Very surprising results!!
LAPPED TESTING
1) INTRODUCTION
After the unlapped (original) testing was carried out back in March 2011 involving MX-4, IC Perihelion and IC Diamond thermal compounds, Q6600 and IFX-14 heatsink were both hand lapped and same testing methodology was implemented and same BIOS settings (vcore, ram speed etc).
This is because it was found from unlapped contact test that the contact area (mating surface) between Q6600 IHS and IFX-14 base was quite narrow and formed an
I shaped pattern. Hence lapping was done to increase the overall contact area and to see if that made any difference to core temperatures and how each thermal compound performed in lapped scenario. However this also affected the mounting force and pressure heavily. More information can be gained by clicking on the links as follows:
UNLAPPED information
-Post 291
http://forums.overclockers.co.uk/showpost.php?p=18814406&postcount=291
-Post 346
http://forums.overclockers.co.uk/showpost.php?p=18941593&postcount=346
-Post 388
http://forums.overclockers.co.uk/showpost.php?p=19118209&postcount=388
LAPPED information
-Post 390
http://forums.overclockers.co.uk/showpost.php?p=19133860&postcount=390
-Post 406
http://forums.overclockers.co.uk/showpost.php?p=19315910&postcount=406
Q6600 Lapped
Thermalright IFX-14 Lapped
The lapped testing was carried out in June 2011 and following are the application and testing dates for each thermal compound:
IC PERIHELION
04/06/2011(Application date) - 12/06/2011(Testing date)
MX-4
13/06/2011(Application date) - 19/06/2011(Testing date)
IC DIAMOND
20/06/2011(Application date) - 30/06/2011(Testing date)
In the case of IC Diamond the testing had to be forwarded further by 3 days as temps went as high as 7C compared to the testing days of IC Perihelion and MX-4. Plus I had exam in the way so couldn't do testing on the target date of 27 June.
Luckily the day temperatures on 30 June were more or less similar to the testing days temps of IC Perihelion and MX-4
.
2) RESULTS
IC (Innovation Cooling) recommends 5-5.5mm pea-sized blob in the middle. However during unlapped testing I found the recommended size insufficient for covering the whole of LGA 775 Q6600 IHS. Although the recommended size did cover most of IHS.
So for lapped testing I have applied slightly more of 7mm blob in the middle and it will be seen that this amount covered the entire IHS especially in the case of MX-4 and IC Diamond.
IC PERIHELION
7-7.5mm Pea-Sized Blob
TEST1 (Afternoon)
Ambient 19.0C
TEST2 (Evening)
Ambient 19.1C
TEST3 (Night)
Ambient 19.5C
Spread Pattern after Tests completion
Spread pattern on IFX-14 base
MX-4
7mm Pea-Sized Blob
TEST1 (Afternoon)
Ambient 19.8C
TEST2 (Evening)
Ambient 20.0C
TEST3 (Night)
Ambient 20.0C
Spread Pattern after Tests completion
Spread Pattern on IFX-14 base
IC DIAMOND
7mm Pea-Sized Blob
TEST1 (Afternoon)
Ambient 20.2C
TEST2 (Evening)
Ambient 21.0C
TEST3 (Night)
Ambient 21.1C
Spread Pattern after Tests completion
Spread Pattern on IFX-14 base
These have been very surprising results indeed!!
Compared to Unlapped raw results and without taking into account ambient for the time being;
IC PERIHELION
-Shown greatest reductions in load temps especially on the hottest cores with as much as 10C !!
-On the coolest cores the reductions was something like 2C overall during load
-Idle temps reduction was something like 1-2C
MX-4
-Also showed impressive performance with load temps reduced by 6-7C on hottest cores
-On the coolest cores the load temp reductions was something like 2-4C
-Idle temps reduction was around 1-1.5C
IC DIAMOND
-Surprisingly for
[email protected], IC Diamond didn't show as much improvements as it did for
[email protected].
-On the hottest cores the load temp reductions were around 1-2C only
-Surprisingly the coolest cores actually displayed increased temps of around 2C
-Idle temps were more or less same with slight decrease
However the above analysis is only rudimentary and we move into
Analysis section to gain better insight and understanding.
3) ANALYSIS
Comparing MX-4 unlapped vs lapped:
-Overall Delta T Idle improved by 1.05C after lapping
-Overall Delta T Load improved by 4.72C after lapping
Comparing IC Perihelion unlapped vs lapped:
-Overall Delta T Idle improved by 2.3C after lapping
-Overall Delta T Load improved by 6.22C after lapping
Comparing IC Diamond unlapped vs lapped:
-Overall Delta T Idle improved by 0.07C after lapping
-Overall Delta T Load improved by 1.40C after lapping
Converting the above tables into graphs
Graph1
In
graph1 IC Perihelion and IC Diamond seem to be on a similar level load testing wise while in idle IC Diamond seems to have given higher temps. IC Perihelion seems to beat MX-4 in idle temps ignoring ambient. However in load testing MX-4 comes on top.
Graph2
In
graph2, in terms of the difference between average load and average idle, both IC Diamond and MX-4 even out each other while IC Perihelion gives higher difference. Regarding the temp difference between Average Idle and ambient, IC Diamond seems to produce higher values overall while IC Perihelion and MX-4 seem to perform more or less similar.
Graph3
In
graph3, IC Diamond seems to give out slightly higher Delta T Idle value overall while IC Perihelion has increased Delta T Load values compared to the other two thermal compounds. MX-4 comes on top in Delta T Load reduction.
Graph4
One thing very striking in
graph4 is that the performance gap between the three thermal compounds has reduced significantly compared to unlapped testing. Now I can't be sure if this is due to improvements in contact or reduction in mounting pressure overall due to lapping.
As stated on innovation cooling website that there are 3 areas for troubleshooting performance of a thermal compound:
1) Pressure
2) Contact
3) Application (of thermal compound)
Application
One thing that can be noticed is that if we only compare raw lapped tests then overall there doesn't seem to be a lot of performance difference between the three thermal compounds. Also from pics above it can be seen that thermal compounds covered the entire IHS with IC Perihelion almost covering it. This may be due to slight mounting error or the way thermal compound spread out under pressure and heat.
MX-4 again seems to behave slightly differently with showing regions of no compound layer in the middle and that is where contact makes the least impact on Cpu IHS. This is strange phenomena.
However since all three thermal compounds cover the overall Cpu IHS very well, I don't think
application will be a major factor that is hindering performance of thermal compounds.
Contact
However one thing that has been stated in this thread is that if the contact and pressure are bottlenecking (less contact and low pressure) the rapid heat transfer from Cpu IHS/Heatsink Base to heatsink pipes or in words creating bigger
Heat Dam, then the performance difference between the thermal compounds may well be only 1-2C.
Referring to
Post 406 in the
Introduction, the 4 cores of Q6600 reside in the area of greatest contact and pressure. Since cores are the dominant factor in generating heat especially during load, this means being in the right place will result in greater transfer of heat thus reducing
heat dam. So the
contact area of Q6600 cores seems pretty good plus the contact area has become broader overall after lapping and achieving
close to contact margins as stated by IC Diamond. However it may be noted that if we had different contact pattern after lapping, then thermal compounds again may well behave differently.
Pressure
We now turn to the largest dominant factor;
Mounting Pressure.
Unlapped Testing,
Post 388:
Average Mounting Pressure =
69.90 PSI
Mounting Force =
38.68 lbf
Lapped Testing,
Post 406:
Average Mounting Pressure =
53.58 PSI
Mounting Force =
20.53 lbf
After lapping the average mounting pressure has been reduced by around 23% while the mounting force has almost halved!! This is significant.
This is mainly due to shaving of metal during lapping so now
slack has occured. It has been discussed that one method to alleviate this reduced pressure is to use
washer as part of mounting system. It has also been suggested that increased pressure should also yield an increase in the size of contact profile.
The reduced pressure may well have affected IC Diamond performance although MX-4 showed greater gains. This may well be due to MX-4 being less viscous and thinner thus adjusting to even lower pressures. IC Diamond being a thicker compound may well require more pressure in order to perform optimally. I think IC Perihelion also benefitted more from improved contact profile than from pressure.
However I think there may be other factors that may well have affected IC Diamond performance and also possibly other thermal compounds.
Weather
During the day of testing for both MX-4 and IC Perihelion, it was raining outside for the whole time. Although this didn't affect the ambient temps much, but overall the environment felt cooler. Since room window was only open to 0.8inch from closing position as this is part of procedure, a lot of air couldn't move into the room.
However it was noted that rainy weather accompanied moderate winds and this would have pushed some cooler air into the room. Thus case fans and heatsink fans also drawing in colder air giving advantage to both MX-4 and IC Perihelion.
The nature of weather, the wind speed and direction are the uncontrolled factors and can affect the performance of thermal compounds.
In the case of IC Diamond, the weather fluctuated throughout the day from being sunny to cloudy to moderate rain accompanying winds to some extent.
Dust
The coolermaster HAF series especially HAF 922 (my case) and HAF 932 are known as dust magnets and tend to attract a lot of dust in short space of time. It is possible that dust build up on fan profiles impeded airflow thus affecting performance.
Fan speeds
In the
Post 291 I gave a general outline of fan speeds during idle and load. We also saw that fans never spin at the same speed and that there is always some give or take fan rotations from the average fan speed. Although this shouldn't affect the performance as the loading is same; nonetheless this factor can't be ignored.
However to summarise the unlapped and lapped testing so far, the results from
graph4 for both testing are presented here:
UNLAPPED Testing
LAPPED Testing
The above values correspond with what I stated when analysing table of results for lapped vs unlapped for overall delta T idle and overall delta T load
4) CONCLUSION
Lapped testing has been very interesting indeed and seems to produce results quite different from unlapped testing. Lapping has had most positive influence on IC Perihelion where hottest core temps reduced by 8-10C without taking into account ambient. This pattern was then followed by MX-4. IC Diamond seems to have been least affected by lapping for
[email protected]. Although for
[email protected], it produced more impressive results. May be that was one off or IC Diamond works better even in more hot conditions.
From the lapped testing it can be seen that
application of thermal compounds in each case was very good as thermal compounds managed to cover almost or all of CPU IHS thus 7mm of blob proved to be more useful than the recommended 5-5.5mm.
Lapping also had positive impact on
contact profile in that it managed to get broader overall and also yielding 'close to contact' margins. Contact was strongest in the area where Q6600 cores reside i.e. in the upper half of the CPU. Thus facilitating improved heat transfer.
The pressure seems to be the weakest link as the dominant factor as lapping not only reduced mounting pressure by about 23-24% but the mounting force itself almost halved.
It has been discussed that placing washer as part of IFX-14 mounting system should help in increasing the pressure and reducing the slack.
As stated weather along with other factors may well have affected IC Diamond performance and this could be true for other thermal compounds aswell.
However overall the performance gap between the thermal compounds have reduced significantly with only 0.91C between IC Diamond and MX-4, with MX-4 coming on top in this investigation followed by IC Diamond followed by IC Perihelion. MX-4 is still a top contender and IC Perihelion still can't seem to top MX-4 despite performance gap closing in.
As the testing involves plenty of numbers, it is likely I may have made a mistake somewhere. Please let me know.
This paves the way for
Washer Modded Lapped Testing
Many thanks
WingZero30
.
and this is before Core 2 series made an appearence.
