Did you see the MotoGP race yesterday? I’m not sure how he managed it but Valentino Rossi’s final overtaking maneuver that secured the win also secured his hero status in my eyes. Awesome!
Despite the great race result yesterday was frustrating for me. I was attempting to write an assignment for my college course. Attempting because my laptop kept overheating and switching off – even with the cooling mat whatsit that I have already bought for it. The technique I discovered to ensure that the laptop remained on was to sit it (on cooling mat) on a flat surface (so not my lap then) and then prop up the front to increase the gap below the vents on the underside. This made typing more difficult but did mean it stayed on for longer.
As I was typing with my wrists at odd angles (carpel tunnel here I come) I got to thinking that my laptop cooling design must be very sensitive. Some days I can sit with it on my lap whilst lounging on the sofa without issue, and yet on other days like yesterday I wonder if there is any cooling going on at all.
That led me to thinking about who on earth designed my laptop and would I ever buy one from the same manufacturer again (I don’t know, looking more and more unlikely though). That in turn got me to thinking about how they had designed my laptop and that really they should have used the Response Surface Optimisation tool in FloTHERM.
The RSO allows a user to change variables within his (or her!) model and to see where the optimum lies. Of course you could do that with the plain ol’ sequential optimiser in FloTHERM too – so why is the RSO special?
Well for starters it allows you to see if your optimum solution is lying in a well.
A well is bad news as any deviation in the variable(s) will cause a rise in the cost function that you as a designer are trying so hard to minimise. An optimum solution that is in a well, may not be the best choice if you cannot guarantee the required degree of accuracy in manufacture for instance.
The RSO also allows you to see just how sensitive your design goal is to changes in the design itself. For instance how sensitive is your cooling solution to the ambient air temperature, how much the vents may be covered or indeed how sensitive your laptop may be to being placed on a lap!
A typical example is a heat sink, with the variable inputs as base thickness, internal fin height, and the thermal paste resistivity.
Here we can see that the cost function (which in this case was a component temperature target) is pretty sensitive to the internal fin height but not sensitive to the heat sink base thickness.
For those of us who enjoy our widgets the response surface viewer allows you to change the 3rd variable (ie the one not on an axis) and see how the response changes – nice
For those who also enjoy getting real results there are 2D graphs. Below are graphs for the same design with the same variables. The first show the response with an internal fin height at around 5mm – here you can see that you will need an expensive low resistance thermal paste to get your low cost function.
The second is with the fin height at around 15mm, here you can have a cheaper higher resistance thermal paste – and still get a lower cost function.
So my advice to my laptop manufacturer (no I’m not going to tell you who it is) is this – get FloTHERM and use the RSO your customers will appreciate it