Air Gesture AC Dimmer/Switch

Non-mains-connected version for demonstration.

Built my CNC milling machine and milled out the switch's front face.

My electrode boards arrived.

Added some LED's and parameterized them.

A little more fine tuning and I'll be happy.

Been working on a LED display idea, the goal is to show the dimming level. I'm using white perspex, with 3mm diameter holes (not all the way through), and 3mm blue LEDs inserted into the holes. The LEDs are driven from a TLC5940. As the LEDs are really bright, I'm able to run them at ~1mA each with great results.

To give you an idea of what I'm hoping to achieve, here is the blank switch with paper in front of the LEDs. The images don't really do them justice. Looks really good in person. Also note that ambient light is shining both sides of the perspex. When the perspex is inside the switch and not being backlit, it looks even nicer.

Added fade-in and fade-out effect between on, off and dimming states.

Here is a quick demo with early firmware. Going to add fading between states, instead of switching instantly to a set dimmer level, or off to on and on to off, etc.

I have canceled my order for the Hillstar Development Kit, but it's not a big deal as I ordered samples of the MGC3130 and PIC18F14K50, built up my own dev board and downloaded firmware for the I2C Bridge from microchip :).

Ok, so the Hillstar Development Kit still isn't here (two and a half months later) and the company says they are waiting for new shipment from their supplier, and when they do get more stock, shipping takes 4-5 weeks, so god knows when this thing will arrive. Here is what I've been working on this week. Been reproducing the development kit myself basically, I'll be done before it gets here :(.

Just a little dimming test.

PWM now tested and trailing-edge looks good. Leading-edge is also achievable by inverting the PWM signal :). That premature trigger (glitch) you see is due to the zero-crossing pulse being centered (not starting) at the zero-cross, therefore the rising edge of the pulse happens just before the zero-cross. This can be corrected by adding a small delay after the rising edge of the pulse in software, to know real position.