Strikingly Bad Design

Ok, I know the assignment said to try to stay away from PC-related examples but this was such a good illustration of bad design that I simply couldn't pass it up. Pictured above is the underside of an Xbox 360 motherboard. We've probably all heard about the massive recalls of these systems due to "hardware failure". I think that this affected enough people worldwide to warrant a mention on the bad design blog.

This summer I tried a little experiment. I purchased a "broken" xbox suffering from the hardware failure that caused Microsoft to recall the system. A little research into what the actual problem was showed that the hardware failure was related to CPU and GPUs overheating on the motherboard. A little more investigating revealed that those shiny metal X shaped clamps were the culprit. Those clamps are attached to four custom bolts that poke through from the other side of the board. The bolts are fastened to heat sinks, and it is supposed to be the job of those X clamps to hold the heat sinks onto the CPU and GPU securely and with even pressure, such that the heat sinks can conduct the heat away from the vital chips and fans can cool the heat sinks; basically those clamps are an essential part of the Xbox's cooling system. Unfortunately, Microsoft appears to have tried to cut a few corners to save manufacturing costs, and the cheap metal they used for the X clamps actually warps under normal operating temperature. A warped clamp no longer holds the heat sink flush against the chip which leads to overheating.

Perhaps the worst part of this design flaw is how easy it was to fix. By simply removing the custom bolts and X clamps, re-applying some thermal-conducting compound, and bolting the heat sinks down using $.97 worth of hardware from Lowes, I was able to securely fasten the heatsinks so that they can do their job. Apparantly, the hardware failure is related to the heatsinks and not the chips themselves, because the "broken" system that I bought for $75 off of ebay now works perfectly, and the repair took about an hour. I understand that the design of this system was supposed to ensure a nice tight fit between the heat sink and the processor chips, but bad production led to failure of a critical element in the system. This is a classic example of a needlessly complicated design that offers no benefits over a simpler solution and is more prone to failure to boot.

I'm sure this one cost Microsoft a whole bunch of money too.

-Kevin