Hey guys, completely vibe coded this,

Hi , taking a look

Hey that looks cool.

schematic is not readable , you can place components closer that would help.

i pushed an update for the schematic (vibe coded this too) is that better or is there something specific you want me to do ?

yea that's better now

and one important thing you should have a keepout under ESP antena

i would suggest to use ground pour and pack your board compact small.

Pushed another revision as per your feedback 😄

LGTM

wow thats amazing, this was completely vibe coded, with just an idea 😮

yea , this is something cool

lot of people here having working vibe coded pcbs from tscircuit

thats freakin awesome

if you found any issues while building board you are open to report.

can i send this straight to jlcpcb and have it manufactured ?

would be good if we get a another review from <@757706909351411845> , we do have a fabrication export .. you just drag drop all the necessary files and give it a goo.

<@174613218772254720> looks good but see these chatgpt comments AP2112K is only 600 mA. ESP32-S3 Wi-Fi current peaks combined with microSD activity could cause brownouts or resets. The microSD socket has no nearby dedicated decoupling capacitor. Card startup/current spikes may make it unreliable. Confirm the PCB contains actual copper routes. The green ratsnest lines could mean some connections exist only in the schematic. If those are genuinely unrouted, the fabricated board will not work regardless of the correct schematic. The generic pushbuttons have no assembly part numbers, so they may arrive unpopulated. My direct answer: yes, the circuit is logically capable of working if all traces are routed and the imported component pin mappings are correct, but power stability is marginal. For a prototype order, I would at minimum add 100 nF + 10 µF at the microSD supply and preferably replace the regulator with a 1 A part.

I gave your input to claude and it suggested me these Bottom line: their direct answer matches mine — logically sound circuit, real marginal-power risk. I'd act on it. Concretely, I'd: Add a 100nF + 10µF pair right at SD1's VDD/VSS. Add a bulk cap (10µF+) directly at U2's output, not just the 1µF that's there now. Swap the AP2112K for a 1A-class LDO — AMS1117-3.3 (SOT-223, huge JLCPCB stock, plenty of headroom for the 5V→3.3V drop) is the easy swap; it's a bigger package so it'll cost a little board space.

sg

just pushed another revision with fixes

im about to place an order on jlcpcb 🤞

Placed the order last night comes in 2 days let’s seee what happens

<@174613218772254720> i think we can use this C55061487

some other from my ai before re ordering

i found a youtube vid on kicad and the guy did gold fingers for the usb -a so i asked tsciruit to do the same by giving it a screenshot, pretty impressed, not sure what i need to tell jlcpcb to do to get those gold fingers

that looks good . maybe you can cancel the order from jlcpcb , and re order

they wont let me cancel i just went ahead with the component swap

oh

I might mill this pcb on my cnc, but i've never soldered in my life so this will be fun haha

looks awesome, yea would love to see a male usb with gold fingers

somebody has figured out how to do it with usb-c also haha

have not tried that myself yet

WHAAAT

I NEED !

yea it's kind of wild

fable couldnt do it !!!! it told me it was impossible

Short answer: USB-C with gold fingers only — no, that one's genuinely impossible, not just hard. But the USB-A style in your screenshot is very doable, and it's actually better than what your board does now. Let me explain both.

trying to find the guys on x

https://x.com/CapacitAnas/status/1982060711670067350

there is apparently an implementation here: https://github.com/vasya-zh/PCB-Edge-USB-connector-KiCad-library

haha this was claude's response - I owe you a correction: I was wrong. I told you USB-C edge fingers were "geometrically impossible" — this repo proves they're not. I over-reasoned the mechanics. Here's what they actually did, confirmed from the footprint file itself.

yea it's ok AI is often wrong in electronics right now

it'll get better

the fact that i even made a pcb with 0 electronics experience blows my mind haha

yea it's super cool

hey another question, if im not using a usb how does one flash firmware to a microcontroller, for my most recent pcb when it arrives ill plugin via usb and press the buttons to get the flashing going, but if we dont have a usb connector what is the way ?

there are tons of options

i would add a male or remale pinheader, then attach the SWD programming pins to a breadboard, then plug that breadboard into an RP2040 XIAO (any XIAO will work)

what we do internally at tscircuit for all our boards is use a JST connector to a XIAO

there are lots of ways but basically you need to be able to plug into the SWD pins, SWD = SoftWare Debug

even with USB, sometimes you can't program