My opinion
Hi Speedo,
WARNING - THIS IS LONG WINDED
From the Intel link:-
"By using the motherboard's rigidity to its advantage, the thermal/mechanical solution is designed to provide structural robustness to ensure mechanical and electrical integrity of the desktop board during shock and vibration events. It does this by placing a compressive pre-load on the CPU package and socket. "
My opinion for what it's worth:- This is a great idea IF all MoBo's were created equal but surely they are NOT. That is to say some are more flexible than others.
If you take the Intel Heatsink/clamp and the ASUS plastic 'Heatsink Cradle' (which is screwed from the top side of the MoBo to a plastic 'stiffener' plate on the MoBo u'side) and assemble them outside the PC then observe the behavior of the 'Heatsink Cradle' mounting pads. Basically the 'Heatsink Cradle' flexes inwards against the aluminium heatsink and transfers this movement to the pads which in turn dishes the MoBo to give Intels 'designer' bend and subsequent "compressive pre-load"
My point...different MoBo's with variable rigidity=variable compressive pre-load. Some boards (like the ASUS P4C800) may be flexing too much and causing problems with CPU/ZIF socket electrical connections. Has anyone out there ever come across this problem with any other MoBo?
In the case of my PC which i've just built using a Chieftec Dragon case, ASUS P4C 800 MoBo and P4 2.8GHz/heatsink, when I switch on the only thing that happens is that the front panel LED's light and the fans come on...no POST, no display, no Hard disk activity and no beeps. UNTIL...I loosen the heatsink clamp arms to within about 5° of perpindicular to the MoBo and bingo things start happening straight away. I immediatly re-clamp (trying not to lose fingers in the fans) and continue as normal with CPU temps varying between 28° at idle to mid 50's for more intense 3d stuff. I can re-start ok but if I shut down I have to go thru the same ritual every time :-\ Funny thing is that the PC worked fine for the first 3-4 days before this
[email protected] started happening.
What next? Having read some of the threads here one guy successfully overcame the problem by filing plastic from the original clamp arms to reduce pre-load. The problem I have with this is that its difficult to predict/measure how much to remove plus you would need to get all four the exact same to ensure square clamping, plus you would probably invalidate your Intel warranty. Another approach would be to replace the u'side plastic stiffener with a metal one to prevent MoBo flex completely, however Intel specifically warn against this, warning that it would apply excessive clamping force to the CPU and subsequently damage it. If you knew the maximum allowable mechanical interference between heatsink and CPU I think you could build or modify a clamp to work with a rigid MoBo. My guess would be to set your clamp to give about 0.2-0.3mm CPU/Heatsink interference. Coolermaster have such a stiffener (available to system integrators) which works ONLY with their heatsinks (the cool looking Aero 4 for example). However this uses a four clip arrangment similar to the Intel idea with no obvious means of mechanically reducing the pre-load should this be an issue with the ASUS P4C800. What I have finally settled on is the Zalman CNPS7000-AlCu (
http://www.zalman.co.kr/english/product/cnps7000a-alcu.htm ), for two reasons 1. 'Gamedude' has tried it and reckons it solved the problem straight away 2. I love the clamp arm arrangment which uses 2 screws and looks as if it could be easily adjusted for pre-load with simple modifications (if nessecary). I've ordered this part and will let you know how it goes when I get it in the next few days.
Sorry for ranting on so long and if this upsets anyone from Intel or ASUS who may have taken the time to read this and take it seriously...ask yoursely this guys...how do you think I feel????
This is the last thing I expected to be trying to solve when I invested my hard earned cash in 'quality' components.
