“In a recent thread on Head-Fi, someone asked how well a paperclip would work for heat-sinking a TO-220 part. Much speculation ensued (much of it from your humble author), including opinions that a penny might work better, and then the argument moved on to exactly how to use the penny and so on. I eventually decided that experimentation was called for, which lead to this article.
I configured six LM317s as 0.125 A current sources propped up by silicon diodes. The diodes are remnants of a discarded idea, so their only purpose in this version of the test was as a place to clip the negative-side power lead. I then attached different heat sinks to each current source, put approximately 18 V across each one in turn, and recorded their temperature over time.”