Hi, I made a further study on this problem.

Actually, the thermal diode is on the die.

As is pointed by [1], the temperature on different part of die is prone to 
be not equal. So, is it hard to get temperature of  `ALU' using a10's rtp 
diode. The prediction and calibration I made is only a close approximation.

As is pointed by [2], there exists a `Theoretical Power Dissipation Ratio'. 
I calculated TPDR of a10 using formula on the paper, only to find that the 
TPDR is not possible to be calculated. If I can figure out how to calculate 
TPDR of a10, then I can fix my model to be more accurate.

Further more, I think my setup looks more like this

<https://lh3.googleusercontent.com/-6-qdBBn92qw/U6XLGSSokII/AAAAAAAABy4/xOoBxFhoDVc/s1600/figure_measure.png>
The packages (red part) are tightly glued together using high thermal 
conductivity silicone. The So DS18B20 is actually measuring the BIG package 
temperature. If the thermal conductivity of the package is good enough the 
temperature on DS18B20 will be the very close to internal diode when system 
is balanced. And it will not lag too much behind internal diode(a shift on 
x axis).

<https://lh6.googleusercontent.com/-pkpTzrdSUvo/U6IcjqFJuSI/AAAAAAAAByY/dJiZi6K3kfQ/s1600/templog.png>
I don't know if I have said it correctly. Are there differences between 
surface and package temperature? Or this should be package temperature.
I do not major in physics, so I don't know too much of heat transfer. But 
intuition tells me that this can get right package/surface temperature.

[1] http://arxiv.org/pdf/0709.1861.pdf?origin=publication_detail
[2] 
http://www.arcticsilver.com/PDF/WhtPr/CalibratingPIII_InternalThermoDiode.pdf

Zhang Tong

On Friday, June 20, 2014 4:32:53 AM UTC+8, Stefan Monnier wrote:
>
> > what if i glued ds sensor to the surface? 
> > does the rule still apply? i.e. does this breaks the condition of the 
> rule? 
>
> It changes the proportionality factor, but not the rule, IIUC. 
> Think of it this way: 
>
> - there's a thermal resistance R1 between the core and the surface. 
> - there's a thermal resistance R2 between the surface and the ambient. 
>
> The heat (think "current") that flows from the core to the surface has 
> to be the same as the one from the surface to the ambient, so the 
> two temperature drops (think "voltage difference") should be 
> proportional to each other (with ratio R2/R1). 
>
>
>         Stefan 
>

-- 
You received this message because you are subscribed to the Google Groups 
"linux-sunxi" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to