|
Presentation of the "DoomsDay" fanless cooling heatsink[ Index ]
DATA :Dimension : L50 x W80 x H80 (mm) ; Where length is measured in parallel with the retainer clip. Weight : 227 grams (0.227 kg), 8 % lighter than the Q4 element. Material : Copper Made for socket 370, Intel Celeron / P3 FC-PGA Cooling capabilities (verified) :
Estimated time to create : 2 weeks. Measurement & Environment factors :All temperature data are normalized to an ambient room temperature of 20 C. Verified Test results and details :- Reality testing started on 23. February 2001. [ read ] - Test of the standard boxed element for Celeron 700 [ read ] [ Index ] Testing of the DD element :After completing the creation of the DD element, it was mounted on top of the CPU with some excitement. Performed different tests to find the characteristics of it. First some standard measurements to see if it holds under normal settings. Then the overclocking possibilities where explored. All testing done with cabinet cover on. Testing of standard settings :CPU clocked at 700 MHz @ 66 MHz FSB. Results for 1.35 Vcore, 3.4 Vio settings : ( Minimal stable setting )
Some performance reducing settings (chipset & cache latency) necessary where necessary for stability. Temperature data are normalized to an ambient room temperature of 20 C. Results for 1.4 Vcore, 3.5 Vio settings :
Full performance and hardware acceleration settings possible. Temperature data are normalized to an ambient room temperature of 20 C. Results for 1.7 Vcore, 3.5 Vio settings : ( Default standard settings )
Temperature data are normalized to an ambient room temperature of 20 C. Testing of overclocking possibilities :Exploring the possibilities for overclocking, showed that they where limited. There is nothing to balance but Vcore levels on the 75 MHz FSB setting. Used together with a large slow moving fan it might be a "killer" cooling element though. This is outside the scope of the project at the moment. Update sometime I need more speed. Vio setting for all measures below are standard (3.5 V), closed cabinet. Initial testing using a default Vcore setting of 1.7 volts. Started out freshly with a 100 MHz FSB setting. This totally failed. Tried with a 83 MHz FSB setting. This flopped as well. (look futher down for an update on this) Clutching at the last straw, different settings for a 75 MHz FSB setting where explored. Testing at 75 MHz FSB : Different Vcore settings1.7 Vcore : System boots, everything fine. High temperature, but usable.No data recorded. 1.6 Vcore : System boots. Optimized system performance settings are
fully possible.
1.55 Vcore : System boots, fairly stable, but system Chipset performance settings had to be tuned down to achieve stability. Works under normal circumstances, but system failed on CPU stab test (v.6.0) : Left unattended. Windows crashed. Temperature exceeded 70 C but below 75 C (auto-shutdown).
1.50 Vcore : Fails after a couple of re-boots. Not stable. Temperature within specs. Don't handle lowered Vcore voltage in this temperature range. 1.45 Vcore : Totally screws up at this setting. Even at most stable settings. Memory problems. etc.etc. Conclusion :Element handles the processor (C-700) at it's standard setting. A Lower Vcore voltage is preferred. An improvement in the processor-element fitting would probably improve the performance of the element. See what can be done. Adjustments :Surface treatment : The element looks pretty cool at the moment. Will a surface polish improve cooling capabilities ? CPU/Element junction : Flatten & polish contact area. More careful mounting of element (no twisting, just downward pressure). Checkout the silver based heat-transfer compound. Clip pressure : Increase fastening clip pressure. (Feather weight at the moment). Customize base spacer fitting. ( Dropped this, to much effort & not enough to gain ) #1 : Updated test results : ( after above adjustments )
Temperature data is not normalized, above. Normalizing temperature data to 20 C ambient gives :
> A noticeable drop in peak temperature. More so at higher temperatures. It is likely that Thermal Junction efficiency becomes more critical at higher temperatures. Results still not optimal, got into some trouble with the base spacer fitting. To thick. Mess. The thermal connection couldn't be made in one go. Wiggling. Not optimal, but better than before. Redoing measurement for the standard settings as well :
Normalizing temperature data to 20 C ambient gives :
#2 : Updated test results : ( whenever element fitting is done in one go ... )Remeasurment of the CPU extreme temperature : Vcore 1.70 FSB 66 MHz Celeron II 700
Testing at 83 MHz FSB : Different Vcore settingsI Later on discovered that by increasing the Vcore by (+0.15 V) to 1.85 volts, it was possible to boot into windows without problems. Any lower settings where unstable. The problem was of course instability at higer temperatures. Have yet not found the temperature limit at which the CPU becomes unstable, but there is a tool available that can stall the processor at this critical temperature limit (Hardware monitor). Will experiment with this later on. Ran at a speed of 875 MHz.
|
|