How it works?

This section explains how the whole system works.

Block Diagram

1. The basic idea

At its core, a KVM device needs to do three things:

1. See what the target computer displays
   — by capturing HDMI video/audio.

2. Control the target computer
   — by emulating a USB keyboard, mouse, or storage device.

3. Send everything to you remotely
   — through Ethernet or Wi-Fi, using a browser-friendly Web UI.

LeafKVM integrates all of this into a compact, wireless, touchscreen-powered device.

Standalone

If the target machine is completely dead, frozen, or fails to boot, LeafKVM still works — because it does not depend on any software running on the target system.

---

2. Input interfaces (from the target machine)

HDMI → CSI bridge

The LTC6911 receives HDMI from the target device and:

• converts video into a CSI-2 stream that the SoC can ingest. Supports up to 4K@30 and 1080p@90.
• splits audio into an I²S stream. (connected to the SoC’s audio interface)

USB (to the target)

LeafKVM behaves like a physical USB device:

• USB HID → Keyboard + Mouse emulation
• USB Storage → Virtual Media / ISO mount

These are implemented through the SoC’s internal USB 2.0 PHY, and each endpoint is emulated by the backend.

---

3. Networking and connectivity

LeafKVM connects to your control device in two primary ways:

• Ethernet (RJ-45) — includes optional PoE for power
• Wi-Fi (802.11ac)

Additional auxiliary interfaces include:

• Local touchscreen — on-device configuration for network parameters
• Extension USB — auxiliary USB-A USB2.0

Note

The device does not require the target computer’s network.
You can connect directly, use LeafKVM as a Wi-Fi AP, or plug it into any LAN.

---

4. Internal architecture

Inside the SoC, several subsystems work together:

Web Backend

• Provides HTTP/WebSocket APIs for the browser UI
• Handles device settings, Wi-Fi setup, USB mode control
• Manages WebRTC sessions for remote video streaming

Video Encoder (VENC)

• Receives CSI frames from the HDMI bridge
• Encodes them into H.264/H.265
• Streams them to the browser with low latency

USB Endpoint Engine

• Implements software-emulated USB HID and USB Storage
• Allows injecting keystrokes, mouse movement, or mounting ISOs

Slint GUI Engine

• Renders the on-device 2.4” IPS touchscreen UI
• Communicates with backend through internal (Web) APIs

---

5. Output interfaces (to the user)

• Web UI
  Responsive interface for video preview, keyboard input, file transfer, and device control.

• Touchscreen UI
  Local configuration options, live HDMI preview, network setup, ISO mounting.

• Debug serial console
  Access through USB-C2 (internal USB-UART bridge, CH340).

---

6. Putting it all together

Here’s the full data flow:

1. Target outputs HDMI → LTC6911 → CSI → SoC video encoder
2. Encoded video → WebRTC → your browser
3. Your keyboard/mouse actions → Web UI → USB HID endpoints → Target
4. ISO images → Virtual USB Storage → Target
5. Local control → Slint GUI → Backend → Network/USB subsystems