other. What's a 10g impact? Dropping a box to the floor from a few feet high. Acceleration toward the ground might be only 1g but deceleration when the box hits the ground can easily exceed 10g.

acceleration is constant and so is mass in this case, so force would be 98.1N.

 

[synphul]

The trouble is all the circumstances involved in this particular issue with skylake. It's multifaceted. One issue is clamping force and doesn't matter whether it's a larger air cooler or smaller aio water block. Another issue is while moving a skylake system and potential damage caused by excessive impacts such as shipping with a larger air cooler installed which isn't a new issue by any means. Yet another issue is human error doing things like installing a cooler with a powered device which can't be as carefully controlled/regulated as tightening with a hand tool.

All of these are potential issues for any cpu to become damaged. Some maybe more than others when dealing with skylake but it's not as if a cpu chip were just sitting there and oops it broke. We're also lacking any real volume of cases of this happening which says it's likely an isolated incident rather than some major design flaw. Of course that won't stop alarmists from immediately claiming the sky is falling.

When there's one or two cases of a vehicle accelerator becoming stuck it's an isolated incident. When 50-100 people report their accelerator being stuck within a relatively short time frame there's a recall indicating a major product failure. One person has an airbag unexpectedly deploy, a major auto manufacturer doesn't immediately issue a recall and that's what we're seeing here. A couple of isolated cases. If multiple people with multiple photos were experiencing this then I would agree intel has a major design flaw.

Just for comparison purposes, one of the cars I owned was a 2007 chevy cobalt. Many people had issues with the ignition switch and power steering modules. At the time there were no factory recalls for the car, it was still deemed an 'ok' car (personal preference or opinion aside). Most of the public are only aware of recalls but before something becomes bad enough to warrant a recall it's listed as a tsb or technical service bulletin. Tsb's are an ongoing list of known issues with higher frequency than an isolated case but less than a recall requires. Within months of buying that car it had over 115 separate tsb's.

115 problems with a single make/model of car costing $18-22k usd and people were fine with it. 2 reports of <$500 skylake cpu's bending and people lose their minds and go into panic mode

 

[InvalidError]

acceleration is constant and so is mass in this case, so force would be 98.1N.

And that 98N is applied at the end of a 3" lever which is itself perpendicular to a 1" lever, which gives it 3:1 leverage and yields ~300N applied to the CPU. 300N = 67lbF, similar to what I wrote earlier.

 

[corei7xrocks]

It occurs to me that we have to bear in mind that any substrate issues are allegations and I personally question how much credence should be given to a finding by pcgameshardware.de. They review and test hardware and software. I really have questions regarding the sample size that they would have been capable of using. Any CPU can be damaged by the same forces being discussed. How much reliance would we give to a finding by an American PC Magazine which made a claim like this. Does anyone think they would or could have obtained a sufficient sample size of Skylakes to prove the allegation, or am I missing something.

 

[Darkbreeze]

Bottom line here though, as mentioned before, is that if you remove the HSF before moving the case around to any significant degree, as with shipping or traveling, there shouldn't be any problem. This does however severely undercut the possibility of using a rig equipped with a Skylake chip and tower cooler as being a good candidate for a LAN party or somewhat mobile machine that can easily be taken from place to place, which we know many users do, unless you are immeasurably careful and it would probably be highly recommended to always keep the unit flat on it's right side panel during such movement. Even then, it might be risky if the problem is as severe as has been indicated, which it probably is not. Is there an issue, probably. Is it the end of the world and is every Skylake chip going to bust if you look at your case sideways, probably not.

 

[atheus]

Just thought this might be an interesting read. Shipment of systems with air coolers installed (other than small/stock) isn't new territory and is mentioned by several places.
"Massive, monolithic air coolers – and even some smaller, more affordable ones – are still extremely popular in the marketplace. While system integrators have largely moved away from shipping air coolers in their high performance systems as the weight of a massive air cooler can cause damage to the motherboard in shipping"
http://www.corsair.com/en-us/blog/2015/april/liquid-cooling-vs-air-cooling-which-performs-better

From an article covering a cooler review right here on Tom's Hardware. "Experience tells us that the odds of a big air cooler either breaking a board or coming loose during shipping are far greater than that of a closed-loop-liquid cooler causing damage from leaks."
http://www.tomshardware.com/reviews/v3-voltair-v3tec120-fc01-tec-cpu-cooler,4165.html

Similar suggestions from 2008.
http://www.overclock.net/t/403525/what-is-safest-way-to-ship-a-computer

Similar issues discussed in 2010.
http://www.silentpcreview.com/forums/viewtopic.php?t=58747

From an article in 2012.
"While these kinds of gigantic tower heatsinks have become favourites of overclockers and other high end enthusiasts, they have created their own set of problems. The first is weight. When mounted in a tower configuration the combination of gravity and the weight of the heatsink can stress the motherboard. This is especially an issue for system builders, who have to ship systems with heatsinks already installed. A bit of jostling during shipping and a weighty heatsink can damage the motherboard, especially around where the heatsink mounts."

Read more: http://www.pcauthority.com.au/Feature/297802,system-builder-watercooling-goes-mainstream.aspx#ixzz3tlNOcM42

The point being that damage from use of a larger air cooler during normal conditions, aka on a desk in an office, not being humped up the himalayas or surviving a round of ultimate frisbee, is remote. During shipping yes it's a real possibility and has been for some time. Maybe people are just now realizing this, maybe the slightly thinner skylake cpu's are more easily damaged by larger coolers in shipping but it's definitely not new. Nor should people act like it is. All of the sudden people are realizing something that's been known for years (obviously by the linked sources and dates) and are blaming a cpu that didn't exist until a couple of months ago.

Improper place of blame seems to be the recurring theme. If in fact skylake chips just mysteriously crush or crumple while a tower is sitting on a desk during normal use then I think there's a legitimate complaint. If the argument is "I dropped my pc off the desk and it broke" I think people need to be more careful with their valuables. Most smartphones can't sustain a drop of a couple feet without the screen shattering and they only weigh a few ounces. A 15-20lb pc tower shouldn't be expected to survive that either.

The thing is, none of the stuff you just quoted, and none of the past stories I have ever read about a computer shipped with a tower cooler mounted make any mention of actually damaging the CPU. There are plenty of cases of the cooler breaking, or maybe PCIE slots cracking... I've never personally read about a motherboard cracking, but I've seen plenty of people speculate that it might happen. It's not until now that the CPU itself bending or breaking is actually a thing that happens. This is why people with broken CPU's would be justifiably outraged — because if it were a 4th gen CPU, chances are it wouldn't have broken. If some sort of shipping insurance covers it that's fine (though not for the shipping service), but for those who experienced a breakage they're responsible for, that's going to hurt — especially with the way i7-6700k's are right now (usually out of stock, and/or priced far above msrp).

 

[Darkbreeze]

I've never once heard of the cooler being damaged, unless the mount itself broke. I've seen and worked on at least a couple of portable LAN party type rigs that had caused the motherboard to be damaged though. Granted, these had massive HSF's attached and probably weren't handled even remotely carefully, but still. When I see evidence that this IS in fact a widespread problem, and not just one or two cases of user error due to improper mounting, that has been sensationalized, as usual, then I might consider glancing towards the torch and pitchfork that are stored in the shed.

 

[synphul]

It's a thing that happened, I haven't seen it happening in plural. Damage from heavy cpu coolers like large air coolers during shipment are real and why most prebuilt pc's that ship use aio coolers. They got sick of covering shipping damages.

Taking a quick look through newegg as an example, read the reviews. Over 100 reviews so far for the 6700k, not a single one mentions a bent/broke cpu. One came doa, another user had bent pins (unknown reason). Most 'cons' are lackluster performance compared to the 4790k. Same with the i5 6600k, another 80-90 reviews, not one mention of a bent cpu/substrate. While doa cpu's can happen it's fairly rare.

Look at the 4690k reviews, around 400 or more (roughly 4x that of the 6700k) and per 100 reviews, more negative reviews on the 4690k which many consider a solid cpu. A few complaints of doa cpu's. So far skylake has fewer negative reviews per 100 than haswell/devil's canyon chips but they're so much weaker. The logical part of me is asking that if this is such a likely problem why aren't more people having this issue?

As it stands we have one review site repeated everywhere that there's a huge issue. It doesn't make sense that such a serious and potential flaw would go unnoticed by so many average users, many of whom are using large heavy air coolers. The evidence just isn't lining up with pcgameshardware's allegations that there's this major flaw. Surely 5-10 of those 100 reviews should have mentioned it.

 

[atheus]

Don't misunderstand me. I'm not making any predictions about how frequently this might happen. I'm just making an observation that for those who do experience it, the blame will fall heavily (though still partially) on Intel's design change, rather than purely blaming whatever mistake was made in handling the PC.

Whether the victim is aware of it or not is another question. I'd say it's good to be aware. I was certainly a bit more cautious than usual while screwing my Be Quiet! Dark Rock Pro 3 on to my 6700k yesterday (especially since there are no springs in the mounting system for it).

 

[Darkbreeze]

Oh screw that. I'm using my cordless impact and running it down til it gets soft again, then another two turns, like I do with everything else. I might even give it an extra couple of bumps with the impact, just to be sure.



(Disclaimer: Just to play it safe against the chance that somebody is dumb enough to believe that, I'm just kidding.)

 

[logainofhades]

Oh screw that. I'm using my cordless impact and running it down til it gets soft again, then another two turns, like I do with everything else. I might even give it an extra couple of bumps with the impact, just to be sure.



(Disclaimer: Just to play it safe against the chance that somebody is dumb enough to believe that, I'm just kidding.)

No no no, you need to make sure to use a torque wrench, to make sure you properly torque the screws.

 

[Darkbreeze]

Ok, so torque them to spec, and then turn them until they get soft again, plus two more turns, right?

 

[logainofhades]

No, I think those bolts are TTY, so you don't go beyond the specified torque.

 

[Darkbreeze]

Oops, they yielded. Heh.

 

[Bamming1]

Oops, they yielded. Heh.

Do they make micro toque wrenches?