Friend at work today asked me when I was going to bring in my televisor for "show & tell". "Tomorrow!" I gamely say, even though I've not actually tested the new LED matrix in battle. So, first things first, I connect up the LED matrix and it works - sort of. I notice the picture is very glitchy and there are regularly spaced vertical stripes over the whole picture. I've seen this before, though - a clear symptom of not enough voltage to the LED array I thought. Not having a variable power supply at home, my choice was to go from 12V to 17V or to use a power brick which is pretty much untested in battle too. I decided to go with the 12.5VAC-->17.5VDC and keep a really close eye (well, finger) on the Arduino - as it quickly overheated last time I tried 17V. Good news is, this totally fixed the picture with regard to the LEDs - a lovely golden white, almost orange like the original pictures were meant to be. So that's good. I only ran it briefly, worried about the current to the LEDs. I calculated and wired in a limiting resistor, and this dulled the picture a lot.
One issue I was getting, though, is that if I had the speaker plugged in at the higher voltage, the system reset - every time. A big "squawk/pop" from the speaker, then a reset. So I did my further testing with the speaker not connected, and all working OK with it not there. No resets.
Next I thought I'd give the 15V power brick that I'd received a short while back. Here was my mistake. Not having the correct female power connector, I decided to wire up just a red/black wire to the main board, and manually hold the wires - one shoved in the center of the power connector, and the other held to the outside. Just for a test, mind you. So I was doing this, kind of dangling the power lead in the air and *POP* and magic smoke at the power inlet of the board. Looking closely, the wires were not clear and they had shorted So I carefully went back to the previous 12V and - no sign of life. I tried the Arduino by itself, connecting to the computer. Dead. Bummer, so that's board #2 dead. I so hate destroying things the day before a demo.
So, dig out board #3 and solder everything on. Everything, that is, except for the IRL540 and the TIP122. Where's my box of FET-things? Search... nothing. Practically tear the whole room to pieces, and remembered I took the box to work on the weekend to do some testing. I distinctly remember bringing it home though. So, search the car - aha! Back to the #3 board, solder in the two components and went for an all-up test. Everything plugged in (except the speaker, I'm now a bit wary of what's happening there). It works! Of course I still have those black stripes in the picture, but time to leave well-enough alone.
Tomorrow I will demo with a bench power supply and feed it 12V at first, and slightly up the voltage until the LED matrix is getting enough juice. Probably will be OK around 12.5V I would guess; it's pretty close now.
So, all's well that ends well - serves me right for being impatient and doing the dodgy "hold it together with your fingers while applying power" thing. Also, on the plus side, I now clearly have some investigation to do into what's happening with the sound at higher voltages. One thing I've been meaning to do is always leave the speaker PWM in a "quiet" state - Im guessing the click/pop I hear on startup/shutdown is where the speaker is just "snapping" to the "far away" value when there's a state change. I might have to put in some sort of smoothing filter for these situations.
Video shows the black banding on picture when LED matrix is under voltage.