Devlog 4 Hydro Garden - Root Grate + BOM

what i did

built the root grate
its a 180mm disc, 4mm thick, 8mm holes on 3mm ribs so roots sit on it and water still gets thru
also chamfered 2mm off both bottom edges. since the cones are 45 and a chamfer is 45 they end up parallel so the grate beds flat on them instead of balancing on a corner
skipped finger holes cuz i can lowk grab it normally

looked into it properly and it doesnt actually work
fiberglass has no threads so theres nothing to tighten a nut against, and thats the whole point of the rods
BUT while checking i found the actual problem. m5 is metric and metric threaded rod is like 15 bucks a meter at home depot
switched to #10-24 which is the imperial one. its 2.98 for 3 feet. saves like 48 bucks and its actually SMALLER than m5 so it drops right into the holes i already made

a lot of them had the wrong description and stuff
the psu i had picked wasnt UL listed, swapped it
led channel was 39.99 and i found the same thing for 19.99
also found stuff i just straight up forgot to buy: tubing for the dosing pumps, wire, ph storage solution, and pvc cement
ended at 481 which is under the 600

4h 14m 34s logged

just built the actual grow module (the main part of the entire build)
it has a 45 degree floor to collect water
a gutter
a area for the pvc pipe to go thru
and 4 plant areas
had to a lot of making sure it fits inside a 256x256x256 square cuz thats the largest printer I have access too (it does now)
dont even need support anymore (might add them but i dont really know yet
made the drop holes slightly larger to make sure it can handle around 3.5 liters a min (pretty much the max my pump should be able to pump after the bypass valve)

i was talking to my dad and he said I could switch from threaded rods to the fiberglass thingys they use when it snows (its a lot cheaper and could save like 50 bucks
looking into ways to save some more money cuz almost 500 bucks (without the pcb included or filament cuz my friends wants me to pay for it)

5h 0m 25s logged

swapped the module joints from bolts and heat set inserts to four m5 tie rods running the full height of the tower. tightening the top nuts puts the whole stack in compression, which plastic handles far better than pull out, and it drops 32 pieces to 20 with no soldering iron needed
the rods pass through a ⌀16 boss at each end of every rib instead of a full height hole. a rod in tension cannot buckle so it needs no support in between, and 20mm of hole prints straight where 200mm would wander off axis
added the level sensor riser. the jsn-sr04t has a 200mm blind zone and a full tank only leaves 82mm of air above the water, so mounted flush it would read nothing exactly when the tank is full. it now sits on a 150mm post with a 250mm fill line instead of filling to the brim
finished the bom. every part has a real link, a real price, quantities matched to actual pack sizes, and a column saying what breaks without it. $496.40, about $100 under the tier cap

the first led strip i picked was rgb. four conductors needing three pwm channels, and my board has one channel for lights. the other three are committed to dosing pumps
the second one advertised 1800 leds and turned out to be 29 watts for the whole 5m reel. that’s 5.8 w/m, about 24 milliwatts per led. it would have put 19w across 16 plants and grown pale leggy lettuce. bought on watts per metre instead and landed at 17 w/m
home depot doesn’t sell heat set inserts at all, and their m4 bolts only come in 2 packs of plain steel that would rust in nutrient solution

4h 57m 25s logged

the project: a modular 3d-printed hydroponic tower where a single pump and a
single pipe water every level. no valves, no emitters. the shape of each part does
the irrigation.
what i did today: set up the onshape document with a variable studio driving
every dimension, then modelled the socket test plate, a coupon carrying one 45°
tapered socket that a net pot wedges into on its own cone.
what didn’t work: sketching directly on a 45° axis fought me for hours.
auto-constraints kept forcing lines horizontal, and dimensions kept measuring
vertically instead of along the axis.
what i changed: gave up on that approach and built the socket standing straight
up, then tipped it 45° with transform. worked first time.
what i fixed: exporting the step and reading the actual surface values caught
three problems the viewport was hiding. the 0.5 mm clearance step ran the wrong way,
so the pot would have jammed below the seat. the “straight” bore had come out slightly
conical. and the tube and bore were 0.55° out of coaxial.
where it ended up: all three fixed. the seat now measures ⌀50.6 tapering to
⌀43.9 at 12.59°, exactly what the net pot needs to sit on.