Quote:
|
Originally Posted by Les
Fascinating stuff as usual.
The overall values seem higher than those in preliminary investigation( ~0.04 for cheapo paste) :my memory or mounting pressure?...
Will re-tune brain and scrutinise.
|
Quote:
|
Originally Posted by bigben2K
Can you expand on that? Not quite sure what you mean there.
|
Les, the MG Chemicals 680-4g paste I used before came with a heatsink or something I bought long ago, mounting pressure is definitely different, I was fully compressed with different springs so I think lower than before but the preliminary number of 0.04 is still reasonably valid although the latest measurements I have were affected by a noisy power supply. I have not included them here. It has changed a small amount due to mounting pressure and the changes I have made to the way I calculate a few things.
Notably, my measurement of the fluxblock earlier was incorrect, I stated that it's dimensions were under 12x12mm. This was due to me measuring with digital calipers outside in my workshop where the temperature was around 5C. I have since discovered that these calipers change value with temperature due I guess to expansion of the metal. A bit annoying but handleable. The true value is something like 12.02x12.03x12.89 (from memory) and I have also lapped the ends so the overall length is marginally less - that has no impact on the TIM value but is relevant for...
...Ben, the fluxblock and heat die have 1.5mm holes drilled into them from the side to accomodate the thermistors. This means the cross sectional area of the block around the thermistors is lower than in the rest of the flux path. This impedes the overall heat transfer and causes the temperature measured to be higher than what it should be in a solid block.
I overcome this by calculating the effect on the C/W of the changing CSA and generating an "effective length" for the thermistor position. The reduction in surface area creates an increase in effective length and these extended lengths are used in my Power calculation (Q=kAdT/L') and interface temperature extrapolation.
I have attached my correction sheet if you are interested. Note that the values are not the current ones but it shows what I am doing. The method is a bit primitive but works. It is also probably not strictly "heat shadowing" and the thermal conductivity of the thermistors, compound in the hole and the epoxy are unknown but I think it nudges me in the right direction.