Night Guard

Everything the board has read so far has been on or off. A button is pressed or it is not. There is no in between.

The world is not like that. Light is not on or off. It is dim, or bright, or somewhere between. Temperature is not hot or cold. To build anything that reacts to the real world, the board needs to read amounts, not just yes and no.

That is what this band gives you, along with the most important lesson in the whole course about working with real things:

That job has a name, calibration, and it is Part 8 of this band. Skip it and your night light will work perfectly on your desk and be useless in the room you built it for.

What you need

• Your board, USB cable and breadboard
• One potentiometer, the round knob with three legs
• One light sensor, a small round part with a wiggly line on top. It may be called an LDR or a photoresistor.
• One resistor of 10 000 ohms. The stripes read brown, black, orange.
• One LED and one 220 ohm resistor. A second LED, a second 220 ohm resistor, and a passive buzzer as well if you are doing the rain alarm or the room monitor Swap instead of the night guard: both of those use a green light and a red light, and the rain alarm beeps. The buzzer is not otherwise a Band 4 part; it arrives properly in Band 5.
• Jumper wires

By the end you will be able to

• Read a number from a sensor and say what range it can be in
• Explain how a knob turns a turning motion into a voltage
• Turn a number from one range into another with map
• Measure the same thing in three real places and write down what you found
• Say why 5 / 1023 gives you zero, and fix it
• Build something that decides for itself, using a limit that a person can set

Activity one: the rope

You need something long and bendy again: a shoelace, a phone charger cable, string, or the rope from Band 1 with its knot untied.

That is a potentiometer. The rope is a strip of material that resists electricity. The two ends are fixed. Your pinching hand is a contact that slides along it. Where your hand is decides how the total is split between the two sides.

Hold on to that. In Part 3 the two ends will be 5 volts and ground, and your hand will be the middle leg going to the board.

Activity two: the brightness walk

Take a pen and paper. Visit five different places, indoors and outdoors. For example: outside in the sun, outside in shade, by a window, in the middle of a room, and in a cupboard or under a cloth.

In each place, write down a number from 0 to 10 for how bright it feels, where:

• 0 means you cannot see your own hand in front of your face.
• 10 means it is so bright it hurts to look.

Write the two ends down at the top of your paper before you start, so you use the same scale in every place.

Keep this paper. In Part 8 you will do the exact same walk again, carrying the sensor, and compare your numbers with its numbers.

Question 3 has no right answer yet, and that is the point. Nobody can know the right switching number without going and measuring. Not you, not your facilitator, not the person who wrote the sensor.

There is a second row of pins on your board, along the bottom, marked A0 to A5. The A stands for analog, which just means "a smooth amount rather than on or off".

These pins do something the digital pins cannot. They measure the voltage on the pin and give you a number for it.

The instruction is analogRead, and it works like this:

int reading = analogRead(A0);

The number you get back is always between 0 and 1023.

• 0 volts on the pin gives you 0
• 5 volts on the pin gives you 1023
• Anything in between gives you a number in between, in proportion

So 2.5 volts, exactly half, gives about 512.

Do not confuse these two. analogWrite from Band 2 sends brightness out, and takes 0 to 255. analogRead here takes a measurement in, and gives 0 to 1023. Different jobs, different ranges, similar names. Write both in your glossary, side by side.

A potentiometer is the rope from Part 1, built as a component. Take yours out and look at it. Three legs, and a shaft you can turn.

Inside there is a curved strip of resistive material. The two outer legs are joined to the two ends of that strip. The middle leg is joined to a contact that slides along it as you turn.

So when you turn the knob fully one way, the contact is right next to the 5 volt end, and the middle leg reads close to 5 volts, which is close to 1023. Turn it fully the other way and it sits next to ground, so close to 0 volts, which is 0. Anywhere between and you get a proportion.

This arrangement, where a voltage is split between two parts and you tap the middle, has a name: a voltage divider. Remember the name, because the light sensor in Part 7 is the same idea with one part replaced.

int knobPin = A0;

void setup() {
  Serial.begin(9600);
}

void loop() {
  int reading = analogRead(knobPin);
  Serial.println(reading);
  delay(200);
}

Here is a problem you now have. Your knob gives you 0 to 1023. Brightness needs 0 to 255. They do not match.

You could do the arithmetic yourself. There is an instruction that does it for you:

int brightness = map(reading, 0, 1023, 0, 255);

Read it as a sentence: take reading, which lives on a scale from 0 to 1023, and give me the matching place on a scale from 0 to 255.

So 0 becomes 0, 1023 becomes 255, and 512 becomes about 127.

int brightness = map(reading, 0, 1023, 255, 0);

int brightness = constrain(map(reading, 0, 1023, 0, 255), 0, 255);

Take out the light sensor. It is a small round part, usually orange or brown on top, with a wiggly line across its face.

It is a resistor whose resistance changes with light:

• Bright light makes it resist less.
• Darkness makes it resist much more.

It has no direction, so it goes in either way round.

You now have a sensor giving numbers. Here is the question that decides whether your project works or not:

What number counts as "dark"?

Nobody can tell you. Not this page, not your facilitator, not the internet. It depends on your sensor, your resistor, the room, the time of day, and which way the sensor is pointing. Two people doing this band side by side will get different answers, and both will be right.

So go and find out.

Suppose you want to turn your reading into actual volts. The arithmetic is easy. 1023 stands for 5 volts, so:

volts = 5 / 1023 * reading

Upload b4_04_buggy_intdiv.ino, which contains exactly that line, and watch the serial monitor.

Nothing is broken. The arithmetic is correct. The problem is what kind of numbers the board thinks it is dealing with.

The fix

Write at least one of the numbers with a decimal point. That tells the board to work in decimals from the start:

float volts = 5.0 / 1023.0 * reading;

int millivolts = 5000 * (long)reading / 1023;
Serial.println(millivolts);

Everything is already wired. The knob is on A0, the light sensor is on A1, the LED is on pin 9. You only need the sketch.

Upload b4_05_night_guard_worked.ino and open the serial monitor.

Read the sketch

void loop() {
  int light = analogRead(lightPin);
  int threshold = analogRead(knobPin);

  if (light < threshold) {
    digitalWrite(lampPin, HIGH);
  } else {
    digitalWrite(lampPin, LOW);
  }
  ...
}

That is the whole idea in four lines. Two sensors, one comparison, one decision.

Notice what the knob is really doing. It is not controlling the light. It is setting the number that counts as dark. The person using the device gets to make that decision, in their room, at their time of day, without touching any code.

Now find the flaw

Turn the knob very slowly until the lamp just starts to twitch, right on the point of coming on. Then take your hand off the knob completely, because your fingers will keep nudging it, and watch for thirty seconds.

This is a real problem and every device that compares against a limit has it. Fixing it is the Gold task in Part 12, and the fix has a name you can look up: hysteresis. It means using two limits instead of one, with a gap between them.

Finding a flaw in something that works is a good day's work. Most people never look for one.

a. Work it out by hand

Fill in this table without running anything. Assume the knob and sensor are wired as in this band.

b. Break it on purpose

In b4_05, change if (light < threshold) to if (light > threshold). Predict what will happen before you upload.

c. Put the lines back in order

These lines have been shuffled:

analogWrite(lampPin, brightness);
int brightness = map(reading, 0, 1023, 0, 255);
int reading = analogRead(knobPin);

Put them in a working order and say why no other order works.

d. Explain one line

In b4_05, the same comparison light < threshold appears twice: once to decide the lamp, and once again lower down to decide what to print. Why not just print inside the first if?

Work on a copy of b4_05.

Bronze

Make the lamp fade as the light falls, instead of snapping on. The darker it gets, the brighter the lamp. You will need map and analogWrite, and you will need to think about which way round the map goes.

Silver

Print the reading in volts as well as as a number, using what you learned in Part 9. Then check it: at the halfway point on the knob, does it really say about 2.5?

Gold

Fix the flickering you found in Part 10. Use two limits instead of one: switch the lamp on when the light falls below the threshold, but do not switch it off again until the light rises above the threshold plus 50. Then find the smallest gap that stops the flicker, and write that number in your log.

The test that matters most in this band

Set it up carefully where you built it, and check it works. Then carry it to a different room and switch it on without changing anything.

You do not need a laptop to do this. A phone charger or a power bank plugged into the same USB socket runs the board perfectly well, and plugging it in is switching it on. The sketch is already inside the board.

Write down what happened. Did it still make sense? Did the person have to reach for the knob straight away? Was the knob even in a place they could reach?

Almost every sensor project in the world fails this test the first time. Yours failing is normal and expected. What you write down about why is the actual work.

Show your work

Post a video, your calibration table, and one paragraph on what happened when you moved it. Then look at one other person's calibration table and ask them one question about how they chose their number.

Code you have not seen. Answer from reading only.

int sensorPin = A2;
int lampPin = 6;

void setup() {
  pinMode(lampPin, OUTPUT);
  Serial.begin(9600);
}

void loop() {
  int reading = analogRead(sensorPin);
  int level = map(reading, 0, 1023, 0, 100);

  float share = level / 100;

  analogWrite(lampPin, level);
  Serial.println(share);
  delay(100);
}

1. What range of numbers can reading be in?
2. What range can level be in?
3. The lamp never gets close to full brightness. Why not?
4. What will share print, for every reading except the very highest? Explain.
5. Fix line by line: what is the smallest change that makes share work properly?
6. The sketch has no wire connected to pin A2 at all. What will the serial monitor show, and what is that called?

Your board can measure the world now, not just be poked at. And you know the thing that separates a project that works on a desk from one that works in a room: the number a sensor gives you is not a fact about the world, it is a reading that has to be interpreted, in the place where it will be used.

You also met the division trap, which will catch you again one day. When a calculation gives you a stubborn zero, look at the types first.

Before you move on, make sure you have

• Your working device, with a video
• Your calibration table with at least three real measurements
• Your paragraph on what happened when you moved it to another room
• The number you found for Gold, if you did it
• Both analogRead and analogWrite in your glossary, with their different ranges
• Your build log for this band
• Five or six correct answers in Part 14, including question 6

Next is Band 5: Simon Says. It is the hardest band so far and the most important one. You will write your own instructions for the first time, and you will find out why delay, which you have used in every single band, has been quietly holding you back the whole time.