Arduino, running the sketch you wrote
arduino-cli compiles it, an AVR8 core executes it instruction by instruction, and the LCD shows what the ATmega328P actually put on the bus. Uno, Nano, Mega and ATtiny85, no board on your desk.

Hardware is all about iteration. Velxio lets you wire a circuit, program the board and run it in seconds, real firmware on a real solver, right in the browser.
DFRobot · M5Stack links on this page are affiliate links: a purchase through them earns Velxio a small commission at no extra cost to you.
Nothing on this page is a scripted animation: the firmware is compiled and executed, and the circuit around it is solved.
arduino-cli compiles it, an AVR8 core executes it instruction by instruction, and the LCD shows what the ATmega328P actually put on the bus. Uno, Nano, Mega and ATtiny85, no board on your desk.

arduino-cli or ESP-IDF compiles the sketch, then AVR8, RP2040, RISC-V, Xtensa or ARM executes it instruction by instruction. What the OLED draws and what the serial monitor prints is what the chip actually did.

ngspice compiled to WebAssembly runs a full nodal analysis about 60 times a second. GPIO pins drive real nets, ADC inputs read solved node voltages, and probes report RMS, DC and current wherever you drop them.

Adders, comparators, decoders, flip-flops and a 1-bit ALU slice: no microcontroller anywhere, just gates, switches and LEDs solved as a real net. The classroom half of electronics, without the breadboard.

The M5Stack Cardputer with its keyboard, Seeed's round display keeping time, Pimoroni's Badger 2350 driving e-paper, and DFRobot's UNIHIKER M10 in the catalogue alongside them. Their makers sent the hardware so the emulation could be checked against the real thing.




35 boards across 6 CPU architectures — AVR8, ARM Cortex-M, ARM Cortex-A, RISC-V, Xtensa, and Linux. Including partner boards from Pimoroni, M5Stack, Seeed Studio and DFRobot. All running locally, no cloud needed.
One institution contract gives every student in the class a full Pro account. No per-seat checkout, no card details in a classroom.
Private projects, BOM and schematic exports and GitHub Sync, for everyone enrolled.
Billed once to your institution, per semester or per year. Volume pricing on request for large cohorts.
Arduino, ESP32, RP2040, STM32, ATtiny and Raspberry Pi, with SPICE analog solved alongside the firmware. Not a sandboxed mock.
The simulator stays free forever, WiFi with internet included. Paid plans unlock the Linux and STM32 boards, custom WiFi networks and the local network gateway. Maker and Pro add the AI agent; No AI leaves it out.
Discover the simulator + light AI help.
The full workbench, zero AI. You write every line.
For makers shipping real circuits with AI help.
Private projects, GitHub Sync and BOM exports.
Questions, bugs and ideas: the Discord and the issue tracker are open, and so is every line of the simulator.
一个基于ESP32的互动机器人,具有舵机耳朵、传感器和OLED显示屏。
An interactive ESP32 circuit simulation you can run free in your browser on Velxio, by zhouzhi0713.
// ---- In Velxio -------------------------------------------------------------------
// This is the firmware of a real desk robot, as its author wrote it:
// https://github.com/davidmonterocrespo24/robot_desktop (GPL-3.0)
// Five things differ from the repository:
// - the files sit side by side (a Velxio project has no folders);
// - Config.h says DHT22 and a light sensor whose output rises with light,
// which is what the parts on this canvas are;
// - loop() rests one millisecond between frames, so the emulated CPU idles;
// - the GIF clips are left out (ClipData.cpp), 776 KB of frames;
// - Buzzer.cpp asks for LEDC channel 6 on Arduino-ESP32 3.x as it does on
// 2.x: left to the core, the buzzer got channel 0, the left ear's.
// Open the 3D button: the model is the robot's own CAD, its screen is the
// OLED of this circuit, its ears are the two servos.
// Desktop robot: OLED face, two servo ears and a few senses.
//
// sensors -> Brain (mood, reactions) -> FaceEngine (emotion graph, animation)
// -> Ears / Buzzer
//
// The same sketch runs on Windows through emulator/ (see README.md).
#include <Wire.h>
#include <U8g2lib.h>
#include <ESP32Servo.h>
#include <DHT.h>
#include "Config.h"
#include "Canvas.h"
#include "FaceEngine.h"
#include "Ears.h"
#include "Buzzer.h"
#include "Sensors.h"
#include "Brain.h"
#include "InfoScreen.h"
U8G2_SSD1306_128X64_NONAME_F_HW_I2C u8g2(U8G2_R0, /* reset=*/ U8X8_PIN_NONE);
Canvas canvas;
FaceEngine face;
Ears ears;
Buzzer buzzer;
Sensors sensors;
Brain brain;
static uint32_t nextFrame = 0;
static uint8_t shown[CANVAS_BYTES]; // what the panel is showing (page layout)
static uint8_t pages[CANVAS_BYTES];
static bool firstFrame = true;
// Sends to the panel only the 8x8 tiles that changed. Moving two eyes touches
// a small part of the screen, so this is several times faster than a full
// refresh over I2C.
static void pushDisplay() {
canvas.toPages(pages);
int tx0 = 16, ty0 = 8, tx1 = -1, ty1 = -1;
for (int ty = 0; ty < 8; ty++) {
for (int tx = 0; tx < 16; tx++) {
if (firstFrame || memcmp(pages + ty * 128 + tx * 8, shown + ty * 128 + tx * 8, 8) != 0) {
if (tx < tx0) tx0 = tx;
if (tx > tx1) tx1 = tx;
if (ty < ty0) ty0 = ty;
if (ty > ty1) ty1 = ty;
}
}
}
if (tx1 < 0) return; // nothing changed
memcpy(u8g2.getBufferPtr(), pages, CANVAS_BYTES);
u8g2.updateDisplayArea(tx0, ty0, tx1 - tx0 + 1, ty1 - ty0 + 1);
memcpy(shown, pages, CANVAS_BYTES);
firstFrame = false;
}
// ---- serial console -----------------------------------------------------------------
// <expression> go to that face (try: happy, love, furious, asleep ...)
// act <name> play an act (try: pacman, pong, sneeze, fly, yes, no ...)
// say <text> speech balloon
// clip <name> play a GIF clip list names of faces, acts and clips
// auto on|off autonomous behaviour info