SEN Blog
Development decisions and lessons grouped by project
40posts
ProjectsLoRaThirty sensors on one shared LoRaWAN gateway, and a node that would not join. With no CRC error to go on, the answer turned out to be the sync word set by a single toggle.
ProjectsLoRaThe radio worked, but the sensors were in the field. A PC program that answers like one turned up four contradictions in the protocol document and a timer bug that split good frames in half.
ProjectsLoRaCalling each device in turn fixed the collisions but cost 42 seconds a round, until we stopped waiting for replies. Plus the uplink monitor and the configuration tool built along the way.
ProjectsLoRaThree wrong theories built on partial logs, and then a line-by-line comparison of the full record explained all thirteen uploads at once.
ProjectsLoRaFrom cutting the enclosure to data that never reaches the server, five posts on a LoRaWAN sensor network - every one of them starting with a signal that said everything was fine.
ProjectsPCB AutomationMeasure the same board with two rulers and you get two answers, and the wrong one was always ours. Ten disagreements found by changing how we measured rather than by fixing the tool.
ProjectsLoRaAn automated clearance check that reads 0 mm3 is not the same as a case you can build. Five assembly failures, and the worst was a check that ran perfectly and reached the opposite conclusion.
ProjectsRP2350 3.5″ Touch LCDWhen the RP2350 Touch LCD 3.5 will not boot or the upload port cannot be found, unplug USB-C, hold the BOOT button, and reconnect the cable to enter USB download mode.
ProjectsRP2350 3.5″ Touch LCDAn RP2350 touch LCD timer that starts its MM:SS countdown only when the inverter is confirmed running - not when START is touched - and handles the STOP command and actual-stop check when time runs out.
ProjectsRP2350 3.5″ Touch LCDA horse-riding machine controller built with the 3.5-inch RP2350 touch LCD, with one manual mode, 11 program modes, 10-stage speed profiles, and RS485 Modbus RTU inverter control.
ProjectsRP2350 3.5″ Touch LCDRUN/STOP and speed control of an inverter over RS485 Modbus RTU, with the RP2350 touch LCD graphing only the speed the inverter actually reported - not the commanded value.
ProjectsRP2350 3.5″ Touch LCDCopying a TTF file into an Arduino project does not put Korean text on a TFT LCD. Here is how we converted only the glyphs and sizes the screen needs into bitmap data for the RP2350 Touch LCD 3.5.
ProjectsRP2350 3.5″ Touch LCDInstead of playing a video, we cycle three PNG frames and redraw only the region where the horse moves, cutting flicker on the ST7796 LCD.
ProjectsPCB AutomationAn unrouted net is not a conclusion but the starting point of the next investigation. Measuring in coordinates, suspecting the tool, unifying the pass criteria, and finally putting the placement choice itself inside the loop.
ProjectsPCB AutomationTen posts on PCB design automation, from generating the schematic to reaching DRC 0, collected on one page in the order the problems appeared.
ProjectsPCB AutomationEvery time we scaled an 8-channel board to 16 channels, we had to redo the schematic and artwork. To cut the repetitive work, we introduced AI automation and built a library and pipeline.
ProjectsPCB AutomationReading PADS ASCII coordinates as 1nm made the parts, the traces and the board all 1.5× too large together, so every internal consistency check stayed green while the object was wrong.
ProjectsPCB AutomationA backslash that invalidated an entire file, pad fields shifted by one slot, and a 3D model rotation we guessed wrong three times - what we hit while moving 1,214 footprints and 857 symbols into a KiCad library.
ProjectsPCB AutomationThe 45° rule, staircases and antennas - none of which DRC catches - each needed a checker of its own, and the per-layer routing directions had to be injected straight into the DSN file.
ProjectsPCB AutomationIf the measurement changes the subject, it is not data. How we caught a check command rewriting the board and a fixer script shaving trace widths to turn DRC green, then measured our own shipping boards for the real numbers.
ProjectsPCB Automation12.9 bends per net and a detour ratio up to 3.0× said what the gates would not: a 16-channel board was being crammed onto two layers. How we moved it to four and re-injected the per-layer directions.
ProjectsPCB AutomationThe longest stretch in automation is not the first 90 percent but the last single item. Ranking the metrics as shorts over unrouted over clearance, and solving a blocked spot with a transaction that moves the wall.
ProjectsPCB AutomationDRC 0, unrouted 0, shorts 0, measured five times with the same result five times. The closing round that took arbitrary angles to zero, plus the seven ways our own tools lied along the way.
ProjectsPCB AutomationThe metrics we measured passing is no evidence that the items we did not measure are correct. The schematic, label and angle gaps a review exposed after we called the board done, and the gates built to close them.
ProjectsBeer Game CabinetNo new parts, just the beer lever again. The design looked elegant on paper and played badly, and the work was translating four vague impressions into four exact numbers - tension, fish stamina, hooking, and haptics.
ProjectsBeer Game CabinetFrom one lever to a cabinet holding four games, seven posts collected on one page - in the order the parts arrived and the games followed them.
ProjectsBeer Game CabinetThe input device picked the game again: an analog joystick meant a vertical scrolling shooter. Climbing a night sky, shooting enemies, and getting stronger on pickups - plus a memory remap, a broken cause and effect, and an interpolation fix.
ProjectsBeer Game CabinetThe ERM vibration motor promised in Part 2 is installed and buzzing, but pulling the lever further changed nothing. After two wrong theories we stopped guessing, logged the actual values, and found the output window was the problem.
ProjectsBeer Game CabinetThe title screen became a game select menu and blackjack became game two. A house rule deals the dealer a card every time you take one, and inheriting the first game's leaderboard and comms layer cut the build time in half.
ProjectsBeer Game CabinetA full day spent making the HMI screen feel alive: five animations, a pub background drawn in code, and a bartender that took four failed shape-built bears before one AI handed the drawing to another.
ProjectsBeer Game CabinetPart 2 of the arcade cabinet. A ball lever modeled parametrically in Blender and printed on a 3D printer, and a Pro Controller haptics transplant that failed - one photo of the real part overturned the whole design.
ProjectsLyra Zero WThe Pi Zero 2 W is sold out and the Luckfox Lyra Zero W is the substitute - but is it fast enough? Same benchmark, same .NET version, both boards. On paper the Pi should have won, and the clock said otherwise.
ProjectsLyra Zero WThe Raspberry Pi Zero 2 W at the heart of ZPi Controller is sold out. We picked the RK3506B-based Luckfox Lyra Zero W as a replacement and actually moved onto it, from flashing the image to rebuilding the kernel and running real CAN control.
ProjectsBeer Game CabinetRoot Bear is a Playdate game about turning a crank to pour the perfect root beer. We moved it onto a physical potentiometer lever and a touch panel, and by the end of one day the inside of the cabinet was genuine plant equipment.
ProjectsArduino Web UploaderOne Raspberry Pi turned into a shared uploader, so a firmware change no longer means carrying a laptop and a USB cable over to the machine.
ProjectsAI Content ProductionThe design tools were never opened. The brochure was written in HTML, exported from Chrome, outlined in Ghostscript, and refined one instruction at a time.
ProjectsBlender MCPThree print-ready STLs and one design abandoned after seven rounds of corrections. The workflow, the parametric structure, and what went wrong.
ProjectsStage HMIPress the wrong switch on a stage and a set misbehaves. Building a panel where a mistake is obvious took four visualisation rewrites and five rounds of GPIO conflict.
ProjectsModbus HMI MigrationA teaching kit becomes an industrial line: no phone in hand, cumulative counts on screen, and a wired link whose loss is actually visible.
ProjectsGoing.IDEFrom an empty solution to v1.2.0.5 in seven days. The feature set, seven troubleshooting highlights, and the AI coding patterns that made the week possible.
