Santana schrijft dat een maffe chinees een stukje heeft geschreven hoe je ervoor kan zorgen dat je minder last krijgt van EMI (electro magnetical interference) bij je SDRAM, blijkbaar een probleem als je op fsb's hoger dan 150 gaat draaien .
Om de electro magnetische storingen tegen te gaan bouwt hij een koperen shield (wordt ook gebruikt bij RDRAM) om het reepje ram heen wat ervoor moet zorgen dat ie het chippie nog wat verder kan tweaken :
The PC133-SDRAM we use is working under 100-133MHz, and do not need shielding and thermal management to work properly. But, from a standard EMI point of view, any signal with a bandwidth over 20~30MHz will give EMI interferences, especially when the conducting wire has metal tips --- this will deteriorate the performance, not only giving out signal, but also receiving stronger EMI. The situation get worse with increasing frequency and will be unavoidable at 100MHz.
The EMI with mess up with the original signal pulses by mixing and modulation, and make new EMI. If EMI induced pulses are stronger than signal pulses, the signal will be suppressed -- one example is the poorly shielded computer will crash if you use a cell phone near by. When our OCers use FSB upto 150MHz FSB (the frequency range that call be called as OC), we are stepping into the "trouble zone" : 150MHz has huge amount of EMI, like from transmitters for pagers. Below is a picture of a pager working at 150.725MHz (the pager does not give EMI, it is the transmitter station).
[...] To add a shield on the RAM module is similar to the 80pins in the ATA66 cable --- to make the system more stable. As far as I know, Appollon (another OCer) in ShangHai is doing the same test and study. I want to remind you that this change will not modify the memory from ground up and can not even be called as reasonable design., so please think over before you do it. Especially, you run the risk of shorting your computer if you did not do a good job.
![]()
Dit is nou wat ze noemen het echte tweaken... toch?