[time-nuts] Thunderbolt temperature spikes
holrum at hotmail.com
Mon Jul 14 18:34:19 EDT 2008
The nature of the spikes are that they show an instantaneous impulse 100mV rise in the temperature readings between two 1 second samples. The rise exists for one sample then decays over around 20 seconds. There is no way that any CPU (or bus) activity can generate a heat pulse that would propogate to the temperature sensor in such a manner... there is just too much thermal mass/spacing involved.
Also the spikes appear to occur randomly and are not correlated with any specific CPU activity. I have seen two spikes with five minutes. I have also seen 8 hours go by without any spikes. I think they are related to some glitch in the temperature sensor system... my bet is something in the firmware rather than the hardware. The way the pulse decays looks a whole lot like there is some software filtering of the readings going on.
>What do you mean by "not caused by real temperature changes"?
The i’m Talkaton. Can 30-days of conversation change the world?
More information about the time-nuts