I like the snorkels for better airflow... especially venting out the top like a landfill. Exhaust fans start ratcheting up the price tag and use additional power (yes it's there in the ATX but it means more wiring) There ought to be a way to configure it for the ATX fan to supply the movement. In combination with convection, placing the atx low in the box with the fan pointed up at the exaust pipe in the roof above should be enough to create the current and eliminate the extra fans and wiring. It makes the final box harder to store out of season though with the holes and exterior piping.
Has anyone done a temp analysis yet? Practically speaking, even with a mega tree of pixelnet nodes time on vs. time off while the sequences are running should average well under 50% of max power consumption. We're building for peak loads and finalies when you get a lot of blinky for a big finish, but really, the overall load should be fairly low... and then only at night, and not continuously. If anyone has the ability to run a test of say 8-10 100 node strings on 'white' for about 10 minutes in an enclosed, then passively vented (open bottome, or non powered vent) case that'll give a good idea of what kind of temp increases 95% of us are ever likely to see. We may all be overthinking this.
