Using a thermal cycler of his own design(circuit below) Brian Blais of Bryant College was able to achieve successful pcr. It’s important to note, because using a simple incandescent light bulb he got ramping of 1C/sec for heating and .25C/sec for cooling which is only about 3 times slower than lab quality PCR machines. Not to mention his cost $25 and most pcr machines are in the >$1k range.
If you are a diybiologist and are planning on adapting Brians lightbulb pcr for your home lab, I’d recommend you adapt the circuit to utilize an arduino (or any prototyping platform with an ADC), that way you wont need to dedicate a laptop to run your thermalcycler as Brian did. My other thought is that by adding a computer fan during the cooling cycle it might help to improve the cooling rate. Either way, if you try doing pcr using a lightbulb, let us know how it works and if you did anything to improve the technique.
Design and circuit via: http://web.bryant.edu/~bblais/projects/cycler/cycler-Feb02.pdf