Wiznet makers

🔍 Original Architecture: Command-Driven Direct I2S Transmission

In the initial system, audio data sent from a PC is received by the W5300 Ethernet controller. The MCU then immediately forwards the data to the I2S peripheral via i2sWrite(). This approach tightly couples reception and playback.

✅ Pros:

Simple and easy to implement

Minimal latency in ideal conditions

❌ Cons:

Tightly coupled timing: audio data must be written exactly when I2S is ready

Risk of underrun or data loss, especially if transmission is delayed

Compressed audio codecs like OPUS or ADPCM introduce decoding latency → causes playback instability

🔄 Improved Architecture: Asynchronous Audio Queue Buffer

To solve timing issues, the system introduces an intermediate 8KB audio buffer, located inside the cmdAudio module. This buffer decouples network reception from I2S output.

Key Components:

Socket buffer (W5300): 8KB

Audio queue buffer (cmdAudio): 8KB

I2S ring buffer: 4KB

🧠 Why It Works

Removes the dependency between command timing and I2S availability

Supports buffered playback for compressed codecs

Allows for background processing, such as FFT or sample conversion

Enables flexible streaming with minimal jitter

🧩 Code Overview: Key Functions

1. cmdThreadUpdate() — Core processing loop

bool cmdThreadUpdate(void)
{
  for (int i = 0; i < CMD_DRIVER_MAX_CH; i++) {
    if (cmd[i].is_init == true) {
      if (cmdReceivePacket(&cmd[i]) == true) {
        ret |= cmdAudioProcess(&cmd[i]);
      }
      cmdAudioUpdate(&cmd[i]);  // Runs continuously in background
    }
  }
}

2. cmdAudioReady() — Returns available space in buffer

static void cmdAudioReady(cmd_t *p_cmd)
{
  data.u32D = (AUDIO_Q_BUF_LEN - qbufferAvailable(&audio_q)) - 1;
  data.u32D = data.u32D - (data.u32D % 2);
}

3. cmdAudioWrite() — Writes to audio buffer only

static void cmdAudioWrite(cmd_t *p_cmd, bool resp)
{
  memcpy(&file_index, &p_cmd->packet.data[0], 4);
  length = (p_cmd->packet.length - 4) / 2;
  qbufferWrite(&audio_q, (uint8_t *)p_buf, length);
}

4. cmdAudioUpdate() — Sends data to I2S when ready

if (state == STATE_BEGIN)
{
  q_Len = qbufferAvailable(&audio_q);
  length = cmin(q_Len, p_type->availableForWrite(p_type->ch));

  if (length > 0) {
    qbufferRead(&audio_q, (uint8_t *)p_buf, length);
    p_type->write(p_type->ch, (int16_t *)p_buf, length);
  }
}

5. cmdAudioInit() — Abstracting I2S/SAI using function pointers

audio_type[AUDIO_TYPE_I2S].write = i2sWrite;
audio_type[AUDIO_TYPE_SAI].write = saiWrite;

🎯 Where Can This Architecture Be Applied?

💡 Why This Matters

This buffered streaming structure enables:

✅ Codec-ready architecture (e.g., OPUS, ADPCM)

✅ Decoupled design for robust audio playback

✅ Multi-platform MCU support

✅ Real-time DSP/FFT possibilities in background

🚀 Even Better with W6300: A Modern, SPI-Based Alternative

While the W5300 delivers excellent performance, it relies on a parallel bus, which requires many MCU GPIOs and introduces routing complexity. An alternative is WIZnet’s W6300, which uses QSPI (Quad SPI) for high-speed Ethernet.

W6300 Advantages:

If you're redesigning your audio streaming system — especially on a compact or resource-constrained board — W6300 is a modern, viable replacement for W5300.