• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

2500k heat spreader quality control

S

smonska

n00b
Joined
Oct 8, 2011
Messages
41
Out of the 100+ PCs I have built, this was a new one to me. Wondered if anyone else has received a new CPU with a warped IHS? I'm super glad I bought my CPU from @m@z0n, who lets you return CPUs!

Built a new 2500k rig, at first using the stock heatsink and arctic silver 5 smoothed with a razor blade. The CPU temps were good at idle, like ~30C, but would spike 40C in like 3 seconds after loading and stabilize around 80C-90C. Cutting the load, they would drop 40C in a couple seconds too...

Hmmmm..... After taking the heatinsk off I found that the distribution of AS on the cpu was very uneven, with no contact in the center. I reapplied the Arctic Silver 5, using a new razor blade to smooth it, put it back together, same deal with temps. I picked up a new Zalman CNPS-10X heatink, and had the same problems, and that heatsink comes with a near-mirror finish flat surface!

I looked up warped CPUs online and found out that the CPU's integrated heat spreader (IHS) itself can be not flat, and you can check flatness by putting a drop of alcohol and the center of the CPU or heatsink and press it up against a piece of glass, like from a picture frame, and watch where the fluid goes. Sure enough, my IHS on the 2500k was concave.

Ordered up a new 2500k, and now my CPU temps "jump" less than 7C when a load is applied. What gives Intel?
 
Eh, I've never had that problem but I'm sure others have. You're going to get a bad piece of hardware from time to time, one bad product slips through the cracks etc.

Lol but you make it sound as if Amazon is the only e-tailer that will replace your CPU if its bad :)
 
A lot of places have "special" return policies on CPUs, that suck. Some will give you a hard time about what is really a defect, etc. Once you've opened the box, CPUs can be hard to return. Been there, done that.
 
Well Intel has a good warranty and they will likely replace the CPU for you. I would go there and try and get another. So if I get this right you have 2 2500K's a good and a bad?
 
ii purchased my 2600k with the motherboard and it ran hot so i exchanged it they didnt bother me... but then when i realzied the 2nd one overclocked worse and ran hotter then the first one i tried to exchange it again.. this is where i had problems on the 2nd time i had to speak to the manager and get the tech at the store on my side before I attempted to take out the cpu and drive to take it but in the end they exchanged it and if anything else happens i will send it to intel
 
I successfully returned my bad one, and purchased a good one. Interesting to note other received chips that "ran hot". Downside of having chips with IHSs I guess is we don't get to put our heatsinks right on the dies, adding one more variable to the equation of heat dissipation, and one we can't easily do anything about.
 
Glad to hear that you got this taken care of, I havent heard of a problem with the heatspeader problem on any of our processors before so that is a new one for me.
 
It's an uncommon problem, but not unheard of. First time I encountered it was while building a PC for a client several years back using an E6420 proc. Most HS/IHS will be slightly concave/convex, but usually not enough so to pose any problems.

I test-bench + stress test every PC I build, just to make sure all the hardware works. Upon beginning CPU stress testing, I noticed it overheating and throttling after 20 seconds of burn. I reduced the test to 50% usage and it still overheated and throttled. It wasn't until I brought the test down to 25% load that it managed to not overheat. I took off the HS, and noticed that the paste had only been squished around the outside of the HS base. Upon closer inspection of the IHS, there was a noticeable 0.75 mm depression in the center.

I've had a few since then that have been fairly bad (~0.5mm gap), and a fair few more that were ~0.25 mm. I usually just lap these chips down to flatness. However, 95% of the chips I get are adequately flat, at least for regular usage.

Some manufacturers even include a slight convex shape to their HS bases.
Thermalright FAQ page

17
Question :I received my Ultra-120 eXtreme the other day and I noticed that the bottom of the heatsink is not perfectly flat and when I put it on a flat surface, it tends to swivel. Did I get a defective product? Should I return it for another one? Please help!

Answer : Thermalright has designed its heatsink base to have a slightly convex bottom. This is not by any means a manufacturing defect. This feature works best on some of the Intel CPUs for their concave surface on the part of the IHS (Integrated Heat Spreader). As long as you have a firm contact between the heatsink base and the CPU, there is no need for concern in regards to swiveling
Click to expand...

However, even TR aren't perfect with their bases. I purchased a TRUE 120 once which had a concave base, which didn't match too well with the Q9550 I was pairing it with that also had a concave IHS. :p

Bottom line, always check all surfaces that are to be mated to heatsinks. ;)
 
Lapping is a without a doubt the best procedure for a performance build or your own machine, but will take more time than the whole rest of a build for a customer's PC, by about 3 or more times probably. I wouldn't expect Intel to expect or even condone the consumer to do that, let alone the OEMs who are the majority of their sales.

That said, my real next question is why the CPU manufacturers stopped have the exposed cores that you put the CPU directly in contact with. It would seem to me to have the least thermal resistance, adding an additional layer in between the core and heatsink can't be helping things.
 
smonska said:
That said, my real next question is why the CPU manufacturers stopped have the exposed cores that you put the CPU directly in contact with. It would seem to me to have the least thermal resistance, adding an additional layer in between the core and heatsink can't be helping things.
Click to expand...

People were cracking the cores installing aftermarket HSF units.
 
Doesn't lapping the IHS void the warranty on a CPU anyway?
 
canna said:
People were cracking the cores installing aftermarket HSF units.
Click to expand...

It was hardly that common back before AMD started using IHSs. I certainly never managed to crack a die and it wasn't a major issue based on threads on this and other forums. With a tiny bit of care you'd have to do something pretty silly or be very unlucky to get it wrong.

This even in the time when installing/removing an HSF meant applying brute force on a tiny metal clip with a screwdriver while praying furiously that it wouldn't slip and bore a hole through the mainboard's PCB :p
 
People were cracking the cores installing aftermarket HSF units.
Click to expand...

Exposed cores would probably work alright on AMD type motherboards with their retention system, as the force applied is applied directly down onto the CPU using a lever. I wouldn't trust using the stock Intel push-pin design with them, with the amount of force variation (all motherboards have slightly different tolerances) and the angular forces that could be applied while pushing in the pins.

As it is, I try not to use the push-pin heatsinks, as I've never liked the amount of warping it causes to the motherboard and the variation of force applied between different motherboards. :p

I think the latest Intel IHS are soldered to the core casings, so the heat transfer loss isn't too bad, and from a company perspective it does make things easier for the end users who can end up being pretty rough and silly with their hardware :S
 
canna said:
People were cracking the cores installing aftermarket HSF units.
Click to expand...

Partially because of the ridiculous clips that AMD heatsinks used. Jesus, I remember having to put all of my weight on a screwdriver to get those things to seat.
 
Chibo said:
Partially because of the ridiculous clips that AMD heatsinks used. Jesus, I remember having to put all of my weight on a screwdriver to get those things to seat.
Click to expand...

lol.. yes... i bought a copper shim that was designed to help prevent the hsf from putting too much pressure on one corner and cracking.


but still i remember being terrified every time I put a hsf on a socket A cpu.

I never wrecked on but several of my buddies cracked the cores of their installing the HSF.
 
Chibo said:
Partially because of the ridiculous clips that AMD heatsinks used. Jesus, I remember having to put all of my weight on a screwdriver to get those things to seat.
Click to expand...

Then again it was the same system on the 462 sockets as with the P4 and PIII sockets of the time, and a continuation of the system which had been used since Socket 3 :)

I do love the current AMD retention system, though. When I first used it with an AM2 system I was delighted at being able to install the HSF with basically a single finger. On the other hand building an i7 920 system with the retail HSF and the push-pin retention system was beyond terrifying. I kept expecting the bloody thing to fall off the moment I put the case up-right and marveled at Intel's ability to market such a fragile system.

I'll take the bulging-muscles-and-screwdriver approach over those push-pins, to be frank. If only for peace of mind :p
 
IntelEnthusiast said:
Glad to hear that you got this taken care of, I havent heard of a problem with the heatspeader problem on any of our processors before so that is a new one for me.
Click to expand...

This is actually a pretty common problem. I am not sure that many people notice it though.

Both my Q6600 and my I7-920 were so unflat it wasn't even funny.

P4 were generally even worse, especially on OEM boxes. One of the most common problems was the CPUs overheating because the thermal paste would seperate right in the center because it was so thick because the IHS was so concave.

There is something not right about whatever machine is used to stamp the IHS's

Generally there will be at least one high spot either near the center or two high spots on either side of the center of the IHS.

The corners of the IHS's generally are higher then the center as well.

It takes a good 2-3 hours of manual lapping to get an IHS flat.. and then you still have to take another 15 minutes or so with higher grit paper to smooth it out.

The first phase goes a lot faster if you are brave and careful enough to be able to use an electric orbital sander.

After lapping, load temps between the cores are a lot more even AND generally at least 5-7c cooler. If you get a really bad one, 10-15c is a closer bet and 20c+ cooler under load can be seen if you get a horrible one.
 
Elledan said:
Then again it was the same system on the 462 sockets as with the P4 and PIII sockets of the time, and a continuation of the system which had been used since Socket 3 :)

I do love the current AMD retention system, though. When I first used it with an AM2 system I was delighted at being able to install the HSF with basically a single finger. On the other hand building an i7 920 system with the retail HSF and the push-pin retention system was beyond terrifying. I kept expecting the bloody thing to fall off the moment I put the case up-right and marveled at Intel's ability to market such a fragile system.

I'll take the bulging-muscles-and-screwdriver approach over those push-pins, to be frank. If only for peace of mind :p
Click to expand...

Apparently you never had one of those CPU socket clips break on you. It wasn't that uncommon, especially if you had a heatsink that oly used the center clip. the ones that used all 3 were the best.. but at the same time expecting the CPU socket to be able to hold the weight and clip tension of the cooler was always iffy to me.

And them having people use flathead screwdrivers to install those things was a really, really bad idea. If you were smart enough to use a small nut-driver so you couldn't accidentially slip, it was a lot safer and not scary at all.

And though I don't much care for the Intel stock LGA775+ retension system, it is a whole lot better then the older style.

It works, and when done properly isn't hard to install at all, and is very secure.
 
cyclone3d said:
Apparently you never had one of those CPU socket clips break on you. It wasn't that uncommon, especially if you had a heatsink that oly used the center clip. the ones that used all 3 were the best.. but at the same time expecting the CPU socket to be able to hold the weight and clip tension of the cooler was always iffy to me.

And them having people use flathead screwdrivers to install those things was a really, really bad idea. If you were smart enough to use a small nut-driver so you couldn't accidentially slip, it was a lot safer and not scary at all.

And though I don't much care for the Intel stock LGA775+ retension system, it is a whole lot better then the older style.

It works, and when done properly isn't hard to install at all, and is very secure.
Click to expand...

The main issue I have with the LGA 775-style push-pin design is that you can not repeatedly remove/install the same heatsink without seriously wearing out those pins and even breaking them. I have never had a clip on the old-style HSFs break on me, and always used a flathead screwdriver with a piece of thick plastic underneath the 'target zone' on the PCB to deal with slippage (never happened, though, I'm just cautious :) ).
 
Elledan said:
The main issue I have with the LGA 775-style push-pin design is that you can not repeatedly remove/install the same heatsink without seriously wearing out those pins and even breaking them. I have never had a clip on the old-style HSFs break on me, and always used a flathead screwdriver with a piece of thick plastic underneath the 'target zone' on the PCB to deal with slippage (never happened, though, I'm just cautious :) ).
Click to expand...

That's not a design issue, but a materials issue.

Aftermarket HSFs (the Hyper212+ is typical) don't use plastic in their construction - also, they don't manufacture as many HSFs as Intel does (Intel doubtless uses plastic to keep the cost per HSF down, as they don't charge for them).

The only design flaw in any HSF using the push-pin design (which Intel has since fixed) was the early usage of 3-wire fan-motors on LGA775 HSFs (C2D and lower).
 
PGHammer said:
That's not a design issue, but a materials issue.

Aftermarket HSFs (the Hyper212+ is typical) don't use plastic in their construction - also, they don't manufacture as many HSFs as Intel does (Intel doubtless uses plastic to keep the cost per HSF down, as they don't charge for them).

The only design flaw in any HSF using the push-pin design (which Intel has since fixed) was the early usage of 3-wire fan-motors on LGA775 HSFs (C2D and lower).
Click to expand...

I see :) That's good to know. I have only seen and used the stock retail HSFs so I wasn't aware that a saner design was being used. Still, I am not sure I'd want the HSF keeping my 100+ Euro CPU from melting down supported by four small points. The AMD stock and bolt-through designs do give me a lot more piece of mind. The two LGA 775 systems I have use the bolt-through system and I wouldn't pick anything less :)
 
You must log in or register to reply here.
Back
Top