You should be sharing a ground between them - and you are the GND on the MOSFET shield attaches to the main ground on the Digispark - for good measure you should attach the PSU ground to the Digispark itself as well.
So you're saying I should run another line from the PSU ground directly to the Digispark ground, in addition to the one from the PSU ground to the MOSFET shield ground (which already ties to the digiground)? I think you lost me... is there a reason I'm missing? Or did I just completely misunderstand your suggestion?
Where the circuit goes wrong is that the Drain (OUT) of the MOSFET is also being attached to ground - effectively removing the MOSFET from the circuit and powering the LED strip permanently.
Ok, so at least I got that much right.
You can power the DIgispark from the PSU as you are now - but whether you do or not - you need to isolate the heatsink attached to the MOSFET - you can't attach it to a grounded case. It would be nice if the MOSFET was made so it could be, but they are sinking the heat out the drain to the case - so it can't.
Ugh. That's about what I was afraid of. That nixes piggybacking it onto any of the existing heatsinks in the PSU, too... and makes it tricky to mount a separate heatsink inside, unless I go the epoxy route. Need to figure out a new way to mount the USB connector, too...
Ah well... at least now I know I
have to redesign it, instead of constantly worrying that there was a simple & sensible option I was overlooking.
Always remember if you are powering something from a supply you must have a common ground with that supply - forgetting that can have some pretty crazy consequences
I think that may be part of what led to my oddly sensitive wire issue... It hadn't occurred to me when I was testing that, while the ground pin of the power cord went to the chassis, with the PCB removed the actual PSU ground was just floating. In retrospect, it's a little disconcerting.
- I recently forgot to connect my common ground on a circuit that was running from USB power and a 5v wall wart - the result was the destruction of everything connected (a Digispark and an Arduino and the wall wart) and the wall wart allowing 120v wall power into my circuit with a huge arc - glad I wasn't touching it!
This is why I'm a lot more comfortable trying this with the Digispark than the Arduino - much cheaper to keep a few spares kicking around. The fingers & hearts... those cost a little more to replace. (needless to say, the DigiX isn't going to be allowed to partake in such reckless pursuits)
I'd never start cutting ground on a PSU, they share a common ground across all voltages on the PSU.
Err, yeah. A 450W Bad Idea™.
Actually, one other (only vaguely related) question... When I was originally breadboarding this (using an Arduino, bench supply, and a circuit nearly identical to the MOSFET shield, but with a TIP120 transistor) I had no trouble measuring the frequency and duty cycle coming off the TIP120 with my DMM (an early model Fluke 87), but on the assembled version, it croaks (00.00Hz & OL, respectively) when trying to make the same measurement off the MOSFET. I'd actually been convinced that I I'd gotten something wrong, until I went ahead and tested it with the strip. Does this have something to do with the difference between the MOSFET and the transistor, or does the lack of a crystal throw off the timing enough to confuse the DMM? Obviously this isn't critical (or entirely relevant), but I've been curious about what causes it.
Thanks for all the help... even if it's not the nice easy answer I wanted to hear.
-Bats
