Devlog 5

• fixed all teh issues the review brought up

Devlog 3 - Smart Bin

• got some feedback from the hardware reviewer
• said that my costs were too high and that I should switch to AliExpress and other sources for cheaper things
• looked into AliExpress and digikey for cheaper sources
• delivery fees are really high, so I swapped only the things that were free delievery
• cut down around 80 dollars from the BOM

Devlog 6

updated the case as per reviewer

btw the screws on the front link up to a bar on teh back to keep it even more secure (reviewer look at this)

Devlog 24: StudyBuddy

so a lot of the feedback said “cool but it needs ur gpu on”.

• that made it seem like a PoC, so now its running on the free tier of openrouter.

making the free tier work

• one request per question was never going to fit a daily cap. extraction now sends up to 10 pages / 25 questions per call, sorting and answer key go 10 questions at a time. ~4 requests per worksheet instead of ~42
• the homepage and upload page count the day’s calls. 80% says busy, 100% stops reading and points u at the samples.

samples

• upload page leads with the three samples as one-click cards; ur own pdf is the second option and says how much of the shared allowance is used.

removed

• phone photo uploads (highly innaccurate on the free tier of models i have)

Devlog 5 - NotGoingToBeLateToSchool

• fixing the repo for submission
• adding mounting holes for the display
• made the entire thing like 4mm larger to make the display fit better

Devlog 4: Adding all the reviewer feedback

went through every review from the last ship and fixed what was asked for.

• search: “LCD display” found nothing because the whole phrase had to match one field. now every word has to match somewhere.
• filters: tag, package and value dropdowns next to the search box, so you can cross-check values without typing.
• remove button: it sat outside the project table. it is inside the row now
• stale pages: you had to reload to see new shopping list items. that was the back button showing a cached page. fixed.
• shopping list export: download it as csv or plain text.
• front page: the root url used to bounce you to the login form. now it explains the idea and shows the bench.
• phones: below 600px the button rows stack full width, forms go one field per line, and wide tables scroll in their own box.
• roster: opt-in from settings. your bench goes on a public list with its part count, busiest tags and biggest stacks.

Devlog 4 - NotGoingToBeLateToSchool

finally finished CADing the board (idk if thats a word)

• i kinda had to redo the entire board but oh well yk

the final case

• its in two parts the top piece which is gonna be glued down and the body of the shape which has posts to hold the pcb
• moved the xiao and the display header to the bottom
• i also decided to use headers cuz this is my first time soldering a pcb so thats smth that could make this slightly easier

the problem (cuz obviusly it cant be a smooth process

• the switch is 15.6mm across the plate flange and a 1u dsa cap is 18.4mm
• my keys were on 15.09mm so thats too small
• so the switches overlapped and so did the switch caps and i didnt want to have to cad the caps too
• rerouted everything to make sure it all fits

the new pcb

• 461 tracks, 14 vias, was 284 (basically cut down hella on the overlapping things)
• i still have to make all the angles not right angles (i didnt know abt that until i say it on slack tdy when i was checking my pings (good to know ig))

things that i updated on the cad itself

• had to shift the usb cutout down a ton cuz i didnt account for the header lengths before

firmware

• finished it. 4 alarms with their own days, info pages for weather, hackatime, now playing and world clocks
• put the spinning globe from my hackpad oled on the clock face

next: get it approved so i can get the kit

Devlog 23 - StudyBuddy

• i just had some minutes left on this so im just making the devlog

• still waiting for my ship to be approved (been waiting a few days)

• no big changes at all (waiting for feedback)

Devlog 3 - NotGoingToBeLateToSchool

penciled the full case design on paper (with dimensions and math cuz its annoying to have to figure that out by myself)

• paper first then onshape cuz thats easier and i can worry about things like pcb fitting and stuff beforehand
• switched to onshape from fusion and its kinda better cuz “create part studio in context” lets me model the case straight off
  the imported pcb step instead of retyping dimensions
• ill still prolly use fusion a lot, but like im more biased to onshape (thanks blare!)

the shape

• its like a right trapeziod which i shelled
• the pcb is gonna sit on those posts you see and then imma screw it in
• the screen is gonna be in the front openign with teh jumpers and stuff
• thats abt it

next: finishing the CAD

Devlog 2 - MiniPad

• built the case on fusion, and guess what. i did lapse the first part of the case like v1, but forgot v2 cuz i left for a late night snakc and forgot to rerecord

• its all on onshape so you can see

the case

• 102.38 x 96.90mm, 16.6mm tall closed
• 1.5mm plate, 14mm switch cutouts on 19.05mm pitch
• pcb sits on four bosses 10mm up, leaving 7.5mm under the board for the reverse mounted leds and the xiao
• m3 heatsets drop into the bosses from the top
• usb-c slot in the side wall, the only overhang in the print
• fillets on the outside corners and chamfers on the top and bottom edges so it doesnt feel like a printed box

minipad (i gave it a name)

• renamed all the kicad files, gerbers, qmk firmware, and repo

restructured the repo to match the ship criteria. everything you need to build one is in production/ now: gerbers, both case steps, the prebuilt uf2. also made a single assembly step with the case and board in it

Devlog 2 - Smart Bin

ml pipeline

• 6 modules, tensorflow only, three dependencies. cut it from 629 lines to 308, and the old one trained in pytorch then converted to tflite for the pi, so it needed both frameworks
• the split is by session, not by frame. frames from one session are near-duplicates, so a random split puts near-copies of your training data in the test set and the accuracy number means nothing
• evaluate.py prints both numbers and the gap. the frame-split one is there as the number not to believe

firmware

• 11 files. trigger, weigh, light, capture, classify, decide, tilt, log, serve
• the safety rule is enforced twice: a confidence gate at runtime, and a config check that refuses to start if the rules file would ever send an unidentified item to recycling
• weight adjusts the result too. paper over 40 g is probably wet, so it goes to trash

bom

priced everything out, then checked the parts against my own schematic. found five problems:

• the mcp3208 had nothing to measure. its only connected input is the battery divider, and there’s no 12 V source in the build, so i dropped it
• the load cell “kit” ships an hx711 module. my hat already has an hx711 on it, and those modules run at 5 V where mine is deliberately at 3.3 V. bought the bare IC instead
• camera module 3 at $36 to module 2 at $9.49. autofocus is pointless when the camera and the pan never move relative to each other

before: $283.52
now: $195.19

Devlog 2 - NotGoingToBeLateToSchool

firmware is basically done

• wifi + ntp so it knows the time without an rtc
• clock face
• 3x3 matrix scan
• alarm with a buzzer that speeds up the longer you ignore it
• 3 snoozes max, then it makes you type a code on the keypad to kill it

the display was the annoying bit. 284x76 is not a size the st7789 library expects, so it needs init(76, 284) and setOffsets(82, 18) or you get the picture shoved into a corner. and Adafruit_ST7789 makes setOffsets protected, so you have to subclass it just to call it.

posting the cad in more detail next time, sorry (scared of losing my streak)

this is my planned cad design

devlog 1 - idk what im calling this

just finished the pcb for my hackpad. its a 3x3 macropad for shortcuts - app
launchers on the top row, screenshot/fullscreen/calculator in the middle,
copy/paste/undo on the bottom

specs

• 9 mx switches on a 3x3 diode matrix (row2col)
• 9 sk6812mini-e rgb leds, reverse mounted so they sit in slots in the board and
  shine up through the switches. per key lighting
• ec11 rotary encoder for volume
• 0.91” 128x32 oled
• seeed xiao rp2040
• 96.38 x 90.90mm, 2 layers, chamfered corners, 4x m3 mounting holes for a
  sandwich case

firmware

also got the qmk firmware compiling. the oled shows a clock and a spinning globe
made from real natural earth coastline data. 24 frames, projected the land
polygons onto a tilted sphere so those are actual continents at 32x32 pixels

next up

the case in fusion 360

Devlog 22 - StudyBuddy

• fixed more bugs and rewrote code by hand

Devlog 1 - Smart Bin

why: recycling bins get contaminated easily, and one bad load can get the whole bin rejected and landfilled. every recycling project i’ve seen is an app telling you which bin to use. i wanted the bin to just do it.

schematic

• 34 components, 26 nets: pi header, load cell (hx711), battery sense (mcp3208), servo rail, led driver, break-beam
• hx711 runs at 3.3 v so both signals are pi-safe, which deletes 4 level-shifting parts
• servo has its own ground meeting pi ground at one 0r link. a shared return browns out the pi and corrupts the sd card
• caught a real bug: led on gpio18 and servo on gpio12 are both pwm0

board

• 65 × 45 mm, 2 layers, pi zero hat footprint. fully routed, ground plane poured. gerbers exported.

cad

• 7 parts on a base plate: pan, hub, idler, servo mount, bearing block, pi mount, plate.

• it couldn’t move. pivot sat 16 mm above the plate, but the pan is 180 mm wide, so at 45° the edges swing 65 mm down and cut through it. raised the pivot to 70 mm

• also measured my actual mg996r instead of trusting the spec (it was actually 39.9 not 40.7 mm, shaft 10.95 off centre not 10.35)

FYI i accidently cadded on fusion and forgot to lapse (there goes like 4 hours of my life) and i imported to onshape so u can see it

Devlog 1 - NotGoingToBeLateToSchool

what it is: xiao esp32c3, 3x3 keypad on top of the case, 2.25in st7789
on the front on jumpers. ambient display by day, alarm by night.

features

• auto rotates through weather / hackatime hours / now playing / world clocks
• pulls real iphone notifications over ble with ancs
• at alarm time it pairs as a ble media remote and tells my phone to hit play
  on spotify, so i wake up to music not a piezo
• 3 snoozes max. on the 3rd you type a code on the keypad to kill it
• screen ramps brighter as the alarm approaches, fake sunrise

the board (schematic + pcb done and routed)

• 69.5 x 97mm, 2 layers, 25 footprints
• 284 tracks, 22 vias, about 2.1m of copper
• keys on the top side at 19.05 pitch, everything else underneath so the case
  only shows keycaps
• xiao at the back with usb-c out the rear, display header at the front edge
  so the jumper run is short
• diodes tucked into the gaps between key rows

stuff i got wrong

• thought i could make it a bluetooth speaker. c3 is ble only, no bluetooth
  classic, so a2dp is impossible. hence the ble remote workaround
• spent ages thinking the extra pins on top of the xiao were spare. they’re
  jtag pads for gpio i was already using
• so bl goes to gnd and rst to 3v3 like the docs say. no hardware brightness,
  sunrise ramp is a dimming palette instead
• wanted a rotary encoder, isn’t in the kit. went 3x3 matrix instead, 9 keys
  off 6 pins using diodes already in the box

next: case cad, then firmware

Devlog 21 - StudyBuddy

• patched even more bugs

• rewrote tons of my code (ai to human)

• cleaned up old files (a lot of old features were removed and reupdated)

• just more QoL fixes

Devlog 20 - StudyBuddy

• more bug fixes

• continous reships (they keep saying its too AI generated even after i rebuilt the UI)

• started working on the API key integration (like almost there rn)

• tons of dead code removal and code clean up

• added support for firefox too (now it works on edge, firefox, chrome, and prolly a ton others i just havent really tested it)