• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

lapping my Ultra-120 Extreme (pics and temp data)

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Yeah TAT sucks really. You should give RMClock a try, it is very good and offers lots of logging options. I'd say your temps are good, but if you want to drop them more, I'd suggest you lap your IHS on the CPU as well as the HS. I knocked off 7C on my coolest core and 10 C on my hottest core (q6600 @ 9x333).
 
Thanks for the tip about RMClock, Graysky. I'd not heard of it.

I'll probably hold off on any lapping for now, especially as I've now got everything reinstalled and running, but your results merit keeping an eye on things and considering that option if I see temps to ever be a problem.

Something that puzzles me...I see RMCLock reports a minimum Vcore of 1.325v required for my 3.6GHz overclock. That's the first time I've seen any tool report what value I supposedly should be using at a minimum, but I've seen that in spite of having Vcore set to 1.4v in my BIOS, most any monitor utility is reporting Vcore (CPU Core) as a lower value than what I defined...1.375-1.38v being typical. I think I've been mistakenly assuming VID and Vcore are the same parameter, but a quick search on Google suggests otherwise. But, this leaves me to wonder if I could actually reduce Vcore to 1.325v from my current 1.4v and still have stable operation. That should also noticeably improve my peak temps under load if so. One way to find out...


Aha!...question answered an my memory is refreshed...I found I couldn't boot without a device error BSOD at 1.325v on Vcore, and oddly I couldn't post, so I then I upped Vcore to 1.35v. At 1.375v, I booted fine but then failed 100% TAT dual-core loading after about 6 minutes...that's what jogged my memory, of seeing that consistently happen yesterday. But, returning to a 1.4v Vcore, I should be stable under load again.

Thanks,

Daryl
 
Originally posted by: graysky
Vdroop.. google it. Most boards have it.
Click to expand...
OMG! No kidding!!!

I run a high-end DFI LanParty mobo. @ 1.70v my board droops to 1.66v -- I gotta push 1.75v to get 1.70ish. 😛

Anyway, what I wanted to tell you was...

Whenever I lap something, I finish it off with Simichrome Polish. Google that! 😀

You can usually find it locally at speed shops and/or custom motorcycle dealers.

Simichrome will give you that desirable mirror finish...
 
It does occur to me that the heatsinks have a convex base that matches up perfectly with the concave CPU heatspreaders Intel is using.

All the x-marker pics I have seen show that CPUs are high around the edges, and the Ultra120 extremes are high in the middle. Since the CPU core is in the middle of the heatspreader, it is really most important to have best contact there.

I think that anyone who laps one and not the other, is likely to see not much improvement.

1-2C variance can easily be explained away by a different mount. If bad contact was due to the shape of the sink, then lapping should make a bigger difference in temps.
 
Originally posted by: Noubourne
It does occur to me that the heatsinks have a convex base that matches up perfectly with the concave CPU heatspreaders Intel is using...
Click to expand...
Bwahahahaha!

That's B.S. but nice try! 😀

Look, the reason Intel CPUs are concave is because they solder the chip to the lid, so the whole center is indented, and the outer edges are raised -- almost in a perfect geometric pattern! They are truely concave, like a basin!

Ultra-120 eXtremes are convex 'cause the heatpipes aren't soldered to the base correctly, so they are convex in straight lines across the base, following the path of the pipe. This makes for irregular patterns, like raised ridges.

These 'basins' and 'raised ridges' hardly match up, even in a perfect world. Mix a concave C2D chip and a convex Thermalwrong eXtreme HSF, and you have created a perfect hell -- the worst of all possible worlds!


Originally posted by: Noubourne
I think that anyone who laps one and not the other, is likely to see not much improvement.
Click to expand...
This we can agree on... 😉
 
Originally posted by: VinDSL
Ultra-120 eXtremes are convex 'cause the heatpipes aren't soldered to the base correctly, so they are convex in straight lines across the base, following the path of the pipe. This makes for irregular patterns, like raised ridges.
Click to expand...


Where do you find evidence that the heatpipes aren't soldered correctly, and that the base is irregular?

Are you aware of the fact that Swiftech is now manufacturing special bases for their waterblocks that bow out in order to create better contact with the consistently concave Intel C2Ds and quads?

Do you think that IBMs recent research may have inspired the new base texturing on this heatsink, or is it your opinion that the manufacturer is unaware of the tiny ridges introduced by the manufacturing process, and rather than rectify the textured base, they choose instead to nickel plate it?

Because you would need to show some evidence and do a lot more research in order to prove that none of these features of the Thermalright design are intentional.
 
Thermalright certainly contends that the slightly askew base of the U120X is intentional:
  • "According to our engineers who tested the heatsink on several processors, convex base fits better with our bolt-thru mounting."
 
Originally posted by: cyburzaki
Thermalright certainly contends that the slightly askew base of the U120X is intentional:
  • "According to our engineers who tested the heatsink on several processors, convex base fits better with our bolt-thru mounting."
Click to expand...
Link?

 
@VinDSL: direct email from Chris Lee at Thermalright (judging by the time stamp, I believe he's in the Taipei office).
 
So far as I am aware, everyone who has tried to lap the TR U-120X has met with calamity initially. I'm not saying that there haven't been issues with this model, because there have been reports of damaged product and fit issues as well. It appears that TR has already altered the mounting system to provide a tighter fit. BTW, I also read on ExtremeOC that someone contacted TR about the convex base and stated that it was by design.
 
Bah! 😛

Phantom authorities that aren't part of the conversation and can't be questioned...

Unfinished, convex bases with machine marks and scratches all over them, supposed designed this way on purpose...

Ad hominem attacks on anyone that questions the quality control of the latest Tuniq Tower Killers -- the demigods at Thermalright...

Sorry, but I'm calling BS on this whole discussion!

See you guys in the next thread... 😀
 
I don't know, I see a lot of speculation in your posts Vin. To call the bottom "Unfinished" isn't even accurate. They're nickel plated after being cut and soldered.

If they intended to lap it because they determined that their cutting process was ineffective at producing an effective cooling surface, they would do so before the nickel plating.

I have yet to see a photo of a covered brand new cooler that has scratches on the bottom either, as you claim.

I agree that there is a textured base. I do not agree that this is necessarily a flaw and hampers cooling performance. It MIGHT be, but I would not agree until I saw proof, and so far no lappers are showing the kind of temperature improvements that would clearly point to this as a problem. For all you know, those channels and grooves in the base assist in spreading the TIM. That IS what IBM's engineers found, and I think you'll see that on the latest Power 6 CPUs, and if not, then probably future AMD chips.

Is it your contention that the unit Anandtech tested performed so well because it was manufactured using a different process from the retail units? I believe they actually did a review of a retail unit to show that their engineering sample was of the same quality, and found it to be worthy of their high rating.

No one is posting ad hominem attacks - I'm not sure that means what you think it means. Ad hominem would be attacking you by saying "VinDSL, you're stupid and you're wrong". No one is doing that. This is a discussion about whether or not Thermalright product in retail exists as designed, or has QC issues, which you seem to be implying about pretty much every aspect of the cooler (base flatness, base texture, heatpipe/base contact, soldering). I, and now a couple others it seems, are simply questioning where you see evidence for this. I don't think those are bad questions to ask.

I would suggest the fact that they have a team of engineers lab testing the product and then shelling out a ton of cash to have it manufactured makes it a fairly safe assumption that some, if not all, of these features are intentional.

If any of the symptoms of the "problem" (I don't concede that there are problems until I have been shown proof) are due to QC issues, then one would assume that not all products would show the same result. Based on the pics posted of lappers, I would venture to say that the product seems very consistent. It could be a manufacturing issue, but as I said before Anandtech found the retail unit to perform as well as the first sample they received... so while it is still a possibility, I would not put a lot of money on it myself.

As for your implication that we are in some kind of cult worshipping TT's products, well, I'm sporting a Scythe Infinity on my Opty, and I was disappointed in the mounting mechanism of the Ultra 120 (non-Extreme), and if you want to find some criticism of their products by me, you are welcome to do a quick search and you'll find it right quick.

However, I do happen to notice that you mention Tuniq Tower (don't remember anyone else bringing that up), and you happen to be running one of those coolers in your own machine (at least according to the link in your post) so perhaps your own motivations for posting should be more closely examined before you start calling foul on others, and shouting "BS!".
 
I have this
http://www.ncix.com/products/i...nufacture=THERMALRIGHT

When it arrived I checked the base and I could see ridges from where they grinned it, when placed on a piece of glass it rocked back & forth and finally the base was painted black as well. So I had no option but to lap it.

Started with 220 grit dry sandpaper until 95% of the copper was exposed. The method of sanding I used was to grip the base with my fingers, not insert my palm under the HS, pushed and pulled in a string line until all the scratch marks when in one direction when looked at under a light, rotated 3 to 5 times before moving on after the 220 grit. The grits I used where 220 - 400 - 600 - 800 - 1000 - 1500 - 2000. At first I used a 1/8'' thick sheet of glass but I saw that was warping so I switched to mirror as that was much thicker with less flex (the face is regular glass). When I was done I had a mirror finish but there where still tiny scratches visible on the surface (IIRC you need a buffing wax to get a perfect finish)

Next I lapped the CPU (AMD AM2 socket). Using the original case + foam it came with (IIRC the intel chips have a better backing to use with lapping) I started with 800 grit sandpaper, found I had too much material to remove from the edges and went to 600 grit. Not knowing how thick the HS on the CPU itself was I removed the bare minimum until it looked flat when checked with a razor blade. This left 70% of the silver layer. The method of sanding was 40 forward & backward strokes in the same direction, rotating 90 degrees until I did a 360. I did this 3 to 4 times with each grit of sandpaper (after the 600). I used 600 - 800 - 1000 - 1500 - 2000 grit sandpaper and was left with a mirror finish similar to the HS.

Sanding proved the center of both the CPU and HS where raised with sunken areas around the center, basically a mirror image that would have resulted in horrible contact.

Using HW monitor, my temps are now ~34c idle, ~39c under load using the crysis benchmark & 2 hours of playing the game (XP high settings, 1900x1200, 16xQ AA) using two S-Flex 120mm fans @ 1200 RPM + P182 case. I did not try my HS before lapping, though my stock cooler ran at ~45c under load with the odd spike to 53c that I could not explain, I forget the idle temps.

-edit- , hmmm, 3 pages you say?
-edit- , hmmm, almost 2 years you say?
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top