EYRotator

Read this first

1System overview

The ESP32 reads the rotator rheostat voltage on ADC1 GPIO34, converts it to antenna azimuth using stored calibration points, and drives separate CW and CCW relays.

The controller supports a 20×4 I²C LCD, a rotary encoder, physical direction and park buttons, a browser interface and USB serial protocols. Optional features include WSJT-X UDP grid-bearing integration, DS18B20 temperature monitoring and a 120-minute temperature graph.

Pin assignments

1AFirmware installation: full BIN

Before applying motor power, connect the ESP32 to a Windows PC using a USB data cable and identify the serial COM port in Device Manager. In the EY Rotator ESP32 Flash Utility, select the ESP32 chip, choose the matching v97.14 full firmware image and flash it at address 0x0000. Wait for completion, then restart the controller.

The merged full BIN contains the bootloader (0x1000), partition table (0x8000), boot_app0 (0xE000) and application (0x10000). Use the correct 4 MB flash configuration and matching partition scheme.

For a firmware upgrade, the retained NVS configuration normally preserves Wi-Fi, calibration and other saved settings. For true factory defaults, a full factory reset or erased configuration NVS is required. This also removes saved calibration and Wi-Fi credentials.

Example esptool command

Adjust the COM port, esptool location and file path. The filename is illustrative; use your actual full BIN.

esptool.exe --chip esp32 --port COM14 --baud 921600 --before default_reset --after hard_reset write_flash -z 0x0 "EYRotator_v97.14_full.bin"

2First power-up and Wi-Fi setup

1. Before applying motor power, verify relay polarity, correct GPIO wiring, ADC input voltage and common/isolated references according to the PCB design. The firmware assumes active-LOW relays on a fresh installation.
2. Power the ESP32. If valid Wi-Fi settings are absent or the connection fails, the controller starts its local configuration access point.
3. Connect a phone or computer to the open Wi-Fi network Rotator-Setup, then open http://192.168.4.1 in a browser.
4. Enter your local 2.4 GHz Wi-Fi SSID and password on the Wi-Fi configuration page. Saving credentials restarts the controller. If the details are wrong, the AP is available again after the connection attempt fails.
5. Once on the LAN, open the ESP32 IP shown on the LCD, or use the configured hostname or network discovery where available. Open Setup to inspect calibration and other settings.

3First-use calibration: two paths

A fresh installation or full factory reset requires valid CCW/CW endpoint calibration before ordinary automatic movement is enabled. Wi-Fi configuration is not a prerequisite for commissioning if the AP-mode LCD calibration screen is available.

3A. Standalone LCD auto-calibration (no Wi-Fi connection)

1. With the configuration AP active and calibration required, the LCD displays CALIBRATION REQUIRED, WiFi AP: ACTIVE, Press ENC to AutoCal and the AP IP.
2. Ensure the rotor can safely traverse to both mechanical endpoints. Press and release the encoder switch once to start.
3. The LCD shows AUTO CALIBRATION, the active CCW/CW leg and the live ADC reading. A deliberate encoder press while running aborts calibration.
4. The firmware drives to the CCW limit, observes stable feedback, then drives toward the CW limit. Allow the entire sequence to complete.
5. After success the LCD displays CALIBRATION COMPLETE, both captured ADC values and Press ENC to SAVE. Press and release the encoder switch again to save.
6. The endpoints are committed to NVS, the calibration interlock clears, and the rotator can be operated locally without a working Wi-Fi connection.

The firmware allows 120 seconds per automatic calibration leg. Endpoint detection uses a stable ADC window of approximately 4 seconds, plus a CW-leg travel/span plausibility check. A failed or aborted sequence does not authorise normal operation. The Save press is separate from the Start press.

3B. Web-based auto-calibration

1. Connect to the ESP32 web Setup page (LAN or configuration AP).
2. Select AUTO CALIBRATE and monitor CCW/CW progress and ADC status.
3. After success, the green SAVE CALIBRATION button appears beside ABORT CALIBRATION.
4. Press SAVE CALIBRATION to commit the endpoints to NVS. Successful detection alone does not save them.

3C. Manual three-point calibration

Use manual capture if the physical −180°, North 000° and +180° reference positions can be positioned accurately. Capture the CCW and CW endpoints from Setup while the rotator is stationary. Then position at true North and capture the optional North point. The North measurement corrects midpoint asymmetry.

The endpoint captures must be plausible. Changing either endpoint invalidates the previous measured North point.

If North has not been measured, the firmware uses (ADC_CCW + ADC_CW) / 2. The Setup button indicates whether North is calibrated and, when measured, shows the signed Delta ADC = ADC_North − calculated midpoint. The stored keys include adcccw, adccw, adcnorth and northvalid. The calibration-required interlock is tracked separately.

4Position measurement and ADC principles

The 500 Ω rotator rheostat produces a position-dependent voltage. The PCB analogue divider and filter scale it for the ESP32 ADC. The original hardware concept spans about −180° at the CCW endpoint, through North (000°), to +180° at the CW endpoint. Actual ADC counts depend on the divider ratio, reference voltage, wiring and ADC response. Never assume the example count of 2994 applies to every PCB.

The ESP32 ADC is 12-bit nominal. The firmware samples GPIO34 repeatedly, applies trimming and filtering, and can distribute samples across complete nominal mains cycles (50 Hz default) to reduce AC interference. It uses different filtering while stationary and while moving, balancing a stable display against timely motor stopping. Test changes to sampling, filtering and overshoot settings carefully.

4A. Two-segment calibration mapping

Let C be the measured CCW ADC, N the North ADC (measured or calculated) and W the CW ADC. For a reading A between C and N, azimuth is interpolated across −180° to 0°. For A between N and W, it is interpolated across 0° to +180°. Negative azimuths are wrapped for the 000–359° compass display. This allows different slopes on the two sides of North.

These explain the three-point method; the actual source also handles wrapping, filtering and limits.

Example: C = 0 and W = 2994 give a calculated N of 1497. If the measured N is 1542, Delta ADC = +45. The midpoint offset is a useful comparison between resistor-divider PCB variants, but it is not itself an angle error without conversion through the calibrated slopes.

4B. Turn and safety functions

• A web, USB or physical-button command requests an azimuth or direction. The LCD command-origin indicator distinguishes WEB, USB and BTN.
• Automatic target movement uses ADC position feedback, a configurable deadband, overshoot compensation and short-turn settings.
• CW and CCW relay commands are interlocked in software. Braking and relay-off delays are controlled separately.
• A 120-second default motion timeout and a feedback-stall safety check protect normal travel. The calibration sequence has its own 120-second-per-leg timeout.
• The controller rejects unreachable headings with TARGET OUT OF RANGE and blocks ordinary movement when calibration is required.
• During an automatic turn, the first physical direction-button press stops the turn. Release before starting manual movement.

4CFirst controlled motor turn

1. Confirm that CCW and CW calibration has completed and has been saved. Check that the displayed azimuth is stable and matches the physical direction.
2. Verify that the mechanical end stops and the motor isolation / STOP arrangements work. Keep the antenna and rotator clear of obstacles.
3. On the Mobile dashboard, enter a target only 5–10° from the present azimuth, well away from the CW and CCW limits.
4. Press TURN TO (target). Changing the target field alone does not command motion. Observe the actual direction and azimuth feedback.
5. Confirm the controller stops close to the target. Use STOP or isolate motor power immediately if movement is incorrect.
6. Repeat a small movement in the opposite direction. Only then test longer moves, PARK, presets and the 180° turn function.

5Web interface

The Mobile dashboard, annotated.

The rotational learning algorithm can be activated in Setup, allowing the controller to estimate turn overshoot more accurately.

6Operating controls

WSJT-X grid bearing integration

To use the WSJT-X interface, first activate the UDP listener in Setup (default is OFF) and enter your own grid location in 6-character format.

1. In the rotator Setup page, turn Grid Bearing / Distance and WSJT-X UDP Listener ON, enter your 6-character local grid locator and keep the UDP port (default 2237).
2. In WSJT-X, open Settings → Reporting and add the IP address of the router in the UDP text box. Save.
3. When you double-click a DX station, its grid locator is transferred automatically to the grid square box on the dashboard. Select Calculate to get the short-path and long-path directions.

7Temperature history and storage

The optional DS18B20 sensor on GPIO4 is read using non-blocking conversions. The Setup page graph covers 120 minutes, with a sample every 30 seconds. Current, minimum and maximum temperatures are shown, and unavailable readings appear as gaps.

The firmware uses a RAM circular buffer (241 positions) and LittleFS snapshots nominally every 5 minutes. It can initialise an unmountable filesystem by formatting the filesystem partition. This may erase other filesystem files but does not intentionally erase the separate NVS partition. History restoration depends on a synchronised clock. The Setup diagnostics report filesystem readiness, time synchronisation, saved sample count and save/restore state. SAVE HISTORY NOW allows an immediate test.

8Setup options and factory defaults

These are the values used when the corresponding NVS keys do not already exist. Existing stored settings override them. Some values are placeholders or dynamically managed, and are not proof of completed calibration.

9Troubleshooting and verification

10Commissioning acceptance checklist

• Relay outputs inactive at power-up
• 5 V / 3.3 V rails stable under relay load
• ADC feedback monotonic across the mechanical range
• CCW and CW calibration completed and saved
• North captured, or calculated midpoint acknowledged
• Physical STOP and direction buttons verified
• Web and USB source labels verified where applicable
• Travel limit, stall and timeout protections checked
• Temperature reading and history storage verified
• Configuration exported / backed up after commissioning

Advanced technical background

For builders who want the detail behind the ADC input circuit and mains-synchronous sampling. Figures are calculated from nominal component values; the bench checks below confirm them.

ADC input circuit: technical evaluation

The front end scales the 0 to 14.25 V position-sensor signal to 0.235 to 2.33 V (Option A) or 0 to 2.91 V (Option B) for ESP32 GPIO34 (ADC1, 11 dB attenuation). It low-pass filters the signal with a corner near 30 Hz and protects the pin with an SMAJ15A TVS and a BAT54S clamp.

Circuit description

1. Input protection. TVS1 (SMAJ15A) connects from the sensor line to ground at the cable entry, ahead of R1. It limits surges to about 24 V and does not conduct at the normal 14.25 V maximum.
2. Divider and offset (node A). R1 feeds node A in series from the sensor line. R2 connects node A to 3.3 V and R3 connects it to ground. Together they scale the input by 0.147 and add a 0.235 V offset, keeping the signal clear of the ESP32 ADC’s poor response near 0 V. C1 (100 nF) shunts noise to ground.
3. Filter (node B). Ferrite bead FB1 couples node A to node B. C2 (1 µF) and C3 (10 nF) connect node B to ground.
4. Pin clamp. D1, a BAT54S dual Schottky, sits at node B: one diode from ground to the node, one from the node to 3.3 V. GPIO34 connects to node B.

Resistor	Option A: offset divider	Option B: plain divider
R1 (series from the sensor)	33 kΩ	39 kΩ
R2 (to 3.3 V)	68 kΩ	not fitted
R3 (to ground)	6.2 kΩ	10 kΩ

DC transfer function (Option A)

V_pin = 0.1469 × V_in + 0.235 V

Condition	Input (V)	Pin (V)
Sensor at 0 V (CCW end)	0	0.235
Mid-scale	7.125	1.282
Full scale	14.25	2.328
End of linear ADC range at 11 dB (about)	15.1	2.45
ADC saturation (about)	19.5	3.1
Clamp starts to conduct (about)	22.9	3.6

Headroom above full scale before the ADC leaves its linear region is about 5.8%. At roughly 0.75 to 0.85 mV per count, the full span is about 2,500 to 2,800 counts, or about 7 counts per degree (0.14° per count) if 0 to 14.25 V covers 360°. The firmware’s endpoint calibration stores the actual counts, so the exact figures are not critical.

Key points

• Loading: Option A presents about 38.7 kΩ to the sensor line and draws at most 0.36 mA. With the 500 Ω rheostat wired as a potentiometer, the loading bow is at most about 0.19% of full scale (about 0.7° over 360°). Confirm the rheostat wiring from the rotator documentation.
• Filtering: a single-pole low-pass with a corner near 30 Hz gives only 6 to 7 dB of attenuation at 50 or 60 Hz mains, so hum rejection relies mainly on the firmware (oversampling, trimmed mean and smoothing). The output settles to one 12-bit step in about 45 ms.
• RF: the circuit does not stop common-mode RF current on the cable. Fit a snap-on ferrite clamp where the cable enters the enclosure. No rejection of 2.4 GHz Wi-Fi is claimed.
• Protection: the TVS limits sensor-line surges to about 24 V, and R1 with the BAT54S holds the pin between about −0.3 V and 3.6 V. For lightning-prone installations add a gas-discharge tube at the cable entry.
• Accuracy: endpoint calibration removes fixed gain and offset errors. The offset follows the 3.3 V rail (71 mV at the pin per volt of rail change), and the ESP32 ADC is non-linear near the ends of its range. Firmware v97.14 has no curve-compensation table (removed in v97.00), so endpoint (and optional North) calibration is essential. Re-run it after fitting these parts.
• Back-feed: if the sensor line can be energised while the controller is unpowered, D1 and R2 feed up to about 0.4 mA into the 3.3 V rail. Check that this does not partially power the ESP32 or disturb its start-up.

Choosing between Options A and B

Option A is the better electrical design; Option B is the pragmatic choice when you want the simplest circuit and the existing calibration.

Attribute	Option A	Option B	Advantage
Resistors	33k / 68k / 6.2k	39k / none / 10k	B (one fewer part)
Pin voltage range	0.235 to 2.33 V	0 to 2.91 V	none
ADC linearity	Whole span inside the linear window	First 5% and last 16% of travel outside it (about 76° of 360°)	A
Counts over the full span	About 2,700	About 3,500	B
Over-range before the ADC clips	About 19.5 V	About 15.2 V	A
3.3 V rail dependence	71 mV per V of rail	None	B
Mains attenuation (50 / 60 Hz)	−5.9 / −7.1 dB	−9.4 / −10.8 dB	B
Settling to one 12-bit step	45 ms	73 ms	A
Sensor-line loading	38.7 kΩ; 7 µA pushed back at 0 V	49 kΩ; no back-current	B
End-of-travel compression	Minimal	Present (~21% of travel); not software-corrected in v97.14	A

Choose Option A if accuracy near the end stops matters, since v97.14 has no curve table to compensate. Choose Option B if you want to keep the existing calibration and the simplest circuit and can accept reduced accuracy near the CCW and CW ends.

Switching options: change R1 and R3 and fit or remove R2, repeat the CCW and CW endpoint calibration from Setup, re-confirm endpoint (and optional North) calibration, then repeat the bench tests below.

Component specification

Ref	Option A	Option B	Specification
TVS1	SMAJ15A	same	400 W unidirectional TVS, 15 V standoff, SMA. From the sensor line to ground at the cable entry, ahead of R1
R1	33 kΩ	39 kΩ	1%, 0.1 W or more, metal film, 100 ppm/°C or better
R2	68 kΩ	not fitted	1%, 0.1 W. Offset from 3.3 V (Option A only)
R3	6.2 kΩ	10 kΩ	1%, 0.1 W. Divider to ground
C1	100 nF	same	X7R ceramic, 16 V or more, at node A
FB1	Ferrite bead	same	600 Ω or more at 100 MHz, DC resistance 1 Ω or less, 0603 to 1206. May be bridged with a 0 Ω link (less RF isolation)
C2	1 µF	same	Ceramic X7R, 10 V or more, at the ESP32 pin
C3	10 nF	same	Ceramic, 16 V or more, at the ESP32 pin. Keep even if FB1 is bridged
D1	BAT54S	same	Dual Schottky, series pair: pin 1 to GND, pin 2 to 3.3 V, pin 3 to node B. BAS40-04 has the same arrangement (check the datasheet pin-out)

Avoid power Schottkys such as the 1N5817 (high leakage) and 1N4148 diodes (about 0.65 V drop lets the pin reach about 4 V).

Layout and installation

• Place TVS1 where the sensor cable enters the enclosure, ahead of R1, with leads under about 10 mm. Lead inductance adds directly to the clamp voltage.
• Place R1, R2, R3 and C1 close to TVS1 with a short ground return. Place FB1, C2, C3 and D1 within about 10 mm of GPIO34.
• Decouple the 3.3 V end of R2 with 1 µF and 100 nF close to the divider, because rail noise reaches the pin at about 7% of its amplitude.
• Route the ADC trace away from the Wi-Fi antenna, relay-driver traces and the switching supply.
• Use twisted or shielded cable for the sensor line, with the shield connected to ground at the controller end, and fit a snap-on ferrite clamp where it enters the enclosure.

Bench verification

1. With the controller powered and the input at 0 V, measure GPIO34. Expect 0.235 V ± 0.01 V.
2. Apply 14.25 V (or the sensor’s full-scale voltage). Expect 2.33 V ± 0.04 V.
3. Apply 7.125 V. Expect 1.28 V ± 0.03 V.
4. Read the ADC counts on the serial monitor at each point. They should rise steadily at about 0.75 to 0.85 mV per count. Then run the CCW and CW endpoint calibration.
5. With the rotator idle and Wi-Fi active, watch the Min, Max and Spread values on the serial diagnostic line for a minute. A spread of more than about 10 counts after processing points to a noise problem.
6. Key the transmitter on each band you use, with the rotator idle. The ADC reading should change by no more than about 2 counts. If it changes more, add or enlarge the cable ferrite clamp, or restore FB1 if it was bridged.
7. Repeat steps 1 to 4 and re-run endpoint (and optional North) calibration after any change to the divider or filter values.

For Option B, steps 1 to 3 should read 0 V ± 0.01 V at 0 V in, 1.45 V ± 0.03 V at 7.125 V, and 2.91 V ± 0.05 V at 14.25 V.