NeuralCard

Brillar0101·NeuralCard·hardware/NeuralCard.kicad_pcb

NeuralCard PCB layout, top view — open source ESP32 board by Brillar0101
Simulate this board

About the NeuralCard PCB

NeuralCard is an open source ESP32 PCB design by Brillar0101, published on GitHub under the CERN-OHL-P-2.0 license. It is a 2-layer board measuring 85.6 × 54 mm, with 66 components from 26 distinct parts.

Its main chip is the ESP32-S3-WROOM-1-N8R2, from the ESP32 family. Other key parts include the LSM6DS3TR-C, ME6211C33M5G-N, ST25DV04KC-IE6S3, USBLC6-2SC6 and MCP73831T-2ACI/OT. By type, the board carries 25 diodes, 15 resistors, 14 capacitors, 6 ICs, 3 switches and 1 transistor.

It belongs with the sensors and wearables designs in this gallery.

From the project

A credit-card-sized PCB that runs a neural network: ESP32-S3, 6-axis IMU, and 24 LEDs wired as the network itself. Draw a digit in the air and watch the activations light up

A business card that runs a neural network.

It is a credit-card-sized PCB, 85.6 by 54 mm, carrying an ESP32-S3, a 6-axis IMU, and 24 LEDs laid out as the network it actually runs: 6 input neurons, 8 hidden, 10 output. You hold the card, draw a digit in the air, and the LEDs light with the real activations as inference runs. Brightest output neuron wins.

The front artwork is the network diagram. The synapse lines are drawn at three stroke weights, the way a trained model's weights differ. There is also an NFC tag with a coil antenna etched into the copper, so tapping a phone opens…

NeuralCard, from the project README
From the project README

Main components on the NeuralCard

NeuralCard bill of materials (BOM)

66 components, 26 distinct parts — part numbers from the project's BOM.

QtyPartRefs
1
ESP32-S3-WROOM-1-N8R2
WIRELM-SMD_ESP32-S3-WROOM-1
Espressif Systems
U1
1
LSM6DS3TR-C
LGA-14_L3.0-W2.5-P0.50-TL
STMicroelectronics
U2
1
ME6211C33M5G-N
Nanjing Micro One Elec
U3
1
ST25DV04KC-IE6S3
SO-8_L4.9-W3.9-P1.27-LS5.9-BL
STMicroelectronics
U4
1
USBLC6-2SC6
STMicroelectronics
U5
1
MCP73831T-2ACI/OT
Microchip Tech
U6
1
AO3401A
Alpha & Omega Semicon
Q1
1
TYPE-C-31-M-12
Korean Hroparts Elec
J2
1
3009
LIR2450
Keystone
BT1
2
TS-1187A-B-A-B
SW-SMD_4P-L5.1-W5.1-P3.70-LS6.5-TL_H1.5
XKB Connection
SW1, SW2
1
MSK12C02
SW_SPDT_Shouhan_MSK12C02
SHOU HAN
SW3
24
KT-0603R
LED-SMD_L1.6-W0.8-R-RD
Hubei KENTO Elec
D1, D2, D3, D4, D5, D6, D7, D8 +16
1
B5819W SL
B5819W
MDD
D25
6
CC0603KRX7R9BB104
C_0603_1608Metric
YAGEO
C1, C2, C3, C4, C5, C11
2
CL10A105KB8NNNC
1uF
Samsung Electro-Mechanics
C6, C7
1
CL21A106KAYNNNE
C_0805_2012Metric
Samsung Electro-Mechanics
C8
2
CL21A226MAQNNNE
C_0805_2012Metric
Samsung Electro-Mechanics
C9, C10
1
CC0603JRNPO9BN680
C_0603_1608Metric
YAGEO
C12
2
CL10A475KP8NNNC
4.7uF
Samsung Electro-Mechanics
C13, C14
6
0603WAF2200T5E
R_0603_1608Metric
UNI-ROYAL
R1, R2, R3, R4, R5, R6
Show 6 more
QtyPartRefs
2
0603WAF5101T5E
5.1k
UNI-ROYAL
R7, R8
2
0603WAF1002T5E
R_0603_1608Metric
UNI-ROYAL
R9, R10
2
0603WAF4701T5E
R_0603_1608Metric
UNI-ROYAL
R11, R12
1
0603WAF1003T5E
100k
UNI-ROYAL
R13
1
0603WAF1003T5E
R_0603_1608Metric
UNI-ROYAL
R14
1
0603WAF2002T5E
20k
UNI-ROYAL
R15

NeuralCard design files

The KiCad project lives in the Brillar0101/NeuralCard repository on GitHub; these links point at the commit this page was built from.

NeuralCard: common questions

What microcontroller does the NeuralCard use?

The NeuralCard is built around the ESP32-S3-WROOM-1-N8R2, from the ESP32 family.

How big is the NeuralCard PCB?

The NeuralCard measures 85.6 × 54 mm, has 2 copper layers and is 1.6 mm thick.

How many components are on the NeuralCard?

66 components, from 26 distinct parts, with part numbers taken from the project's own BOM. The full bill of materials is listed on this page.

Where can I download the NeuralCard design files?

From the Brillar0101/NeuralCard repository on GitHub, which has the KiCad layout, the schematic and the BOM; the links under Design files point to each one.

Can I use the NeuralCard design in my own project?

Yes, under the terms of its CERN-OHL-P-2.0 license, which Brillar0101 chose for the repository.

Can I test firmware for the NeuralCard without the hardware?

Yes. HardLabs builds a simulation of the board from its netlist and BOM, so you can run and debug ESP32 firmware against it before you order a PCB.

Similar boards