Dont forget Planck's law
H
H
Skickat från min Samsung MobilColin Reese <colin.re...@gmail.com> skrev:I'm not
sure I caught all of that, but at steady-state:
Pdissipated = qconduction + qconvection
= k (Tdevice - Tamb)/ (thickness of boundary layer) +
hA(Tdevice - Tamb)
If you are dissipating some amount of power, the device will heat to a
temperature above ambient so that the equation above is valid. k, h, and
A are fixed, so Tdevice will increase until the equation is balanced,
i.e. temperature is transferred to the environment at the rate it is
produced. If the environment is 50C, the device will heat above 50C
until it transfers the balancing amount of energy via conduction and
convection.
C
On 10/5/2013 02:02, Colin Law wrote:
> On 5 October 2013 09:31, Colin Reese <colin.re...@gmail.com> wrote:
>> Without comment on the details of this situation, that is not how
>> convection and conduction work.
>>
>> Convection is typically described by Newton's law of cooling: q=hA(T-Tamb)
>> Conductions is q=k*dT/dx ~k (Tsurf-Tamb)/dx
>>
>> Thus the heat transfer rate depends explicitly on the difference between
>> temperature of the surface and ambient. Think about it - if the room is
>> 100C, the heat transfer direction is reversed, not to mention the magnitude.
>
> Still not getting it I am afraid. You are right that the heat transer
> rated depends on the difference between the surface and ambient, not
> the absulute temperature, so with your example of the room at 100C
> (and ignoring the fact that the chip will be fried) suppose the sensor
> was not sampled for some time, after a while the sensor will then also
> be at 100C. Now start sampling, some power is put into the sensor, so
> its temp will rise above 100C. Now imagine a room at 0C, again if
> left for some time not sampling the sensor will also be at 0C. Now
> start sampling, some power will be put in so the temp will rise. I
> believe by approx the same amount. the temperature difference between
> the it and the surroundings will be approx proportional to the amount
> of power, and not related (to the first order) to the absolute
> temperature.
>
> Colin
>
>>
>> C
>>
>> On 10/5/2013 01:16, Colin Law wrote:
>>> On 4 October 2013 20:28, Jan Kandziora <j...@gmx.de> wrote:
>>>> [snip]
>>>> In addition,
>>>> measuring low temperatures with parasite power is bad, as it heats the
>>>> sensor and thus, gives you slightly wrong temperature values.
>>>
>>> I don't understand why this is worse when measuring low temperatures
>>> than high. If measuring with parasitic power puts in a certain amount
>>> of power (say 10mW average for example) then I would have thought this
>>> would raise the temperature of the sensor by an amount depending on
>>> the thermal conductivity to the environment (say, 0.5 degree, for
>>> example). It should not depend on the absolute temperature. In other
>>> words it should raise the temperature by 0.5 degrees (or whatever the
>>> value is in any particular case) whether the ambient is -10C or +30C.
>>>
>>> Colin
>>>
>>> ------------------------------------------------------------------------------
>>> October Webinars: Code for Performance
>>> Free Intel webinars can help you accelerate application performance.
>>> Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most
>>> from
>>> the latest Intel processors and coprocessors. See abstracts and register >
>>> http://pubads.g.doubleclick.net/gampad/clk?id=60134791&iu=/4140/ostg.clktrk
>>> _______________________________________________
>>> Owfs-developers mailing list
>>> Owfs-developers@lists.sourceforge.net
>>> https://lists.sourceforge.net/lists/listinfo/owfs-developers
>>>
>>
>> ------------------------------------------------------------------------------
>> October Webinars: Code for Performance
>> Free Intel webinars can help you accelerate application performance.
>> Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from
>> the latest Intel processors and coprocessors. See abstracts and register >
>> http://pubads.g.doubleclick.net/gampad/clk?id=60134791&iu=/4140/ostg.clktrk
>> _______________________________________________
>> Owfs-developers mailing list
>> Owfs-developers@lists.sourceforge.net
>> https://lists.sourceforge.net/lists/listinfo/owfs-developers
>
> ------------------------------------------------------------------------------
> October Webinars: Code for Performance
> Free Intel webinars can help you accelerate application performance.
> Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from
> the latest Intel processors and coprocessors. See abstracts and register >
> http://pubads.g.doubleclick.net/gampad/clk?id=60134791&iu=/4140/ostg.clktrk
> _______________________________________________
> Owfs-developers mailing list
> Owfs-developers@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/owfs-developers
>
------------------------------------------------------------------------------
October Webinars: Code for Performance
Free Intel webinars can help you accelerate application performance.
Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from
the latest Intel processors and coprocessors. See abstracts and register >
http://pubads.g.doubleclick.net/gampad/clk?id=60134791&iu=/4140/ostg.clktrk
_______________________________________________
Owfs-developers mailing list
Owfs-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/owfs-developers
------------------------------------------------------------------------------
October Webinars: Code for Performance
Free Intel webinars can help you accelerate application performance.
Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from
the latest Intel processors and coprocessors. See abstracts and register >
http://pubads.g.doubleclick.net/gampad/clk?id=60134791&iu=/4140/ostg.clktrk
_______________________________________________
Owfs-developers mailing list
Owfs-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/owfs-developers
