Three wires, one channel

Three wires

• BCLK, the bit clock, ticks once for every bit. The ESP32-S3 makes it.
• DIN carries the bits, one per BCLK tick, most significant first. Each sample is a signed 16-bit number.
• LRCLK is LOW while a left sample goes by and HIGH while a right one does. It changes once per sample, so at 16 kHz it runs at 16 kHz.

The chip is always the listener: all three are inputs on the TK100, and all three clocks come from the ESP32-S3. A wrong pin does no harm; it just makes no sound.

Every frame has two samples

One frame is a left sample and a right sample: 32 bits at 16-bit stereo, so at 16 kHz the bit clock runs at 16000 × 32 = 512 kHz, worked out. The NS4168 takes any rate from 8 to 96 kHz and a bit clock of 16, 24, 32, 48 or 64 ticks per frame, so 16-bit stereo fits.

A mono chip picks one

The NS4168 has one output, so it plays one half of each frame and throws the other away. As shipped it plays the right half; bridging the L/R jumper makes it play the left.

That choice can silence a sketch that did everything else right. A program that puts sound in the left half and zeros in the right is silent on a TK100 as shipped. So the sketches in this book write each sample twice:

frame[0] = frame[1] = clip[i];      // same sample, both channels

Then it does not matter which half the chip takes.

Stereo from two boards

A stereo song on one TK100 plays one channel only, the right as shipped. Two TK100s on the same three wires make stereo: bridge the jumper on one so it plays the left, and give each its own speaker. Each still needs CTRL driven HIGH.