Rohit (@.) on 12/15/11 wrote:
---------------------------
>Konrad Schwarz (no.spam@no.spam) on 12/15/11 wrote:
>---------------------------
>>You write:
>>
>>As the equation makes clear, subthreshold leakage increases exponentially with temperature.
>>
>>The exponent in the equation is -B * V / T
>>
>>If T tends to infinity, the quotient tends toward minus zero,
>>and the exponential tends toward one. Isn't the dominant factor in the equation then the T squared?
>>
>>This would mean that subthreshold leakage increases >>quadratically with temperature.
>
>That dependence is the asymptotic limit. Normally, you are in the regime where
>V is MUCH larger than the thermal voltage, which is exponential matters.

So ironically one of my friends who has a PhD in math made this same point. As Rohit mentioned, it's not the asymptotic behavior that is really a concern.

Tongue in cheek:

Asymptotically as T-->infinity, the leakage of the chip goes to 0 since it burns up : )

David