Is there a possible Advisory? Skylake's more fragile CPU PCB substrate

[BonzaiDuck]

Most of the really good heatpipe coolers probably weigh around 900 grams to 1000 grams without the fans. A 140mmx25mm fan should weigh somewhere in a range around 175 grams.

An advisory at ThermalRight mentions reports that some heatsink retention mechanisms may cause a deformation of the PCB of the processor -- leading to problems.

In March, 2016, TR then issued a communication to their resellers with a supply of clear plastic shims or spacers, designed with a taper from the hinge side of the retaining bracket to the side with the latch screw where the plastic is thinnest. The spacer fits snugly around the bottom lip of the processor cap. This supposedly distributes the pressure from the latch-plate evenly across the PCB.

TR has a photo gallery to explain how to install the shim. The gallery only shows the motherboard, CPU, socket and latch mechanism, in a sequence for installing the TR shim. There are no other parts than what you would expect with a traditional retail-box motherboard.

Meanwhile, I have an ASUS motherboard and instruction guide, in a box that includes a square brown plastic frame called an "installation tool." The name could be misleading: the instructions tell the user to leave the installation tool clipped to the CPU as directed before it was dropped into the socket. One is then supposed to close and lock the latch-plate on top of the processor and "installation tool."

Using the TR shim will cause the spring-lever to bend more than it would seem designed to do. This may or may not be a concern, but I followed up with both TR and ASUS.

TR recommends removing the installation tool and using their shim. ASUS recommends following the specifics of their guide.

Can anyone contribute more information or intelligence about this? What steps had been taken by other cooler-makers pertaining to "deformation of PCB" or similar damage risk? What motherboards are shipped with an "installation tool," and which are not?

 

[whm1974]

The last piece of advice I recall from reading about this last year was for non K CPUs to use the stock HSF. Well that said, Intel should not have made the CPU PCB substrate as thin as they did. In any case, take your sweet time installing the HSF.

 

[CHADBOGA]

I'd be looking at AIO water coolers.

 

[BonzaiDuck]

As was I, and still am. However, a survey and cross-reference of benchmarks and the expected or reported temperature drop from my delid/relid with CLU will make a heatpipe cooler a better performer as compared to a stock CPU with an EKWB Predator 240. The EKWB likely lags behind the Swiftech H240 X2. I think I can beat the EKWB by between 7C and 10C. Meanwhile, my case has been prepared for either a Predator, an H240 X2, or a custom assembly. At that point, I would either pursue it at a leisurely pace, or just ignore the additional possibilities. With the latter outcome, I might save between $70 and $150. With the former, even if prices remain constant, there will be new entrants, new kits, and more time for me to think about squeezing blood out of water.

But it might not matter. The force or pressure at the mount points -- not the weight of the cooler -- provide a potential risk. The motherboard maker's attention to an "installation tool" remaining in the socket is another indication, and the TR shim yet another. Greater pressure at the mounts would address the torque the cooler exerts on the socket. You could damage a socket 1151 CPU with a waterblock -- theoretically.

Frankly, I don't see so much risk there, and of course an AiO or custom water system offers a different regime of risks and maintenance. I just became too curious about what I might squeeze out of the LG Macho cooler. That thing weighs in at 900 Grams same as the D15 shed of fans. It's supposed to be passive (or ducted) cooler with a front-fan 120-135-140mm option, touted for socket 2011 systems with a bigger base surface. But with this new unexpected development, we're dealing with the CPU PCB and not the motherboard.

What I'd like to find out is whether other 1151 motherboards of different manufacture do not include a shim or "installation tool" that stays under the locked latch plate. Or how many people may have ignored the instructions for installing their CPU to the motherboard, and failed to include a bundled part or removed the "installation tool."

There's also another story about Skylake, and I can't tell if it's been mentioned on the voluminous "Skylake/KabyLake" thread. Supposedly, you cannot easily apply your waterblock to a bare-die because the Skylake die-height is below the square socket-frame wall or plastic square protrusion that holds the CPU in place. Otherwise, you'd have to trim plastic in part of the motherboard that makes it risky or difficult. So delidding requires re-lidding.

With the Macho cooler, I think the comparison using the same CLU-treated CPU would still have the Macho in a tie or just a degree or two behind the EKWB -- with my own TLC attentions to the building project. Or, it might exceed the EKWB Predator by a couple degrees lower load temperatures.

To sum up, I went out on a safe limb, with the $70 heatpipes and the $50 delid/CLU-treatment. These stories about cooler mounts and the CPU PCB come late in the planning stage. The CLU treatment should be a permanent thing, as will be the use of the IHS. If I can exceed performance of an EKWB and save space in the case where it counts for me, I'll do that. When or if I buy an AiO, AiO custom or custom kit, I'll get exactly what I want and need. For now, I might only need this.

 

[superstition]

It's scandalous that they thinned the substrate and incentivize enthusiasts to delid to compensate for sub-par TIM. But, since Intel has no strong competition in the enthusiast space they will continue to get sales despite these tactics.

 

[BonzaiDuck]

It's scandalous that they thinned the substrate and incentivize enthusiasts to delid to compensate for sub-par TIM. But, since Intel has no strong competition in the enthusiast space they will continue to get sales despite these tactics.

When tightening the cooler, I didn't feel any more pressure in the retention mechanism with the screws twisted in than for a 212EVO, or any other cooler I've used. But adding to the lb/ft or newton-meters of the 2-lb cooler, it doesn't seem different from any other.

It just seems that the center of gravity for the LG Macho is a bit further from the HS base.

This could either be the dumbest thing I've done in recent years, or it could turn out as well as I imagined. Meanwhile, I may have another look at the Swiftech cooler for testing it by running a molex and or PWM out the right side-panel and into a vinyl tub.

If there's an expensive disaster, it will just take me longer to get this all working. I'll postpone until I'm running on my 2017 budget.

 

[LTC8K6]

My recollection is that the substrate is thinner, but the same strength, and that no problems could actually be linked to the substrate itself being thinner.

The rare problems seen were from improper installations, or ridiculously heavy coolers, or from dropping/shipping the computer with the heavy cooler attached.

https://forums.anandtech.com/threads/skylake-coolers-can-damage-the-cpus.2456465/

 

[nerp]

I've never seen this happening. If this were a widespread issue wouldn't we see stories of enthusiasts whose Skylake chips were damaged due to coolers? I'd say that unless you're in an extreme situation this is a non issue.

 

[superstition]

My recollection is that the substrate is thinner, but the same strength

That's not possible unless the substrate material has changed. I've never heard anything about the material being stronger. Any data on this claim?

 

[BonzaiDuck]

I've never seen this happening. If this were a widespread issue wouldn't we see stories of enthusiasts whose Skylake chips were damaged due to coolers? I'd say that unless you're in an extreme situation this is a non issue.

Yes. You would see such stories. It could depend on what you mean by "extreme situation." I think I remember all-copper heatpipe towers that may have weighed in at 1.5 kg. It seems normal for the double-towers now to fall in the range around 900 gm.

Just because they looked like a good fan and I need some 140mm units, I ordered a pair of Corsair 140mm Mag Lev -- fairly new to the market, or so I came away with a sense of it.

There are plenty of 140mm fans weighing closer to 175 grams, but these weigh around 227 grams.

I've got more ambitious ideas that will leave the cooler bare of extra weight. First I have to test my hardware in the next couple days.

The LG Macho is massive enough with a base considerate of socket-2011 heat-spreaders. It's compact enough that you might not complain of a "crowded case" using a mid-tower specification. The $7 accessory duct seems to just fit without bending between a 120mm exhaust and the cooler. But it's likely you'll have to install the duct with the cooler, and any time you need to remove the duct, you would want to remove the cooler. Maybe not, but extraction and replacement look nearly impossible otherwise. I just can't see a need to remove the duct often, though.

The only thing that sort of gives me the willies is the center-of-gravity. Compared to other coolers, those pipes and the fins weigh more. But it could be a "come-back" for ThermalRight.

 

[Fanatical Meat]

I was going to attempt to say something funny about huge coolers like this old P4(?) cooler but just realized....

Compared to what other posters have said this cooler is pretty small

 

[BonzaiDuck]

I was going to attempt to say something funny about huge coolers like this old P4(?) cooler but just realized....

Compared to what other posters have said this cooler is pretty small

It's the copper. It has better thermal resistance and thermal conductivity than aluminum. But it's a lot heavier. that looks like a 92mm fan, likely less in weight than your high-end 120x25 or 140x25.

The LG Macho (please -- if you don't mind -- call it the "LGM") -- is promoted as a "passive" or "semi-passive" cooler -- "semi" for the ducting solution. I believe the comparison tests used a fairly limp case exhaust fan. The LGM RT version is simply a "deluxe" bundle with a 135mm "round" fan. The regular LGM comes with the standard TR clips to install a fan of your choice. I have an Akasa Viper 140mm "round" unit, and a Cougar Vortex 120mm -- both PWM. The Vortex only throws about 75 CFM; the Viper specs at 100 to 105. If the exhaust is rated around 110, either should work. I'd just rather not hang a fan on that cooler.

I can tentatively think of about three or four possible ways I can do it, and they may actually be fairly neat. But the Viper fan is also likely to be light compared to the Corsair Mag Levs. Confirmed: 170.1 grams for the Viper.

Well, this is about weight, coolers and sockets. I guess it belongs in this forum. Or you could move it to "C & C".

 

[LTC8K6]

That's not possible unless the substrate material has changed. I've never heard anything about the material being stronger. Any data on this claim?

Intel said they had not changed the specs. Max mounting pressure is 50 pounds for Haswell/Broadwell/Skylake.

http://www.legitreviews.com/intel-releases-statment-on-bent-skylake-cpus_175944

Intel still has not changed the specs as far as I know. The problem was with too much mounting pressure, very heavy coolers, or high G loads.

Intel provides a recommended maximum mounting pressure for third-party cooler designs, which for Skylake remains the same 50 pounds (22.6kg) of static load. This is the same pressure (50lbs) that was recommended on Haswell, so the mounting pressure hasn’t changed despite the thinner substrate. From what we are seeing after looking into this situation is that this problem seems to be exacerbated when the CPU cooler is installed on a system and it is shipped or transported. Some 3rd party enthusiast heatsinks also have more mass than what Intel’s design spec/criteria recommend, so that could also be causing some problems in a shipping situation. The max heatsink mass spec is 500 grams with a 25.4mm or less center of mass from the IHS surface. We aren’t sure if any heatsinks are breaking specs off the top of our head, but it wouldn’t be too shocking to find some on the market. It should be noted that none of the issues appear to have happened with Intel OEM cooling solutions.

http://www.legitreviews.com/some-cpu-coolers-are-bending-intel-skylake-cpus_175933

 

[Burpo]

It's scandalous that they thinned the substrate and incentivize enthusiasts to delid to compensate for sub-par TIM. But, since Intel has no strong competition in the enthusiast space they will continue to get sales despite these tactics.

More intel bashing? Supserstition/FUD, as usual.

 

[Phynaz]

It's scandalous that they thinned the substrate and incentivize enthusiasts to delid to compensate for sub-par TIM. But, since Intel has no strong competition in the enthusiast space they will continue to get sales despite these tactics.

Uh-huh

 

[BonzaiDuck]

Intel said they had not changed the specs. Max mounting pressure is 50 pounds for Haswell/Broadwell/Skylake.

http://www.legitreviews.com/intel-releases-statment-on-bent-skylake-cpus_175944

Intel still has not changed the specs as far as I know. The problem was with too much mounting pressure, very heavy coolers, or high G loads.



http://www.legitreviews.com/some-cpu-coolers-are-bending-intel-skylake-cpus_175933

I have a good touch with a screwdriver. It seemed as though I didn't need to exert any more effort to twist in the mounting screws more than would be comfortable using thumb and forefinger. Before any greater torque was needed, the screws reached their stops and couldn't turn any more anyway.

But the indications are not mutually exclusive, and a choice that exceeds a spec by double when the spec was applied from since 15 years ago -- is a hazard to shipping the cooler.

So it would seem that any cooler with the weight of an NHD14 or NHD15 would pose no risk by itself except for (a) exceeding the weight, (b) exceeding a torque spec defined as distance from the processor cap or greater than 25+mm, and (c) carelessly shipped or moved imparting sudden momentum to the cooler in its mounted position -- additional "G's" of force.

The weight spec of 500 grams has never proven to be a barrier with heavier coolers on motherboards through generations. Nor has the torque or center-of-gravity spec, but this LGM cooler does have a center of gravity further from the processor than any I remember.

I've discovered something else about this Macho cooler and its design for a front-mounted pusher fan. There's little need to duct a fan hoping to eliminate the dead spot at the exhaust hub. The design of the cooler has a recess at the center of the forward cooling fins which accommodates the width of most 140mm fan hubs and gives between 1/4" and 3/8" separation between fins and hub.

But that probably belongs in the Cases and Cooling . . .

 

[superstition]

Intel said they had not changed the specs.

The specs changed. The substrate was thinned.

So, the only way it could be equally strong as before is if it's a different material, is reinforced with an additional material(s), or is somehow engineered with a stronger form factor.

None of those things have any data behind them, though, that I've seen.

 

[superstition]

More intel bashing? Supserstition/FUD, as usual.

That's not a rebuttal.

Uh-huh

Nor is that one.

 

[Burpo]

Not interested in rebuttals of superstition.. Just pointing out the known intel basher

 

[superstition]

Not interested in rebuttals of superstition.. Just pointing out the known intel basher

I have a rebuttal for your spammed ad hom. Corporations aren't people. But what they are are entities whose purpose, in terms of a consumer's relationship with them, is to provide the least product/service for the highest cost to the consumer.

So, corporations are in business to bash the consumer's pocket book. Have a nice day, Burpo.

 

[Excessi0n]

The specs changed. The substrate was thinned.

The mounting specifications are the same. It is probable that the newer substrate is weaker than the old one, but that just means that the old one exceeded the required strength by a larger amount than the new one. A cooler which was fine on the old processors but damages the new ones is a cooler which was grossly out of spec in the first place.

 

[Burpo]

.

So, corporations are in business to bash the consumer's pocket book. Have a nice day, Burpo.

I am.. Bitter superstition doesn't sway me..

 

[superstition]

The mounting specifications are the same. It is probable that the newer substrate is weaker than the old one, but that just means that the old one exceeded the required strength by a larger amount than the new one. A cooler which was fine on the old processors but damages the new ones is a cooler which was grossly out of spec in the first place.

So, the claim is that Intel changed the substrate thickness to make its product in-spec? It's been producing CPUs for years that are grossly out-of-spec?

 

[Phynaz]

That's not a rebuttal.

Nor is that one.

You were given a suitable rebuttal, but you are blinded by your hate of Intel.

So since there isn't a rebuttal that would satisfy you, why would anyone try? It's obvious there's no way anyone will be able to convince you of anything you don't already believe.

You don't want a discussion, you want to troll. The tone of your initial post makes that evident.

 

[superstition]

You were given a suitable rebuttal, but you are blinded by your hate of Intel.

So since there isn't a rebuttal that would satisfy you, why would anyone try? It's obvious there's no way anyone will be able to convince you of anything you don't already believe.

When you have a rebuttal for what I've written I'll be happy to respond to it. So far I haven't seen rebuttals from you or from Burpo.

You don't want a discussion, you want to troll. The tone of your initial post makes that evident.

See this:

http://rationalwiki.org/wiki/Tone_argument