We decided on going with a PTC because it protects the entire circuit. So, even if by some assembly error the FET was pinned high, the PTC would not pass more current than the FET and inductor are rated for at continuous DC. And PTCs are pretty cheap nowadays, ~20 cents each. :)

ancient hacker, the chip has a 'always on WDT' and the only place the dog is kicked is a function where we also set the timer & port registers fresh. if a register freaks out, say the one that has the port direction or the PC, then either the dogkicker will be called (and reset the registers) every 2 seconds OR the chip is completely latched in which case the WDT will reset the chip. same with the BOD, which cuts in at 1.8V (the chip can run down to 1.8V) however, there is -always- a possibility for cosmic waves, so our design accomdates a pulldown to the fet gate and a 200mA 0.1 polyfuse after the DC jack. all shipping kits have this (we'll update the schem shortly). experimenting with wreaking havoc on the FET and causing a dead-short, the fuse kicks in and 'stabilizes' below the rated DC current for the inductor and FET. thanks for helping us think about all this again, if you're still reading this send us a note support@adafruit and we'll send you out a kit once we ship the first round of orders out.

yes its a cool trick! sadly those circuits work only with AC. if you put DC in you'd be disappointed with the output :) there are DC 'switch cap' converters, they seem to show up in RS232 converters a lot

it is firmware controlled but its set once, its not like an interrupt has to go off at the exact right time. AVRs are pretty good about not freaking out as long as you have BOD set and a WDT (which it does). also, to make the clock display brightness adjustable you cant just use an RC oscillator...how would you set the PWM duty cycle? could it *possibly* screw up? sure. just like logic chips sometimes die or glitch. but its not medical equipment, theres no interlocks, its a DIY desk clock kit in a plastic enclosure.

Ha! good luck adding a radio receiver to a clock with both a multiplexed display and 32khz boost converter. WWVB receivers wont work within 10 feet of this clock. (yes, we've tried it)

Yup! If you look at the schematics or design document you will see how this works - its described in detail there but basically: If power is unplugged, the microcontroller kills the display and runs at 50uA off of a coin cell for a couple of weeks until power is restored. the low power RTC is used when theres a blackout or the clock is moved, etc. Its a big pet peeve, all clocks should do this!

if you read the design document, schematic and source you'll see theres a lot more in there than just a counter. theres day calculator, date, alarm, low power mode, RTC, HV boost, menu system with configuration, etc etc. a $2.50 microprocessor isnt overkill when you consider its doing -everything- except drive the HV VFD.

if it was just a timer divider, it would be a rather boring clock! if you read the design document, schem or source code you'll see theres a lot more in there than just a counter :)

the PWM is hardware controlled and there's a WDT its as reliable as any 'off the shelf' chip...even 555's latch up if set up wrong ;)

Hi Anon. I designed the wave bubble project in mid 2004 as part of my research work. While the Bubl Space project is named somewhat similarly, the wave bubble is not based on it or stolen from it or derived from it. The BublSpace project is, in fact, conceptual and never functionally existed: http://www.textually.org/textually/archives/2004/08/005007.htm Whereas the wave bubble is a fully functional design, completely documented, and open source.