On 01/02/2017 22:11, rickman wrote:
> On 2/1/2017 3:36 PM, druck wrote:
>> On 31/01/2017 23:03, rickman wrote:
>>> There would appear to be something wrong with your numbers.  If you
>>> raise the ambient temperature 1 degree, the core of the CPU should rise
>>> 1 degree.  If you lower the ambient temperature 1 degree, the core of
>>> the CPU should lower 1 degree.
>>
>> Only if there is a perfect heat transfer between the chip's core and the
>> surrounding environment.
>
> I'm not sure what *you* mean by "perfect heat transfer", but I can
> assure you, all heat created in the CPU is transferred to the environment.

In perfect convective heat transfer, the heat transfer coefficient will
be constant and "Newton's law of cooling" applies, so then yes, with
changing ambient temperature a new equilibrium will be reached where the
temperature difference is the same as before; i.e. 1 degree hotter room
= 1 degree hotter cpu.

BUT... to quote Wikipedia: "The heat transfer coefficient is often
relatively independent of temperature in purely conduction-type cooling,
but becomes a function of the temperature in classical natural
convective heat transfer. In this case, Newton's law only approximates
the result when the temperature changes are relatively small."

In open cases, Newton's law will almost certainly be a good
approximation. Maybe the lack of circulation in druck's fridge explains
the slight divergence there, and I don't know WHAT happened in the car.

--- SoupGate-Win32 v1.05