Lapping the Thermalright XP-90 Revisited

[MadScientist]

I checked my Thermalright XP-90 for flatness with a borrowed metal machinist's rule and it was prefectly flat. The only reason I lapped it was because it had deep scratches, down to the copper core, in the middle of the part of the heatsink that comes in contact with the cpu. This is the way it came out of the box. I probably should have rma'ed it. Before I lapped it I placed it back on the cpu with AS5, waited an hour and recorded temps.
I used this lapping technique, Link, with 400, 600, 800, 1000, and 1500 3M wet/dry sandpaper and a thick piece of flat glass. I stopped at 1500 grit. It took about 4 hours. It took 2 full sheets of 400 grit to remove the nickel plating, and I used one full sheet for the other grits. Luckily I saved some money because my local Ace Hardware sells in by the sheet. I checked it again for flatness with the steel rule after I was done and cleaned it with 99% isopropanol.
The nickel came off around the edges first. Those over at OCforum see this as an indication that the heatsink is concave. IMO, this is just the way it laps or the nickel is thicker in the middle.
After lapping it, I placed it back on the cpu with AS5, waited 1 hour and recorded temps.

The temps before and after lapping were the same. NO DIFFERENCE.

The nickel plating on the bottom of the copper heatsink must be thin enough to not affect it's performance.

So if you are thinking about lapping your XP-90 or XP-120, if it's flat, DON'T

 

[Fallengod]

Sounds like you had some fun for 4 hours for no reason.

 

[MadScientist]

I was watching the Steelers beat the Bengals so it wasn't a total loss.

 

[GalvanizedYankee]

I retract and apologise for everything i posted on lapping anything, for any reason, at any time. Only a fool would do it.

When my XP-90C gets here tommorow,according to UPS tracking, i will be a fool if it is not dead flat.

The reason the nickel flash came off the edge first is that is what meets the abrasive first.
The human hands natural tendency is to make a sphere. That is why changing position and doing figures of eight is reccomended. Plus nickel is one tough metal. It resists abrasive action very well, not quite as good as tin but close. Breaking the edges is to be done.
AS 5 needs 100 hours to take it's set. After a number of heating and cooling cycles it is drawn into the pores of the CPU and HS base. This did not happen during a quick and dirty test.

As i've read many times YMMV

Galvanized

 

[Greenman]

You should have posted before you did it, could have saved yourself four hours time.

 

[GalvanizedYankee]

Originally posted by: MadScientist
I checked my Thermalright XP-90 for flatness with a borrowed metal machinist's rule and it was prefectly flat. The only reason I lapped it was because it had deep scratches, down to the copper core, in the middle of the part of the heatsink that comes in contact with the cpu. This is the way it came out of the box. I probably should have rma'ed it. Before I lapped it I placed it back on the cpu with AS5, waited an hour and recorded temps.
I used Galvanized Yankee's lapping technique with 400, 600, 800, 1000, and 1500 3M wet/dry sandpaper and a thick piece of flat glass. I stopped at 1500 grit. It took about 4 hours. It took 2 full sheets of 400 grit to remove the nickel plating, and I used one full sheet for the other grits. Luckily I saved some money because my local Ace Hardware sells in by the sheet. I checked it again for flatness with the steel rule after I was done and cleaned it with 99% isopropanol.
The nickel came off around the edges first. Those over at OCforum see this as an indication that the heatsink is concave. IMO, this is just the way it laps or the nickel is thicker in the middle.
After lapping it, I placed it back on the cpu with AS5, waited 1 hour and recorded temps.

The temps before and after lapping were the same. NO DIFFERENCE.

The nickel plating on the bottom of the copper heatsink must be thin enough to not affect it's performance.

So if you are thinking about lapping your XP-90 or XP-120, if it's flat, DON'T

"I used Galvanized Yankee's lapping technique" no longer exists. It has been edited away.
I should have known better. Some ppl have a difficult time using a hammer, so expecting them to be able to take on a task like lapping is way out of the question.

Galvanized

 

[JEDIYoda]

Originally posted by: GalvanizedYankee

Originally posted by: MadScientist
I checked my Thermalright XP-90 for flatness with a borrowed metal machinist's rule and it was prefectly flat. The only reason I lapped it was because it had deep scratches, down to the copper core, in the middle of the part of the heatsink that comes in contact with the cpu. This is the way it came out of the box. I probably should have rma'ed it. Before I lapped it I placed it back on the cpu with AS5, waited an hour and recorded temps.
I used Galvanized Yankee's lapping technique with 400, 600, 800, 1000, and 1500 3M wet/dry sandpaper and a thick piece of flat glass. I stopped at 1500 grit. It took about 4 hours. It took 2 full sheets of 400 grit to remove the nickel plating, and I used one full sheet for the other grits. Luckily I saved some money because my local Ace Hardware sells in by the sheet. I checked it again for flatness with the steel rule after I was done and cleaned it with 99% isopropanol.
The nickel came off around the edges first. Those over at OCforum see this as an indication that the heatsink is concave. IMO, this is just the way it laps or the nickel is thicker in the middle.
After lapping it, I placed it back on the cpu with AS5, waited 1 hour and recorded temps.

The temps before and after lapping were the same. NO DIFFERENCE.

The nickel plating on the bottom of the copper heatsink must be thin enough to not affect it's performance.

So if you are thinking about lapping your XP-90 or XP-120, if it's flat, DON'T

"I used Galvanized Yankee's lapping technique" no longer exists. It has been edited away.
I should have known better. Some ppl have a difficult time using a hammer, so expecting them to be able to take on a task like lapping is way out of the question.

Galvanized

Pardon me....
Something I don`t understand....
#1--- what was the temp before you decided to lapp your heatsink?
#2---did you have AS5 burnt in correctly before checking the temps?
**note IMO AS5 is over rated-it really doesn`t matter**
#3---after lapping did you apply the AS5 and under the exact same circumstances including testing with the same ambient room temp retest the CPU?

 

[MadScientist]

#1--- what was the temp before you decided to lapp your heatsink?
#2---did you have AS5 burnt in correctly before checking the temps?
**note IMO AS5 is over rated-it really doesn`t matter**
#3---after lapping did you apply the AS5 and under the exact same circumstances including testing with the same ambient room temp retest the CPU?

My Athlon64 3200 was running 38C idle before lapping after 200 hours AS5 burn in. The temp will rise to 48-50C under full load.
I tried to keep all variables down to a minumum, so I reinstalled the HS with AS5 before lapping to negate any burn in effect, and kept the ambient (room temp) at 26C. AS5 was apllied in all instances according to AS's instructions on their website.
My system temp ran a constant 28C. I saw no difference in cpu or system temps between the unlapped (burned in), unlapped, and lapped XP-90.

As I stated above, I lapped it only because of the deep scratches on it. The lapping went well. It's still perfectly flat, but I didn't see the temp differences that others over at OCforum were claiming.

 

[JEDIYoda]

so if I may....26c is approx 79 degrees...thats downright hot in a home..lol
then you are saying that your idle temps are only 2c higher than ambient room temp.

The reason I am asking is there is usually in my experiences with heatsinks as well as helping friends there is usually a 4c--5c difference between ambient and CPU idle temp...
unless liquid cooled...

Anyways Im just asking!!

 

[Zap]

Originally posted by: MadScientist
The lapping went well. It's still perfectly flat, but I didn't see the temp differences that others over at OCforum were claiming.

I think the differences are like that of different thermal compound. They range from no difference to one degree of difference. I suppose if the bottom of the heatsink was horrible, may have made a difference.

 

[nealh]

I believe as stated above AS5 is overrated...it is better the stock crap...but I used it on several cpus and never saw a temp drop after 100 hrs or so....

In the future I will use just AS ceramique(which also I never a saw any real drop after long use)

I also follow AS website instructions on placement on my A64s and chipset

lapping always seemed like alot of work too me

 

[MadScientist]

Originally posted by: JEDIYoda
so if I may....26c is approx 79 degrees...thats downright hot in a home..lol
then you are saying that your idle temps are only 2c higher than ambient room temp.

The reason I am asking is there is usually in my experiences with heatsinks as well as helping friends there is usually a 4c--5c difference between ambient and CPU idle temp...
unless liquid cooled...

Anyways Im just asking!!

79F is comfortable if the humidity is low. My inside case temp, or motherboard temp as Everest reads it, is 28C -26C = 2C higher than ambient (room) temp. The cpu idle temp is 38C - 26C = 12C higher than ambient temp, 10C higher than case (motherboard) temp.
I have an Epox 9NPA+ Ultra motherboard. I do not know how accurate the cpu temp sensor is on this board or if 38C is a normal idle temp for the Athlon64 3200. My system runs fine, no crashes or BSODs.

What I was primarily looking at was the relative temp difference between the unlapped and lapped heatsink.

I realize that my case cooling may be inadequate. I have, I can hear the laughter, an old modded Enlight 7237 case. It's built like a battleship and has some sentimental value. It's been through 4 builds, but I think it's time to retire it. I have a Lian Li PC-7A Plus on order.

I also agree that AS5 is overrated. I purchased the retail version of the XP-90 that came with Thermalright's heatsink paste. It appears to be a silicone/zinc oxide based paste. I used it before I purchased the AS5 and it gave me cpu idle temps 2C lower than the AS5. I'm going to give the Thermalright paste another try when I change out cases.

I don't advocate lapping your heatsink unless you are 100% certain that it needs it. If you do, these instructions are the best I have found.
Lapping Instructions

 

[BonzaiDuck]

Question submitted to ThermalRight Tech-support a month or two previous:

HERE?S MY QUESTION. A minor matter. I?m getting ready to install the SI-120 on a Prescott 3.2E processor. I want to squeeze the last bit of cooling I can from the sucker. There are very fine scratches on the bottom of the heatsink base. I understand that this new base is copper with a nickel coating.

I cannot remember for sure exactly what others said ? I seem to recall a remark that the nickel-plating was meant to discourage corrosion. I can?t be sure, so I?m asking you: ?Can I lap-sand the bottom of the SI-120 heatsink base with 400-wet-or-dry sandpaper and water??

Please confirm. I?m getting ready to install it, and I?ve been debating about this final step before I go forward.

ANSWER from "Bob" at ThermalRight:

Hi,

DO NOT LAP!! The nickel coating allows for the soldering of the various parts that make up the SI-120. What is important is that the base be flat, in the old days we did not have heat spreaders between our heatsinks and the core of the CPU's, thus the base was more critical.

For Thermalright,

Bob

 

[MadScientist]

From a one of my previous posts:
Thermalright finally got back to me on why they nickel plate the XP-90 and XP-120. Their response was "for aesthetics only". They also mentioned that the heatsink's performance with and without the plating was negligible.
This was from, I'm assuming, the same Bob from Thermalright.

The nickel plating is also on the side of the heatsink on the XP-90 where the clips are soldered. Lapping the thin nickel plating off the bottom of the XP-90 did not even come close to where the clips are soldered on to the side. IMO, it looks like Thermalright soldered the clips and heatpipes to the copper heatsink core and then nickel plated the sides and bottom of the core for appearance.

The SI-120 though may be assembled differently.