Sensors

[Charles Steinkuehler]

Brian Densmore wrote:
>> -----Original Message-----
>> From: Jonathan Hutchins

>> Come to think of it, I have CPU and MB sensors, but I don't 
>> think there is a 
>> third sensor.  Good call.  Where's the .4 come from though?

> That's a complicated question that would be hard to answer without
> a lecture in binary logic, digital electronics, and binary to decimal
> conversion. That said, I suspect the .4 is just a random fluctuation
> specific to your specific MB and could very well vary over time in your
> MB and between yours and other MBs of the same series. If I am correct
> and the circuit is converting temperature from an analog or decimal 
> representation and storing it as a binary. It is because not all decimals
> can be represented 'exactly' in binary logic. For example try to represent
> 1/37 using base 2. 

I suspect one (or both) of the following:

- There are actually more than 8 bits of resolution.  For instance, 9 
bits of all ones would potentially convert to 255.5 (511/2), pretty 
close to the 255.4 seen above.

- The temperature mapping is not linear, and the value reported is a 
result not only of the missing sensor, but of the math needed to turn 
the low-level sensor reading into an actual temperature.  This is highly 
likely, as the diodes typically used for temperature measurements have 
an exponential response to temperature, which is usually compensated for 
by software (where it's 'cheap') rather than hardware (which is expensive).

Of course, with the sensors code available, you could actually dig 
around and see what's going on, if you really wanted...

-- 
Charles Steinkuehler
charles at steinkuehler.net