Using the 18 bits MCP3421 I2C ADC with Arduino.

A while ago, I published a post about expanding the ESP8266 ADC capacities with the PCF8591 8 bits I2C ADC multiplexer. 8 Bits is really enough for most work I would guess, but if you need more accuracy, say 18 bits for a  high precision Digital Voltmeter, there is the (relatively expensive) MCP3421. I say ‘relatively coz it is still cheaper (and easier to use)  than the LTC2400 (controlled by SPI). This is only a one channel chip, so if you want more channels, you would think you have to get more (which is OK as generally they are sold in webstores in a lot of say 5 or 10), but the chip has no address selection. It has a fixed address of 0x68.
If you do not like soldering though, there is also an ‘evaluation module’ available.

The MCP3421 has an in-program sample rate selection from 12-18 bits and in inbuild Programmable Gain Amplifier that can also be controlled from  your program. There is a library available, several in fact which makes using the chip very easy. If one channel is not enough, try the MCP3424 18-bit, 4-channel, multi-address ADC


Arduino Watchdog timer

I will not go too much into the benefits of a Watchdog timer as I presume those are known, but in short a Watchdog timer resets a processor  when it hasn’t seen any activity of that processor for a while. This is  specifically handy for processors  that are in a less accessible space, where otherwise you’d have to go to, get access and  press reset.

Basically there are 2 types of  Watchdog timers: Hardware and Software. About Software WDT’s I can be brief, they work, but  as the very goal of a WDT is to reset in case of software failure, using a software WDT seems a bit counterintuitive. If however you opt for a software timer, you will find a good article here.

Many of the Atmel chips, such as theAtmega328 in the Arduino UNO have an internal hardware Watchdog timer. In reality that watchdog timer is just a simple counter that gives a pulse when it counts up. This pulse can be used either to generate interrupt or simply reset MCU (or both). The watchdog timer can be reset to zero at any time with the WDR command.

The Internal Watchdog timer of the Atmeag328 is set by by programming WDTON fuse when flashing the MCU. You will find more information here.

Other than having to set fuses when flashing your chip, this internal hardware timer has a drawback similar to the software WDT: If you make a mistake, your chip could get into near continuous reset which makes it hard to re-use it again.

From that point of view an External Hardware WDT might be a better option. If you dont need it anymore, or made a mistake, you simply disconnect it.
There are specific  WDT chips available, but the good old NE555 can be used as well.

Watchdog timer
Watchdog timer

The workings are as follows:
In this circuit the 555 is configured as an astable oscillator. When free running it will charge C2 till about 2/3rds of Vcc (so about 3.33V) and then generate a negative pulse on pin 3 that is connected to the RST of the Arduino (or other chip). Diode D1 is not essential, but it does protect the Arduino.
The frequency, and thus period of the oscillator is determined by R2, R3 and C2. In its current config the time period til reset is about 69 seconds. At the end of that period the  circuit will generate a LOW pulse that  will reset the attached microprocessor. As the dutycycle of the circuit is far from symmetrical (fortunately) the LOW pulse will take some 15 mSec which is a useful time. Not too short, not too long.

However, we do not want the processor to restart every 69 seconds, we just want it ro  do its thing unless it hangs and therefore we let the processor restart the  555 oscillator continuously to signal that it is still alive and kicking. We do that by discharging C2 via R1 and we do that simply by using an output pin that we make LOW. So if during each loop of the firmware in the processor we pull that Output pin LOW, the Watchdog knows the processor is still doing its work and only when the program hangs, the watchdog will do a reset after 69 seconds.

It is not sufficient to just make the Output pin LOW to restart the timer, but afterwards it needs to be set HIGH again. However, even if an output pin is HIGH, there still is an internal impedance that could drain C2. Therefore after setting it as OUTPUT and LOW, we set it back to INPUT mode again, because as input the I/O pins of the Arduino have a fairly high impedance.

R1 limits the current that can flow through the sinking pin to abt 9mA (in reality some 6mA). Ofcourse R1 adds to the  time needed to discharge C2, but that does not pose a problem here. (If not given enough time though, the Watchdog will need less than 69 seconds before it is able to send a reset signal again.)

int HeartbeatPin = 8;

void heartbeat() {
// Sink current to drain C2
pinMode(HeartbeatPin, OUTPUT);
digitalWrite(HeartbeatPin, LOW);
// Set pin back to High Impedance
pinMode(HeartbeatPin, INPUT);

void setup() {
// Send an initial heartbeat.

void loop() {
//your code

The above is just a very simple example. Ofcourse it is also possible to check specific sections of the program such as say succeeding an internetconnection or  attaching to a serial port. That way you could avoid  that only part of the program (e.g. connecting to internet)  failed, while the processor is still happily kicking the dog on every loop
The end result however will always be a total reset

Reverse polarity protection with a P-MOSFET

Reverse voltage protection with P-FET
Reverse voltage protection with P-FET

Reverse polarity protection can easily be done with a single diode. The problem with that however is the voltage drop (usually 600mV) that with  large currents can mean a substantial powerdissipation. Using a Skottky diode for its smaller voltage drop is risky as the reverse current through a Skottky is not negligable.
A better way is to use a P-MOSFET as shown in the circuit above. This method is explained very well  in an Afroman youtube video.
Though the direction of the P-Fet may seem a bit counterintuitive, but realise that when the Vgs is in conductive range, the FET will  basically operate as a Switch or in fact a  resistor with very low resistive value, allowing current in both directions.
However you need to pick a FET with low VGS(threshold) so that the device is still offering a tiny volt drop at low battery voltages AND you’ll need a FET with low RDS(on) so that at  your peak current  the FET is dropping less than say 100mV. Other than the FQP4706 the DMG3415U may be a good pick. With 1.8V gate drive it has 0.071Ω resistance.

Switching a humidity sensor off and on

Gardeners who use a resistive moisture sensor to measure the moisture of the soil, sooner or later get faced with corrosion of their metal sensor. Though practically all metals corrode in the soil, in this case it is sped up by the fact that a current flows through the sensor, causing electrolysis.

Several solutions are suggested to solve that like: ‘you have to feed them with AC’ but those solutions are not always practical.

There is however a fairly simple method to minimize the electrolysis and that is to switch off current through the sensor when you do not need it.
Moist soils do not need constant measuring, maybe 3-4 times a day is enough, so if you switch off the sensor except for the say 3x a milisecond or so that makes a big difference.

Easiest might be to use an I/O pin to feed the sensor, but not everybody feels comfortable doing that and not all microcontrollers are suitable for that (although the arduino probably is with 40mA per pin).

But it is in fact quite easy to use a transistor -controlled by an I/O pin- to switch the sensor on and off. Figure 1 gives an example of how to do that with either low side switching or high side switching.

Figure 1a is a very basic transistor switch: Pull the base of the transistor HIGH and current will flow through the sensor, pull it LOW and the current will be switched off. Remember there will be a slight voltage drop (say 250-450 mVolt) over the transistor.

Fig 1b is also a rather basic transistor configuration, it is the emitter-follower or common collector that is often used in voltage buffering. Here also a LOW on the base will switch the transistor OFF and a HIGH will switch it on. Also here one has to take a voltage drop into account (basically the voltage on the I/O pin minus 0.6 Volt). Mind you that Fig 1&2 just give a basic circuit. You’ll need to add a base resistor and possibly a pull-down or pull-up  resistor, depending on  what state you want the circuit to start in.

Figure 2 shows two other circuits you may be considering to try, but that’s not necessarily a good idea.
Fig 2A is basically identical to Fig 1A, but with the sensor and series resistor switched position in the voltage divider.
That does not work because even when the transistor is switched off, there still could be a current flowing through the sensor via the internal impedance of the microcontroller input.
Figure 2B shows a PNP transistor used. That should work right? After all it is basically figure 1A but then reversed in polarity. Well, it is not that simple. The I/O pin is tied to a microcontroller running at say 5V, so its voltage level will vary between 0 and a bit less than 5 Volt. At best it might be 4.2 Volt.*

To keep the PNP transistor switched off, we need to keep the base voltage at nearly the same level as that “+” line and that is hard  in this configuration. Therefore opening the transistor is not a problem, but closing it might, although it will ‘probably’ work

Mind you that Fig 1 and 2 just give a basic circuit. You’ll need to add a base resistor and possibly a pull-down or pull-up  resistor, depending on  what state you want the circuit to start in.

*A pFET on the other hand is quite suitable for high side switching



The LM335 and it’s serieresistor

alibrated mode
Fig 1: LM335 in Calibrated mode

The LM335 belongs to a range of a fairly accurate and simple to use temperature sensors.

The LM335 operates from -40°C to 100°C. It’s output is lineair with 10mV/°Kelvin (¶ 6.6 in the datasheet). It can be used in two ways: Calibrated and uncalibrated mode. The difference is a  10kΩ potentiometer that is used to regulate the output voltage to 2.98 Volt at 25°C (25°Celsius is 298.15°Kelvin) . With more and more devices running off of 3.3 Volt, the question is sometimes raised: will the LM335 run off of 3.3 Volt as well. Going through the datasheet you will not see a min or max value stated for the voltage to be used as all depends on the current through the LM335 and that current is determined by a series resistor (R1 in Fig. 1)

If we want to use the calibrated configuration of the LM335 then this is how to calculate the series resistor for the LM335:
According to the datasheet, the potentiometer needs to set the voltage over the LM335 at 2.98 Volts at  25°C (¶ 6.4 in the datasheet). The LM335 needs between 400uA and 5 mA (¶ 6.2 in the datasheet). Suppose that the max temperature we want to measure is 30 C (=303.15 °K). The output then will be 3.0315 Volt (let’s say 3.03 V).
At 3.03V the current through the 10k potentiometer will be 3.03/10k=303uA. The minimum current that needs to flow through the LM335 is 400uA, so R1 needs to supply 703uA over a voltage of Vcc-3.03V.

So at 5 Volt that would be 1.97/0.703 = 2.8kΩ
At 3.3 Volt it would be 0.27/0.703 = 0.384kOhm =384Ω
(Note 703uA= 0.703mA = 0.000703A. I chose to divide by mA so I get outcome in kΩ).
Now we still have to check if the current will not be too high at the lowest temperature we want to measure:
Let’s say our minimum temperature is 0°C that is 273.15. At 10mV/°K that means an output voltage of 2.73 Volt.
At 5 Volt the resistor current will be (5-2.73)/2.8kΩ = 810uA. We still need to subtract 2.73V/10kOhm=273uA which flows through the potentiometer so we have 810-273=537uA, which is well below the 5mA max.

If we calculate that for 3.3Volt, we find the following:
The resistor current will be (3.3-2.73)/384=1.5mA. Subtract 273uA that flows through the potentiometer and that gives roughly 1.23mA, well within limits.

If you want to measure temperatures higher than 30°C then R1 has to be smaller. When using 3.3 Volt the  theoretical maximum temperature measurable is 330°K which is about 57°C. In practice however that does not work as you need a drop voltage over R1 so the value of R1 cannot be zero, or, in other words: you cannot connect the LM335 directly to Vcc and still expect to get an output lower than Vcc.

Bet let’s see how that would be at 50°C (323.15°K) and 3.3V Vcc :
Well again you need 703uA but this time over 0.07V.
That gives a value  for R1 of 100Ω.
But then at 0 degrees the current is 5.7mA. Still need to subtract 2.73uA which gives 5.4mA which is out of spec.

A program would be as follows:

//A0 is used to read the LM335
void setup()
void loop()
int rawvoltage= analogRead(A0);
float millivolts= (rawvoltage/1024.0) * 5000;
float kelvin= (millivolts/10);
Serial.println(" degrees Kelvin");

float celsius= kelvin - 273.15;
Serial.println(" degrees Celsius");

float fahrenheit= ((celsius * 9)/5 +32);
Serial.println(" degrees Fahrenheit");


Beware though that the ESP8266, in spite of it being a 3V3 device only wants a maximum of 1 Volt on its ADC

Upload Data to Thingspeak through MQTT with an ESP8266

Sending  values to Thingspeak via the Thingspeak API is well known. There is another way as well: through MQTT. Thingspeak has recently (5 dec 2016) added a (one way) MQTT broker for this at

There are two topics one can use:
To upload more than 1 field in one session use:

To upload an individual channel use:
channels/<channelID>/publish/fields/field1/<channelAPI> (just using field1 as example)

In the first case, the payload string is as follows:

In the second case the payload string is just <value1>

In the program below I am using the PubSubClient from Knolleary. The “credentials.h” file is a file that defines my WiFi credentials, you can either create such a file yourself or just insert your wificredentials.

I am using an ESP8266 to make the connection but ofcourse it is also possible to use an Arduino with Ethernet connection when you make the proper changes  in this file in order to connect to Ethernet.

To avoid using again a DHT11 as an example, I show uploading variables by using micros() and a counter

#include "PubSubClient.h" //Knolleary
#include  <ESP8266WiFi.h> //ESP8266WiFi.h
#include   <credentials.h> //This is a personal file containing web credentials

const char* ssid = WAN_SSID;// this constant is defined in my credentials file
const char* password = WAN_PW;// ditto
//char* topic="channels/<channelID/publish/<channelAPI>
char* topic = "channels/123456/publish/T8I9IO457BAJE386"; 
char* server = "";

WiFiClient wifiClient;
PubSubClient client(server, 1883, wifiClient);

void callback(char* topic, byte* payload, unsigned int length) {
  // handle message arrived

void setup() {
  Serial.print("Connecting to ");
  WiFi.begin(ssid, password);
  while (WiFi.status() != WL_CONNECTED) {
  Serial.println("WiFi connected");  
  Serial.println("IP address: ");

String clientName="ESP-Thingspeak";
  Serial.print("Connecting to ");
  Serial.print(" as ");
  if (client.connect((char*) clientName.c_str())) {
    Serial.println("Connected to MQTT broker");
    Serial.print("Topic is: ");
    if (client.publish(topic, "hello from ESP8266")) {
      Serial.println("Publish ok");
    else {
      Serial.println("Publish failed");
  else {
    Serial.println("MQTT connect failed");
    Serial.println("Will reset and try again...");

void loop() {
  static int counter = 0;
  String payload="field1=";
  if (client.connected()){
    Serial.print("Sending payload: ");
    if (client.publish(topic, (char*) payload.c_str())) {
      Serial.println("Publish ok");
    else {
      Serial.println("Publish failed");

The file is available for download here. Whether this is a better method than with the api remains to be seen.
Currently the connection time is limited because of the limited number of sockets on Thingspeak so it is ‘connect->upload->disconnect’. Thingspeak currently cannot be used as a ‘broker’. The traffic is one way only. If your client is already connected to an MQTT network on your own private or public broker, then this method cannot be used without ‘bridging’ the two ‘networks’

Calculating Sunrise and Sunset on Arduino (or other microcontroller)

Knowing the hours of sunset and sunrise may be handy in a variety of situations, an automated chicken coop door might be only one example.

There are several ways to get the proper times: a lookup table in EEPROM, the Timelord library or one of its successors, the Dusk2Dawn library, a rather complicated calculation including the Julian calender, or a fairly simple approximation that I will discuss here.

This method uses the average of the earliest and latest sunset and then for any given day  adds or subtracts a certain amount of time with a maximum of half  of the difference between the earliest and latest sunrise.

Rob Tillaert discusses the method here. It presumes that the sunrise time follows a (co)sinoidal function. I will try to visualize it with a simple example:

Say that on June 23 the earliest sunrise of the year occurs at 4am, and that the latest sunrise of the year occurs at 23 December at 6 am.

Then you know that on any other day the sunrise is between 4 and 6 am. If you take the average that is 5 am, then you know that every other sunrise that year is either 0-1 hr later than 5 am or 0-1 hr earlier than 5 am.

It is the latter that is captured in the formula:


  • avg is average sunrise time, in minutes since midnight
  • Δ the difference between the earliest and latest sunrise time
  • doy is the day of the year
  • the 8 is there because we start on the wintersolstice: 23 December is 8 days before jan 1
  • 58.09 is 365/2π. That is necessary because the cosinusfunction has max 2π as input.

If you live in a DST zone, the earliest sunrise wil be under DST, however you need the non-DST corrected time: the sun knows no DST. Calculate firstm then add DST later

For my location the earliest and latest sunrise are:

earliest sunrise is at 4.19 am
latest sunrise is at 8.51
in order to use them in our equation, we have to calculate them in minutes past midnight:
4.19= 4×60+19=259
8.51= 8*60+51=531
The average is (259+531)/2=395
the difference or delta is 531-259=272. We need half of that which is 137.
The equation then becomes:


To check the accuracy of the approximation, I plotted the actual sunrise times (blue curve) against the calculated sunrise time (red curve).

As it shows, the first half of the year is a perfect fit, the second half of the year seems to follow a more linear curve with the max deviation being 20 minutes, that may or may not be accurate enough for your project. With the aid of this curve though I could opt for a linear approximation for the 2nd half of the year.

For sunset we can practically use the same formula, be it that we now have to subtract the variable part rather than add it.
For my location the sunset is as follows:
latest: 22:07 =1327
Earliest: 16:27= 987
delta=240 ->170

That gives the following graph:

This time I didnt bother to enter all the  real sunset times, but it is clearly visible that there is a reasonable fit that could maybe be enhanced a bit by shifting it slightly more to left or decreasing the delta a bit. Again Red graph is the calculated sunset, the blue is the actual sunset. None of the graphs has been corrected for DST.

A procedure for the Arduino would look as follows:
Where DST is a byte indicating whether DST is active (1)  or not active (0).
The day of the year I pull from my RTC library but it can also be calculated as follows:
int(((month-1)*30.5)+dayOfMonth)  (that is an approximation though)

The sunrise and sunset are both given in minutes after midnight. The hour and minutes of the sunrise (and sunset) can be calculated by: