CHATGPT My new best friend!

[Tmaster]

Hello guys.
Years after MySensors release , and we still here talking here about, on on my electronic engineer opinion ,the best/usefull open source library ever!!! i say that because all my outdoor appliances, work with this stable library . Water heating solar panels, irrigation systems , outdoor light , main gate and basement gate can report status and command, water level from the irrigation system etc etc. Even for save water because i have timeouts and safe features for not water water or forget plants watering.... i only don't use the inside because my hardware don't fit inside concrete wall boxes....
Now since i had to debug one of my sensors and realize the power of chatgpt now i just got lazy and don't program my sensors myself. I just explain what i want and the AI do it.
have you tried it? Amazing .
This morning i did a water level sensor to replace my sonar one. Asked one mysensors code for 16 hall effect sensors connected to a mux cd74hc4067 and it suggest ; why no use interpolation,why not use sensor
hysteresis? hyster... what? i didn't its called like that. want cms or percentage....- DO it!! . 10 minutes later i had a hell of compacted code that do what i want WITHOUT BUGS and compile errors!!! yes,it's amazing because i'm not programmer and i can program c++ on arduino code but not much more than this....
So don't stop your creativity because you are not very good programmer and you don't even know how to start you project. mysensors library it's well known from AI . .thank you my sensors team

[fsgraz]

Fully agree with you about the library. The best one for sensors, amazingly stable and openly useable with lots of controllers.

And the combo ATmega328P + nRF24L01+ just great.

[OldSurferDude]

I agree with @fsgraz !

@Tmaster Here's the challenge with AI, it doesn't know what you want. Also, it only knows what has been done. That said, most of what we do has already been done and it can infer what you want.

Not knowing what the code does makes errant code difficult to debug. And debugging someone (something?) else's code is even more difficult. The upside of that is debugging errant code is that it is a great learning tool.

A forum like this is a great way to share knowledge and you are sharing your great success with AI. I'm impressed and I'm going to try it. Key to this sharing is having a good command of the language one uses to communicate. One could use AI to formulate a succinct description of the code, too. (Note that some languages express some ideas better than other one's native language, thus using other languages is also a good communication technique.)

Thanks for sharing and stimulation!

ps, can you share the code that the AI generated?

-OSD

[Tmaster]

It's not tested because i didn't finish the hardware yet ,but should be pretty close from what i want ..and clean..

#include <MySensors.h>

#define CHILD_ID_LEVEL 1
#define CHILD_ID_DIRECTION 2

#define TOTAL_SENSORS 16
#define TOTAL_HEIGHT_CM 150.0  // for internal reference only, we won't send cm

// Pins for the CD74HC4067 multiplexer
const int S0 = 4;
const int S1 = 5;
const int S2 = 6;
const int S3 = 7;
const int SIG_PIN = A0;

MyMessage msgLevel(CHILD_ID_LEVEL, V_LEVEL);
MyMessage msgDirection(CHILD_ID_DIRECTION, V_STATUS);

float lastReportedPosition = -1;
float previousReading = -1;
int confirmationCounter = 0;
const int CONFIRMATION_THRESHOLD = 3;

unsigned long lastReadTime = 0;
const unsigned long readInterval = 5000; // 5 seconds (adjust as needed)

void before() {
  pinMode(S0, OUTPUT);
  pinMode(S1, OUTPUT);
  pinMode(S2, OUTPUT);
  pinMode(S3, OUTPUT);
}

void presentation() {
  sendSketchInfo("Interpolated Water Level Sensor 49E", "1.1");
  present(CHILD_ID_LEVEL, S_LEVEL);
  present(CHILD_ID_DIRECTION, S_INFO);
}

void setup() {}

void loop() {
  unsigned long now = millis();
  if (now - lastReadTime < readInterval) return;
  lastReadTime = now;

  int readings[TOTAL_SENSORS];
  for (int i = 0; i < TOTAL_SENSORS; i++) {
    selectMuxChannel(i);
    delay(5);
    readings[i] = analogRead(SIG_PIN);
  }

  // Find the pair of consecutive sensors with the highest sum of readings
  int maxSensor = -1;
  int maxValue = -1;
  for (int i = 0; i < TOTAL_SENSORS - 1; i++) {
    int sum = readings[i] + readings[i + 1];
    if (sum > maxValue) {
      maxValue = sum;
      maxSensor = i;
    }
  }

  if (maxSensor < 0) return;

  float v1 = readings[maxSensor];
  float v2 = readings[maxSensor + 1];
  float frac = (v1 + v2 == 0) ? 0 : (float)v1 / (v1 + v2);

  // Relative position [0, TOTAL_SENSORS-1], inverted (top = 100%)
  float sensorPosition = maxSensor + (1.0 - frac);

  // Calculate level as percentage [0..100%]
  float levelPercent = (1.0 - sensorPosition / (TOTAL_SENSORS - 1)) * 100.0;

  // Hysteresis with 0.5% margin
  if (abs(levelPercent - previousReading) < 0.5) {
    confirmationCounter++;
  } else {
    confirmationCounter = 1;
  }
  previousReading = levelPercent;

  if (confirmationCounter >= CONFIRMATION_THRESHOLD && abs(levelPercent - lastReportedPosition) > 0.5) {
    send(msgLevel.set(levelPercent, 1)); // send percentage with 1 decimal place

    int direction = (levelPercent < lastReportedPosition) ? 1 : 0; // 1 = rising, 0 = falling
    send(msgDirection.set(direction));

    Serial.print("Level (%): ");
    Serial.print(levelPercent, 1);
    Serial.print(" - Direction: ");
    Serial.println(direction == 1 ? "RISING" : "FALLING");

    lastReportedPosition = levelPercent;
  }
}

void selectMuxChannel(int channel) {
  digitalWrite(S0, bitRead(channel, 0));
  digitalWrite(S1, bitRead(channel, 1));
  digitalWrite(S2, bitRead(channel, 2));
  digitalWrite(S3, bitRead(channel, 3));
}

And theres more.... i asked for he..she..it.. explain me the behavior of the code and this is what it explain:

🎯 Purpose of the Code
This Arduino sketch reads analog values from 16 linear Hall effect sensors (49E) connected via a CD74HC4067 16-channel analog multiplexer to measure the position of a floating magnet (e.g., on water). It calculates the water level as a percentage (0% = bottom, 100% = top) and detects the direction of change (rising or falling water level). The data is then sent via MySensors to a gateway/controller.

🔌 Hardware Involved
16 Hall Effect Sensors (49E) arranged vertically to detect a magnet's position.

CD74HC4067 multiplexer to connect all 16 sensors to a single analog pin (A0).

MySensors library for wireless communication (e.g., with an MQTT or home automation gateway).

⚙️ Main Code Behavior
1. Initialization
before(): Configures the multiplexer select lines S0 to S3 as outputs.

presentation(): Sends node and sensor info to the MySensors controller.

setup(): Empty in this version (can be used for future initialization).

2. Main Loop (loop())
Runs every 5 seconds (readInterval = 5000 ms):

a. Read Sensor Values
Iterates through all 16 channels of the multiplexer.

For each channel, selects the sensor via selectMuxChannel(i) and reads its analog value from pin A0.

Stores all values in the readings[] array.

b. Find Position of the Magnet
Finds the pair of consecutive sensors with the highest combined reading (indicating proximity of the magnet).

Uses those two sensors to calculate a more accurate interpolated position (sensorPosition), including fractional values.

c. Convert to Water Level Percentage
Translates the sensor position to a percentage level, with 100% at the top sensor and 0% at the bottom.

Uses:

cpp
Copiar
Editar
float levelPercent = (1.0 - sensorPosition / (TOTAL_SENSORS - 1)) * 100.0;
d. Hysteresis Filtering
Uses a confirmation counter to avoid sending data for small noise fluctuations.

Only reports new data if the level has changed significantly (by more than 0.5%) and remains stable for 3 readings.

e. Send Data (if changed significantly)
Sends the level as a float with 1 decimal point.

Sends the direction:

1 for rising water level

0 for falling water level

Prints this info to the serial monitor as debug output.

🧠 Smart Features in the Code
Interpolation: Instead of picking a single sensor, it uses interpolation between the two strongest sensors to increase resolution.

Noise Filtering: Uses hysteresis and confirmation logic to avoid spamming updates due to noise.

Direction Detection: Reports whether the water is rising or falling.

📤 Output (via MySensors & Serial)
V_LEVEL message: Water level percentage (e.g., 72.5%)

V_STATUS message: Direction (1 = up/rising, 0 = down/falling)

Serial example output:

mathematica
Copiar
Editar
Level (%): 72.5 - Direction: RISING
🧪 Use Cases
Smart water tank monitoring

Sump pit level monitoring

Reservoir or hydroponics water level sensing

Any linear position tracking using a magnet + Hall sensors

Let me know if you’d like a diagram or simulation example of how this works!

[OldSurferDude]

@Tmaster I think it writes better code than a lot of code I've seen, and the documentation is a lot better. The latter, of course, is because most coders don't document.

some key elements:

• good statement of work -- Purpose of the Code is key (did you write that? Good job!). This will guide the AI to write what you want.
• descriptive variables
• good documentation
• code is independent of reading sensors up-to-down/down-to-up

I spent a couple hours analyzing, researching and writing and re-writing this and all I can say is that the AI didn't catch is, as far as I can see, if your sensors are too far apart or your magnet is too weak, you could get false readings.

You, being the author of AI directive, are responsible for for the code. The AI is just a tool.

I started my coding with assembly language, though at that time we still had to enter the binary on some machines (set 16 switches, then press commit). ForTran and COBOL were the first real high level languages and subsequent languages, pascal, c, java, etc. were improvements. AI is a quantum step. It's still coding, but you have to learn how to talk to the AI to get what you want.

Good project! Let us know how it turns out and if you had to tweak the code.

-OSD