Originally posted by: apocalypse
Originally posted by: Idontcare
Originally posted by: clarkey01
?
Intel allows Vcc of 1.5V on G0 Q6600. Intel wouldn't allow a Vcc that would violate the 10yr lifetime internal spec. (3yr public warranty)
So your voltage of less than 1.2V will probably give you a lifetime of 100-500yrs provided there isn't some other intrinsic flaw in your chip.
Decreasing vcore by 0.3 will give a 10 to 50 fold increase in chiplife time??
Originally posted by: LOUISSSSS
where can i get more concrete info on getting my Q6600 to last 500 years assuming all else is constant (working)
Electric-field induced breakdown failure mechanisms fall under a broad reliability category dubbed TDDB for Time Dependent Dielectric Breakdown.
TDDB covers both temperature (thermal) induced failure mechanisms as well as voltage (electric-field) induced mechanisms.
These classes of failure mechanisms involve the physical relocation of matter (diffusion of atoms) and as such are rate-limited by the activation energy required to do so.
[Activation energy]http://en.wikipedia.org/wiki/Activation_energy[/L] involves activation barriers which generally entail exponential functions relating to the probability of a random event involving reactants possessing the necessary energy to surmount the barrier to form the transition state as well as the geometry (angle of entry into the transition state).
Electric fields assist in polarizing the electron clouds in the atoms involved, slightly reducing the amount of additional energy needed to reach the transition state...in other words you can take the viewpoint that electric-fields are like a catalyst to lowering the activation barrier of thermally induced defect mechanism. (although it's not the true definition of a catalyst as catalysts must not be consumed on the process).
At any rate, what you get is failure mechanisms that vary as the exponential of the voltage (electric-field). Decrease the applied electric field by 10% and you can see the lifetime of the chip double.
This phenomenon is actually used in the opposite sense for determining chip lifetime, i.e. the field of accelerated aging is dependent on the fact a controllable elevation of the operating voltage by slight amounts results in the chip failing in weeks or months. This data allows the manufacturer to then know what max voltage to allow the chips in the field to operate at in order to give themselves some confidence that the return rate will be low. (i.e. how Intel came to know that a 1.5Vcc is the max allowable VID for a G0 Q6600 without waiting 10yrs to confirm 1.5V is ok...they test at elevated voltages (1.8, 2.2, 2.5, etc) and work out the weibull plot specifics for failure rates under accelerated aging conditions and then back-out what voltage is needed to afford a 10yr lifetime at standard operating conditions)
At any rate the 100-500yr lifetime is an order-of-magnitude reference as in O(3) lifetime vs. the O(1) that is targeted and is based on my experience in the field of accelerated lifetime modeling with SUN ultrasparcs I stand by it.
Is there a link on the web that proves this statement? Doubt it, you'd need a couple college courses plus some hands-on experience to gain a deeper understanding and appreciation for all the science topics being touched on here. Not trying to talk down to anyone here, just saying people get paid >130k a year for these job descriptions for a reason and you aren't going to find too many
correct laymen-type articles on the topic. Not unless CTho9305 has been busy making some more wiki pages
Facebook
Twitter