Overclock.net › Forums › Cooling › Air Cooling › Lapping a Thermalright Ultra 120 and an AMD Opteron 170’s IHS
New Posts  All Forums:Forum Nav:

Lapping a Thermalright Ultra 120 and an AMD Opteron 170’s IHS - Page 2

post #11 of 17
Thread Starter 
Step 4: Lapping the Opteron 170’s IHS

The Opteron’s IHS didn’t require as much effort to smooth out. I started dry lapping with the 280, then moved on upwards. The procedure really didn’t differ compared to what I’d done with the Ultra 120’s base. As with the HSF, I stopped at the 1000 grit level.

After completing lapping both contact surfaces and being satisfied with their flatness, I put the CPU back in its socket, and mounted the HSF onto the CPU. For the record, I used Thermalright’s provided TIM, which is much smaller-grained and is far easier to spread over the IHS of the Opteron. I’m theorizing that the lapped surfaces, with their improved smoothness, would be better served with a smaller-grained TIM than the much more popular Arctic Silver 5 or the Shin-Etsu TIM I used to have.

A note, however: The change in TIM may skew the results somewhat. This is obviously a change of one of the variables in the equation. The more effective move may have been to compare results with just using Shin-Etsu TIM in both lapped and unlapped configurations, but I decided to “cheat†by a step and go with the Thermalright TIM right after lapping.
    
CPUMotherboardGraphicsRAM
Phenom II X3 720 BE Gigabyte GA-MA790FXT-UD5P Sapphire Radeon HD 3870 2 x 2GB Crucial Ballistix Tracers PC3 10600 
Hard DriveCoolingOSPower
Almost 8TB of storage AMD Stock Cooler Windows Home Server CoolerMaster SilentPro M 700W 
Case
Lian Li PC-V2000B (with thlnk3r mods) 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
Phenom II X3 720 BE Gigabyte GA-MA790FXT-UD5P Sapphire Radeon HD 3870 2 x 2GB Crucial Ballistix Tracers PC3 10600 
Hard DriveCoolingOSPower
Almost 8TB of storage AMD Stock Cooler Windows Home Server CoolerMaster SilentPro M 700W 
Case
Lian Li PC-V2000B (with thlnk3r mods) 
  hide details  
Reply
post #12 of 17
Thread Starter 
Step 5: The results

So now we’ve come to the moment of truth: Was the exercise worth it? How effective is lapping for reducing operating temperature readings?

Here are my results:

Ambient temperatures: 23 degrees Celsius (74 degrees Fahrenheit), estimated
CPU Idle Temperatures (Core 0 – Core 1): 33 degrees Celsius – 32 degrees Celsius (idle temperature gathered 30 minutes after initial boot-up into WinXP Home SP2)
CPU Load Temperatures (Core 0 – Core 1): 54 degrees Celsius – 51 degrees Celsius (with spikes to 58 degrees on Core 0) (load temperatures attained through a 30-minute OCCT run)

CPU temperatures were measured with SpeedFan, which was calibrated to CoreTemp .95 readings. I prefer CoreTemp .95, but the program (even in older .94 spec) crashes my machine.

Conclusions

Lapping resulted in a measurable temperature decrease in similar ambient temperature conditions and with all other variables identical to the initial baseline measurements taken at the onset of the experiment. A 3 – 4 degree Celsius drop may not seem significant, but it’s still a reduction in heat dissipation.

Is it worth the effort? Only the end user can answer that question, really. For me, personally, the answer would have to be yes. This is especially true with misshapen contact surfaces which reduce the efficiency of heat transfer, and therefore heat dissipation. Eliminating these deformations on the contact surfaces should result in improved efficiency for your cooling system. This was certainly my finding.

I hope this has proved to be helpful to someone.

Good luck with your own lapping adventures!
    
CPUMotherboardGraphicsRAM
Phenom II X3 720 BE Gigabyte GA-MA790FXT-UD5P Sapphire Radeon HD 3870 2 x 2GB Crucial Ballistix Tracers PC3 10600 
Hard DriveCoolingOSPower
Almost 8TB of storage AMD Stock Cooler Windows Home Server CoolerMaster SilentPro M 700W 
Case
Lian Li PC-V2000B (with thlnk3r mods) 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
Phenom II X3 720 BE Gigabyte GA-MA790FXT-UD5P Sapphire Radeon HD 3870 2 x 2GB Crucial Ballistix Tracers PC3 10600 
Hard DriveCoolingOSPower
Almost 8TB of storage AMD Stock Cooler Windows Home Server CoolerMaster SilentPro M 700W 
Case
Lian Li PC-V2000B (with thlnk3r mods) 
  hide details  
Reply
post #13 of 17
Thread Starter 
Will add pictures tomorrow, when I disassemble the cooling system yet again for another personal experiment.

Thanks for reading!
    
CPUMotherboardGraphicsRAM
Phenom II X3 720 BE Gigabyte GA-MA790FXT-UD5P Sapphire Radeon HD 3870 2 x 2GB Crucial Ballistix Tracers PC3 10600 
Hard DriveCoolingOSPower
Almost 8TB of storage AMD Stock Cooler Windows Home Server CoolerMaster SilentPro M 700W 
Case
Lian Li PC-V2000B (with thlnk3r mods) 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
Phenom II X3 720 BE Gigabyte GA-MA790FXT-UD5P Sapphire Radeon HD 3870 2 x 2GB Crucial Ballistix Tracers PC3 10600 
Hard DriveCoolingOSPower
Almost 8TB of storage AMD Stock Cooler Windows Home Server CoolerMaster SilentPro M 700W 
Case
Lian Li PC-V2000B (with thlnk3r mods) 
  hide details  
Reply
post #14 of 17
Thread Starter 
I wasn't too thrilled with the temperature readings I reported in my last post, so I decided to redo the HSF mount.

I applied a little less Thermalright TIM this time around (about a quarter of an uncooked grain of rice, for those familiar with Arctic Silver 5's famous instruction), spread it over the IHS with an old credit card, then reseated the Ultra 120.

Here are the updated results:

Estimated ambient temperature: 22.78 degrees Celsius (73 degrees Fahrenheit), estimated
CPU Idle Temperatures (Core 0 - Core 1): 33 degrees Celsius - 32 degrees Celsius
CPU Load Temperatures (Core 0 - Core 1): 49 degrees Celsius - 47 degrees Celsius

Here, then, is a direct comparison between pre-lapped and post-lapped temperature readings:

CPU Idle (Pre-lap) (Core 0 - Core 1): 37 degrees C - 37 degrees C
CPU Idle (Post-lap) (Core 0 - Core 1): 33 degrees C - 32 degrees C
Temperature delta (Core 0 - Core 1): 4 degrees C - 5 degrees C

CPU Load (Pre-lap) (Core 0 - Core 1): 57 degrees C - 55 degrees C
CPU Load (Post-lap) (Core 0 - Core 1): 49 degrees C - 47 degrees C
Temperature delta (Core 0 - Core 1): 8 degrees C - 8 degrees C

Methods of measurement were kept identical, as were the methods of achieving these temperatures. Therefore, we can eliminate two variables from the analysis of the experiment.

Ambient temperatures, of course, are only best guess estimates, as I do not actually have a means to measure my bedroom's temperature. However, just judging from actual "feel," I don't believe there were significant differences in ambient temperatures which would significantly affect the experiment's results. Just the same, I must say I wish ambient temperature was a controllable variable in this experiment.

Eight degrees is a bigger improvement over the four degree drop observed earlier. If before I said lapping was worthwhile, these results are even more convincing to me that lapping will result in significant drops in operating temperatures and improved performance of your cooling solution.

My promise of some photos ought to be fulfilled later, when I get home from work and I get a chance to upload some photographs.

Again, I hope this thread is helpful to some! Thanks for reading.
    
CPUMotherboardGraphicsRAM
Phenom II X3 720 BE Gigabyte GA-MA790FXT-UD5P Sapphire Radeon HD 3870 2 x 2GB Crucial Ballistix Tracers PC3 10600 
Hard DriveCoolingOSPower
Almost 8TB of storage AMD Stock Cooler Windows Home Server CoolerMaster SilentPro M 700W 
Case
Lian Li PC-V2000B (with thlnk3r mods) 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
Phenom II X3 720 BE Gigabyte GA-MA790FXT-UD5P Sapphire Radeon HD 3870 2 x 2GB Crucial Ballistix Tracers PC3 10600 
Hard DriveCoolingOSPower
Almost 8TB of storage AMD Stock Cooler Windows Home Server CoolerMaster SilentPro M 700W 
Case
Lian Li PC-V2000B (with thlnk3r mods) 
  hide details  
Reply
post #15 of 17
Thread Starter 
Here are some pictures.

I took a few more, but they're not very good. Too blurry.

As you can see, though, both the IHS and the heatsink's base look quite different compared to how they looked at stock.

More importantly, though, as they are today, they're also much more effective at heat dissipation.

Thank you once again for your time and attention.


    
CPUMotherboardGraphicsRAM
Phenom II X3 720 BE Gigabyte GA-MA790FXT-UD5P Sapphire Radeon HD 3870 2 x 2GB Crucial Ballistix Tracers PC3 10600 
Hard DriveCoolingOSPower
Almost 8TB of storage AMD Stock Cooler Windows Home Server CoolerMaster SilentPro M 700W 
Case
Lian Li PC-V2000B (with thlnk3r mods) 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
Phenom II X3 720 BE Gigabyte GA-MA790FXT-UD5P Sapphire Radeon HD 3870 2 x 2GB Crucial Ballistix Tracers PC3 10600 
Hard DriveCoolingOSPower
Almost 8TB of storage AMD Stock Cooler Windows Home Server CoolerMaster SilentPro M 700W 
Case
Lian Li PC-V2000B (with thlnk3r mods) 
  hide details  
Reply
post #16 of 17
Very nice and super informative writeup. Thanks for the coverage!

David
post #17 of 17
yes, very nice writeup, you have excellent prose!
cannabuild
(13 items)
 
  
CPUMotherboardGraphicsRAM
Core2Duo e6400 @ 3.2ghz Gigabyte GA-965P-DS3P (rev. 2.0) evga 8800gt 512mb @700/1050 4gb Patriot ddr2 800 4-4-4-12 1:1 
Hard DriveOptical DriveOSMonitor
WD 250gb Samsung Writemaster Windows XP Media Center HP w2207 22" 
PowerCaseMouse
OCZ Gamexstream 600 lian li pc-60B w/side window Logitech G5 
  hide details  
Reply
cannabuild
(13 items)
 
  
CPUMotherboardGraphicsRAM
Core2Duo e6400 @ 3.2ghz Gigabyte GA-965P-DS3P (rev. 2.0) evga 8800gt 512mb @700/1050 4gb Patriot ddr2 800 4-4-4-12 1:1 
Hard DriveOptical DriveOSMonitor
WD 250gb Samsung Writemaster Windows XP Media Center HP w2207 22" 
PowerCaseMouse
OCZ Gamexstream 600 lian li pc-60B w/side window Logitech G5 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Air Cooling
Overclock.net › Forums › Cooling › Air Cooling › Lapping a Thermalright Ultra 120 and an AMD Opteron 170’s IHS