TRUE -> Lapped = Finally Working!

DirkVoodoo

Well, a small update to those who were paying attention all those months ago.

When I took delivery of my Thermalright Ultra 120 Extreme, I was shocked by the performance.

The base was completely convex, ants could have played see-saw on the razor blade I placed on the hump!

So, I decided to lap it, no pretty pictures im afraid. Its still slightly convex and I didnt bother to give it a mirror finish (more of a pre-lap job) but the results were amazing.

Convex TRUE: Idle -> 32C Load (Running Orthos for 20 mins) -> 53C (CPU @ 3GHz)
Lapped TRUE: Idle -> 27C Load (Running Orthos for 40 mins) -> 41C (CPU @ 3GHz)

Hardly scientific, I'm assuming my room temp to be approximately the same, but there is definitely a big impovement, especially at load. I also ran speedfan in both instances with CPU fan on max.

Anyway, if you have one and are not happy, definitely lap it but lap it well, it seems the internet is full of people not aware of "how" to lap and botching their heat sinks!

awhir

DirkVoodoo wrote: »

Well, a small update to those who were paying attention all those months ago.

When I took delivery of my Thermalright Ultra 120 Extreme, I was shocked by the performance.

The base was completely convex, ants could have played see-saw on the razor blade I placed on the hump!

So, I decided to lap it, no pretty pictures im afraid. Its still slightly convex and I didnt bother to give it a mirror finish (more of a pre-lap job) but the results were amazing.

Convex TRUE: Idle -> 32C Load (Running Orthos for 20 mins) -> 53C (CPU @ 3GHz)
Lapped TRUE: Idle -> 27C Load (Running Orthos for 40 mins) -> 41C (CPU @ 3GHz)

Hardly scientific, I'm assuming my room temp to be approximately the same, but there is definitely a big impovement, especially at load. I also ran speedfan in both instances with CPU fan on max.

Anyway, if you have one and are not happy, definitely lap it but lap it well, it seems the internet is full of people not aware of "how" to lap and botching their heat sinks!

nice one bud.i also noticed a large drop in temps when i lapped my tunqi.what grit did u go upto also try lapping ur chip.

DirkVoodoo

Like I said it was a very rough job (no pun intended).

I started at 600 grit in the UCD surface engineering lab. I don't know exactly what speed the grinding wheels were moving at but they removed all the nickel in a few minutes.

I then went by hand using 600, 800 and 1000 grit. I should have started manual grinding at 400 or lower as it was pretty slow this way. The surface is nearly planar. I also did the "dull smear of AS5" approach, mainly to offer some sort of oxidation protection to the copper surface. I noticed when I brought it home fom college first, the surface was already quite oxidised and some greasy fingerprints had resulted in preferential oxidation.

The chip was already pretty flat, but I may lap it just for kicks and I like the way they look when lapped to a mirror finish.

I have the easypckit premium kit which goes all the way up to 10 micron, so it would be a shame not to use it, i'll even post pics this time

To anyone with a TRUE and a convex base, don't believe thermalright's excuse about it being intentional, it is total rubbish! They warp it when attempting to weld the heat pipes to the base then machine it, it is not a curved base through machning!

nibble

I'm quite surprised at this to be honest, the bow really should provide better performance. Both the Apogee GTX and D-tek Fuzion waterblocks show significant improvements in temps with a bow versus unbowed. The bows aren't machined in these either they're bent.

DirkVoodoo

From the thermalright Ultra 120 FAQ:

Question 5:

The bottom of the Ultra-120 eXtreme isn’t very shiny. Would I get better results if I lapped it?
Answer:

A shiny surface does not necessarily mean a flat surface. Lapping only eliminates some of the larger gaps between the grains. We have demanding criteria for high quality especially when it comes to our heatsink base. The Thermalright heatsink you purchased does have a relatively flat surface therefore we do not recommend risking damage to your heatsink or your CPU DIE by lapping because results may not always be positive.

Yet they then said that the curvature was intentional when people started complaining.

DirkVoodoo

Also, according to apogee the GTX is 3mm to make it more bendy, so pressure in locking it down allows it to flex and mate better with the heat spreader. The TRUE does nothing of the sort and the curved base simply sits on top of the IHS, even with the bolting force applied.

I'm saying that thermalright simply failed to heat treat the bases, resulting in warping which they are now passing off as this precision machined base you see on waterblocks. Otherwise performance SHOULD drop when people lap their TRUE, and for the vast majority it improves.

Anti

1000grit ??? Jesus thats not enough for lapping. shoulda gone to 4000/8000 or even 2000 would do fine though. But its pretty hard to find.

Here is a semi decent kit: http://www.kustompcs.co.uk/acatalog/info_3714.html

L31mr0d

what kit did you get off easypckits? The premium one? Why didn't you use the higher grit paper?

DirkVoodoo

The premium kit.

As mentioned above, I got bored..it was a Sunday afternoon and time was dragging on. I'm gonna finish it off, I just wanted to see was there any difference in temps.

MooseJam

Anti wrote: »

1000grit ??? Jesus thats not enough for lapping. shoulda gone to 4000/8000 or even 2000 would do fine though. But its pretty hard to find.

Here is a semi decent kit: http://www.kustompcs.co.uk/acatalog/info_3714.html

thanks for the link, ordered a kit there, can't go wrong for 20 beans, i'm gonna turn my silent knight into a mirror