Week 9 — Input devices

Group assignment for this week is on the Chaihuo Week 9 group assignment page. The rest of this page is my individual work.

Individual assignment

I read a rotary encoder on the milled board from Week 6 and Week 8. The input device is a KY-040 module, driven by pcb_testing on the W9 INPUT page.

Assembled circuit

The encoder connects to the XIAO carrier over jumper wires, with an AMS1117 branch for 3.3 V power after the on-board rail was damaged during soldering.

Powered bench stack with USB and module indicator LEDs

Wiring

Pin assignments come straight from pcb_testing.ino. The XIAO is powered over USB-C; encoder signals go to D1–D3. Because the on-board 3.3 V pad was damaged in Week 8, the encoder’s + pin is fed from a separate AMS1117 branch — not from a GPIO.

KY-040 → XIAO (signals)

KY-040 pin	XIAO pin	In code
SW	`D1`	`PIN_SW` — long press (2 s) switches mode
DT	`D2`	`PIN_DT`
CLK	`D3`	`PIN_CLK`
GND	GND	Common ground with XIAO and AMS1117

AMS1117 → KY-040 (power only)

The regulator exists because the PCB 3.3 V trace was lifted during soldering. Feeding the encoder from raw 5 V would also pull CLK/DT/SW highs to ~5 V and overstress the XIAO GPIOs.

AMS1117 pin	Connects to	Notes
VIN	5 V (USB / bench)	Bench 5 V rail
OUT	KY-040 `+`	Regulated 3.3 V — encoder outputs stay within MCU logic levels
GND	Common GND	Tied to XIAO GND and encoder GND

Learning the KY-040 rotary encoder

Before wiring I looked up how an incremental encoder behaves on GPIO. The module exposes quadrature signals CLK and DT plus a push switch SW. I used:

Wokwi KY-040 reference for CLK/DT timing, internal pull-ups, and quadrature direction.
The open-source KY040 Arduino library (module photos below) for pin naming and the three 10 kΩ pull-up resistors on the PCB.
The Seeed XIAO ESP32-S3 wiki to map Arduino pin names (D1–D3) to safe 3.3 V logic levels.

Reference module photos (front and back) from the codingABI/KY040 library — the rear side shows R1–R3, the 10 kΩ pull-ups that tie CLK, DT, and SW to VCC. That is why powering the module at 5 V can produce 5 V-high outputs even when the MCU is 3.3 V.

KY-040 module back reference showing pull-up resistors

Power workaround: AMS1117 3.3 V module

During assembly I accidentally damaged the on-board 3.3 V trace (Week 8). With that rail unusable I powered the bench from 5 V, but the KY-040 module is often sold as a 5 V–tolerant board whose output highs can reach ~5 V on CLK/DT. The XIAO ESP32-S3 GPIOs are 3.3 V only and are not 5 V tolerant, so feeding 5 V encoder outputs directly would risk damaging the MCU.

The fix was an external AMS1117 3.3 V step-down module: 5 V in, regulated 3.3 V out to the encoder’s + pin only. CLK, DT, and SW still connect to the XIAO inputs, but now swing within 3.3 V logic levels. Common ground ties the regulator, encoder, and board together. The IC itself is an AMS1117 fixed 3.3 V LDO (Advanced Monolithic Systems); the blue breakout board adds input/output capacitors and a power LED around that three-pin regulator.

How the firmware reads the encoder

Sketch pcb_testing.ino polls CLK and DT on every loop pass. When CLK changes state, the firmware compares it to DT to decide direction (+1 or −1) and updates an encoderSteps counter. On the W9 INPUT page the LCD prints the live CLK/DT/SW levels and that step count; serial logs mirror the same fields every 500 ms. The shaft switch uses a 35 ms debounce; a two-second hold cycles through sketch pages (W8 → W9 → W10 → W11).

W9 INPUT: live encoder readout

Hold the encoder switch for two seconds to step through sketch pages until the LCD shows W9 INPUT. The second line then tracks the KY-040 in real time:

C — CLK, quadrature clock line (0 or 1)
D — DT, quadrature data line (0 or 1)
S — SW, the shaft push switch (UP or DN)
d — cumulative step count from rotation; increases or decreases with direction and can go negative

Rotating the knob toggles C/D and steps d; pressing the shaft flips S. The clip below shows those fields updating on the LCD (audio removed).

Download Arduino sketch (pcb_testing.ino)

Reflection

Less about finding the perfect sensor than making a damaged production board still teach something. The AMS1117 workaround turned a soldering mistake into a clear lesson on level shifting; the encoder is simple on paper and harder on a hand-wired milled board.

The group probing session helped interpret what showed up on the LCD: quadrature as two time-offset square waves, and a logic analyzer catching missed edges that serial prints hide. I did not need an LA on my own bench to pass, but knowing what clean CLK/DT transitions look like helped when counts jumped on a loose jumper.