On Saturday 07 January 2006 10:50, Tim wrote: > jludwig: > > It takes some time for the temperature sensors to respond and some time > > for the mass of the heat sink to warm. > > > > This being the case, from a "cold" (like say a half hour or more) start, > > one can get an idea on the accuracy of the temperature sensor(s). > > In general, heat sensors are going to be close to the hot device, and > measure the temperature at the point where heat will be a problem. > > Measuring the temperature of a heatsink, itself, is next to useless. > There's no point in having a sensor further along the metal. Quite > apart from the delays involved in heat transfer, there'd be heat loss, > too (you'd be reading colder temperatures than at the device that's > getting too hot for comfort). > > I'd expect sensor readings to react fairly quickly; mere moments, not > minutes. Of course, they could be too late to react to something > getting too hot to shut it down, that sort of thing is best handled in > another way. But you should have a useful gauge of whether the device > is cold, warm, hot, or overheating, on average. > > You can see the sort of times involved between sensors registering > changes and devices changing temperature if you stop the fan on > something. Depending on the device, the temperature climbs at a modest, > and predictable rate, with the sensors showing it accordingly. > > NB: Anybody thinking of testing that, do so with care. Be it on your > own head if you fry your CPU. I did my testing while everything was > still well under maximum operating temperatures. > > -- > Don't send private replies to my address, the mailbox is ignored. > I read messages from the public lists. Actually, (having been in the aerospace business thermal cycling satellites etc.) the thermal 'inertia' is much greater than you would expect.
