seen from Malaysia
seen from Canada
seen from Malaysia
seen from Malaysia
seen from Sweden
seen from China
seen from China
seen from China
seen from Canada
seen from China
seen from Panama
seen from Russia
seen from United States
seen from Panama
seen from Canada
seen from United States
seen from Lithuania
seen from China
seen from Italy
seen from Türkiye
And all 5 are programmed! Took a bit of fighting to get the fancy pogo pin programmer to play nice, (turned out I had a tiny solder bridge,) but once that was cleaned up, the pogo pin programmer made it a total breeze to get all 5 up with the new firmware (including the new thermal diode base seed generator.) Con is next week so we should be good to go!
What's that? A demonstration of several working devices in their final form? Yes... Yes I do think so!
On the Matter of Design Verification
So with one device physically made, replacement parts and tools for sidestepping my layout error in the mail, and my brain already churning on the next project it's time to reflect on what went well, what went poorly, and how to make the next project go better.
In general, this is probably one of my most successful projects. Every error in the system could be corrected without having to iterate the board, and as long as the battery covers come in within the next week, I will absolutely have at least one known-good device. As projects go this means I easily hit the 80% mark for design accuracy and completeness on the first iteration. For a one man team, working between calls, with less than a week to produce the schematics and layouts, this is an amazing feat despite the simplicity of the project.
But it was indeed only 80%. The programming port was laid out incorrectly, one of the parts failed to make it into the BOM and by extension wasn't ordered, and not all of the pieces of battery holder were ordered. The last error I blame on poor documentation from the manufacturer but the first two are all on me.
Since both of these errors are due to a lack of meticulousness on my part, I believe that while I'm waiting for parts to arrive I need to work on designing a design verification process. The first step is probably actually breaking out of the lone wolf operating style; catching your own mistakes is nearly impossible. That said, peers in my field willing to do this kind of work are few and far between, though this might not be a bad excuse to start trying to network with other hobbyists and EEs.
Demo video of the first assembled flash toy. Missing a few parts, and not everything went exactly to plan, but that's the reason people build prototypes and usually work in teams. Going forward I think I'm going to be working up a plan for verifying every aspect of a design before I go to fab.
That said, for hobby work this came out remarkably well. I'm quite pleased.
Putting it all together.
Boards are in! It's always nice to see concepts and ideas that exist only in software find physical manifestation. I'll be trying to build up at least one tonight, and hopefully I'll get more done so I can see how they play together.
Rigged up a prototype using a similar ATTiny24 microcontroller and worked out a few bugs in the code. At this point it's looking like we're all green on the board for a working project when the boards come in. They're in the mail as of today so I should have them before this time next week.
So with a bit of thought I think I've got an idea for a seeding method. The watchdog oscillator operates completely independently from the main CPU oscillator. As such, at boot up it would be easy enough to simply start both timers and then use the current value of one of the on-board timers as the initial seed for the linear shift register. I'll just have to do a quick check against 0 (since LFSRs don't like 0 as an initialization vector) and I'll be golden.
