Reset a program from a ‘freeze’

While doing a test run on software for an incubator the software did fine, until suddenly on the ninth day weird things started to happen: the LCD showed some odd data and in spite of a low temperature the heating had not kicked in. I resetted the software, wondering what it could be but only an hour later again something off happened. This time in spite of the temperature being well over the upper limit, the heating was not switched off.
I checked my code, which really wasn’t so complicated: Read DSB1820, read DHT11, compare with high and low limit, switch a pin on or off and write the value to an LCD and I couldn’t find a single mistake and don’t forget, it had run flawlessly for 9 days. I suspected my display. This was a 20×4 LCD with I2C module with 4k7 pull up. Nevertheless there might still have been some ‘static’ on the SDA and SCL lines that supposedly can cause the Arduino to freeze.

Obviously that is not good if you are running an incubator as you dont want to check it and find your eggs boiled.
So, other than maybe straightening out the LCD cables a bit, I decided that I needed some software protection against ‘freezes’
The only (and possibly best) way to do this is with the watchdog timer. I dont want to go into the specifics and the background of the watchdog timer, but just keep it on a practical level.
What we do is to set-up the watchdog timer to initiate a system reset after say 4 seconds. Then in our loop we do a reset of the watchdog timer so it starts counting from zero again. So as long as the program tells the watchdog timer “I am still running” nothing will happen. Should the program freeze up, it will not reset the watchdog timer and then after 4 seconds the watchdog timer will reset the entire system.
It is very well possible to use the watchdog timer by manipulating the various registers yourself, but it is much simpler to use the watchdog libary that is part of the avr libraries.
We do this as follows:

#include <avr/wdt.h>

void setup()
{
	wdt_disable();
	//wdt_enable(WDTO_1S);// 1 sec
	wdt_enable(WDTO_2S);// 2 sec
	//wdt_enable(WDTO_4S);// 4 sec
	//wdt_enable(WDTO_8S);// 8 sec
}

void loop()
{
	wdt_reset();
	//  your program
	........
}

Solar powering an Attiny or Arduino with a capacitor, or just use AA? Part 2

Solar Charging a capacitor
Solar Charging a capacitor

As I had a 10.000uF (0.01F) capacitor and was awaiting my supercapacitors, I was wondering if I could perhaps use that, just to carry a circuit through the night.

With the aid of a 9 Volt Solarpanel (and a resistor) the 10.000uF capacitor was charged to 10Volts fairly quickly (it was a sunny day).

The charge in that can be calculated by 1/2*0.01*10² =0.5Ws. But as we can only use it at 5 Volt level in reality that is 0.125 Ws.
Also, as we take 1.8 Volt as our lower limit, we have to subtract 0.0126 Ws, leaving us with 0.109Ws.

With the earlier stated power consumption of 13.8uWatt that leaves us with about 7900 seconds or 2 hours and 12 minutes, which is obviously not enough to carry us through the night.
Yet, I wanted to test whether I was on the right track, so I used the following circuit to connect my circuit to the fully charged capacitor:
5volt

As it turned out, my circuit lasted for 1 hour and 25 minutes before it appeared to be dead, which i guess is close enough to the calculation, considering it was an old (>20 yrs) capacitor that might have suffered and maybe  had a bit of a leak current. Now before I get criticized that I am not taking into account that the capacitor is also discharged  during the day because it has to feed the circuit then as well… yes that is true, but I merely wanted to see if the capacitor could store enough load to begin with, to carry me through the night, which it cant. The charge -and thus the capacitor- would need to be at least  some 9 times bigger (so a 100.000uF minimally). I could have figured that all out just by calculation, but nothing beats a field test to see if calculations are right.

Anyway, while waiting for my supercapacitors to arrive (aliexpress) I had some time to think about what solarcell to use. The 9 Volt (5 Watt I seem to remember) seemed definite overkill, but from a few old gardenlamp I had some small solarpanels that delivered 2.5 Volt. I could put 3 of those in series (as they only give 2.5 Volt on a sunny day). I had no idea what Wattage they are but they are used to charge a 1.2V 40mAh NiCad, and some 300mAh NiCad. The latter seemed a better choice.
Ofcourse I could buy a solarpanel, but I first wanted to try with what I had available.

In part 1 I calculated a 1.5 Farad capacitor at 5 Volt is charged up to 19.5 Joule or 19.5 Wattseconds.
So if one wants to charge that capacitor with say a 5 Volt 1 Watt solarpanel one would need  a minimum of 19.5 secs to charge it (if we forget about a series resistor for a while).

IMG_20160505_172103Thriftstores sell these small solar Garden lights  usually for less than a euro. They have a small solarcell and inside a circuit build around an YX8018 IC that charges a single NiCad cell. Sometime sometimes a very small one (40mah)  sometimes a bigger one (800mAh). They need a full day of sun to charge and at dark an LED  lights up for a few hours. Now suppose the small solarpanels from my garden light are indeed enough to charge a 1.2 V 40mAh cell if left to charge for say 8 hrs.
The cells are  1.2 *40 mWh =48 mWh. If charged over 8 hrs that leaves a solar cell of 6mW. As the cells give 2.5 Volt in bright sun light, I would need 2 which then gives a total of 12mW (this is just theoretical, it is very low for a solarcell). So if I need a charge of 19.5W.s, that comes down to (19500/12=) 1625 secs or 27 minutes or 0.5 hrs to charge the 1.5 Farad capacitor. That is doable (if we  forget about the poweruse in the daytime).
In that 0.5 hrs obviously 13.8uW*0.5=6.2uWh is used just for the working of that circuit, which is small enough to ignore in terms of charge time. So, if a 1.5 Farad takes 0.5 hrs to charge, my 10.000uF should take 20/100 m or  12 secs. Lets see.
Now before I go on, just a word of warning: the calculations I do are partly limit calculations and do not always take all practical realities into account. For instance, a 5 Volt 12mW cell will be able to deliver 12/5=2.5mA. However if one tries to load an empty capacitor  with a series resistor of say 100 Ohm, then the max current that is requested is 50mA, obviously the photo cell cannot deliver that so the voltage drops and therefore the classical natural log curve of  charging a capacitor will not be followed. For any given moment the curve will follow the natural log curve for the  voltage at that moment, leading to a longer charge time.
Just a word: One could use the innards of the cheap garden light so, including the circuit around XY8018 IC to charge the caps, if one brings it up to 5 Volt. Various  circuits for that exist. I however decided to keep it simple and do a straight laod from  one or more  fotovoltaic cells.
Nevertheless, as halfway through my experimentations my Ultracaps arrived from Aliexpress (very fast delivery), it didnt seem much use to experiment with the 10.000uF any longer, so in Part 3 We go to the real work.
By the way, someone has already done some work on this.

 

Solar powering an Attiny or Arduino with a capacitor, or just use AA? Part 1

If you have a processor in a remote location that you cannot or will not bring a psu cable to, you have several options of supplying it with energy.
Batteries come to mind and these could be rechargeable or not, or one can use a capacitor. After all, those store energy as well. To make a capacitor a viable option, obviously it needs to be a big one (in capacity, not especially in size), and one has the option of adding a solar cell to it.
As I wanted to feed an attiny that is somewhere in my garden and that is asleep most of the time and as I had some idle solar panels  (9 Volts, 2 Volts) those seemed  suitable to use, but as they do not provide much energy during the night, I needed to store that energy as well. As storage Nicads or even a LiPo came to mind. A Lipo I discarded as it takes some extra circuitry to do any decent charging. A nicad obviously needs some consideration in charging and discharging  as well, but they are more tolerant albeit that a deep discharge is usually not appreciated by them, nor is a constant charge.
Therefore I also considered using a “supercapacitor” I have a capacitor of 10.000uF but one can hardly call that a supercapacitor, but nowadays supercapacitors are readily available at modest prices. They would be easy to use in a charging circuit (basically a resistor would be enough) and they don’t mind frequent  charge and discharge. As their charge is  limited, obviously I would need a solar-cell, so the only thing I needed to see is if they would carry my application ‘through the night’

The Charge of a capacitor is expressed in Coulomb (symbol C).
this can be expressed as:

C=F * V  (where of course ‘F’ stands for the capacity in Farad)
Energy in Joule is expressed as:
J=C*V
Which is:
F*V*V  of F.V²
However, at a given voltage and capacity only half of the energy is available to charge the capacitor hence the energy in the capacitor will be:
J=½ * F *V²
and Joules can also be expressed as ‘Wattseconds’:
J=W *s

So if we calculate the number of Joules in the capacitor at 5.1 Volt:
J=0.5 * 1.5 * 5.1² = 19.5 J
Now obviously we will not all use those as the attiny kinda stops working at 1.8 Volt. If we calculate the number of Joules still left then we come to
J=0.5 * 1.5 * 1.8² = 2.43 J
So the amount of usable energy comes to 19.5-2.43= 17J (rounded down)

From an earlier project on  an attiny13 that was put to sleep only to awake every so many seconds, I found out that the current consumption was 5.5 uA at 5 Volt. Now obviously it will consume a bit less when the voltage drops, so lets assume 4uA at an average of 3.45 Volt (The average voltage being (5.1+1.8)/2=3.45 Volt).
We can calculate the average power consumption as V*I:
3.45 * 4*10⁻⁶ = 13.8 uWatt
as we have 17 Joules available and 17 joules being 17 Ws (Wattseconds) we can calculate the time:
J=W*s
s=J/W
s=17/(13.8 *10⁻⁶)
s=17/13.8  *10⁶
s=1.23 *10⁶
s=1230 000 sec (1,23 million seconds)
that is:
20.531 min, is
342 hrs is
14 days
Now this is just an approximation as there might be some current leakage. The processor will need to do something when awake, like flash a  led or send something, so it will probably not make the full 14 days, but that is not important as it is obviously enough to carry us through the night and probably an overcast day as well.

So how would this be if I used normal AA batteries or NiCad batteries?
As example for regular alkaline batteries  I will take the ubiquitous Philips  LR6. Fortunately someone else already calculated the energie stored in these batteries  at  1986 mAh at a discharge of 100mA. As the performance of these batteries is increasing when the discharge decreases, I will use 2000mAh for my calculations.
If one uses 3 of those to get to 4.5 Volts that will be a total of 9000mWh or 9 Wh.
Again, we will not be able to use this all as we have 1.8 Volt as lower limit, so we will be stuck with 3.6 Wh that we cant use, leaving  5.4 Watthour usable energy.
The average Voltage on the  processor will be:
(4.5+1.8)/2=3.15 Volts.
the average power consumption thus is:
3.15* 4*10⁻⁶ = 12.6uW
as we have 5.4 Wh available we can again calculate the time:
h=5.4/12.6 * 10⁶
h=0.428571 *10⁶
h=428571 hr
=17857 days
=592 months
=50 years
Wow. Why would anybody want to use anything else but LR6 batteries ?
Well for several reasons: 3 of those batteries take more space and eventhough they tend to perform well at low temperatures and low discharge rate, we all have experienced that Alkaline batteries left in a hardly used device tend to leak, even destroying that device by its corrosive leaking. Advantages and disadvantages of alkaline batteries are discussed here.
Nevertheless using alkaline batteries can be a good choice, but don’t forget to change them.

Using NiCad batteries gives a somewhat similar calculation. Suppose we use 3 cells of 1.2 Volt 2000mAh (or 2Ah).
Total energy will be 3.6*2= 7.2 Wh
useable energy will be (3.6-1.8)* 2=3.6 Wh
the average voltage we set at (3.6+1.8)/2=2.70 Volt
The average power consumption will be:
2.7* 4*10⁻⁶ = 10.8uW  (V*I)
The time will thus be:
h=(3.6/10.8) * 10⁶
=333333 hr
=38 yrs
Now if that only was true. My experience with  for instance solar garden lights (that often only have an 800mAh cell)  is that they usually dont last through the winter… and mind you, those are being recharged (which could contribute to their demise).
Nevertheless they may be a good choice.

In a following article I will discuss some practical examples

 

1 Wire LCD interface: Part 1

The one wire LCD interface has been around on the internet for some time already and made popular by Roman Black, although he claims he is not the inventor of the principle. His article is mainly PIC processor oriented, but it gives a good explanation of the principle. Myke Predko is also a name that  should not be unmentioned as pioneer in 1 wire LCD driving. Also this article about  one wire driving of a shift register is very informative

The ShiftRegLCD123 website from Raron used to give a lot of information as well, along with circuits, but that (the Wiki) seems to have disappeared. But his library is in codebender.

In brief, the One Wire interface uses a latched shift register in which the clock and the latch are taken from the data signal through two RC networks that produce the necessary delay. Circuits are a plenty on the interweb, but finding the proper driver for it can be confusing sometimes.

Although the interface on the data, clock and latch lines is rather standard, the connection between the shift register and the LCD has thre major variants that for now i will just call the Roman Black, the LCD3Wire circuit (yes, it is 1 wire) and the ShiftRegLCD123 circuit. Also Elektor has published a 1 wire interface (Detlef Hanemann)(Published in issue 9/2015 on page 92) with yet again a different shift register to LCD connection. Their setup is said to have several advantages over the setup used by Roman Black, it uses the Q7′ direct output of the shift register to trigger the output latch register to carry over automatically the shifted data to the output pins. But for that it requires a monoflop built around a BS170 FET to generate the E-pulse. The setup to be used in Francisco Malpartida’s LCD library is yet again different from the other 3 and uses a diode to create an AND gate

1wireLCD
Roman Black

1wireLCD-shiftreglcd

ShiftRegLCD123

1pinlcd-elektuur
Elektuur

onewiremalpertida

 

Malpartida LCD library

. Roman Black LCD3Wire Shiftreglcd Elektuur Malpartida
RS  Qc  Qb  Qc  Qg  Qg
R/W  GND  Qc  GND  GND  GND
E  Qd  Qd  Qh  Qh’  Qh’ sort of
D4  Qe  Qh  Qd  Qa  Qe
D5  Qf  Qg  Qe  Qb  Qd
D6  Qg  Qf  Qf  Qc  Qc
D7  Qh  Qe  Qg  Qd  Qb

Driving these OneWireLCD’s thus differs per configuration. The ShiftRegLCD123 library can do both the ShiftRegLCD and the LCD3Wire protocol. The Liquid Crystal Library of Francisco Malpartida  does have a One Wire protocol but make sure you have the latest version. However, it relies on connections different from the others mentioned:
// Bit #0 (QA) – not used
// Bit #1 (QB) – connects to LCD data input D7
// Bit #2 (QC) – connects to LCD data input D6
// Bit #3 (QD) – connects to LCD data input D5
// Bit #4 (QE) – connects to LCD data input D4
// Bit #5 (QF) – optional backlight control
// Bit #6 (QG) – connects to RS (Register Select) on the LCD
// Bit #7 (QH) – used for /CLR on the HW_CLEAR version (cannot be changed)
// (Q’H) – used for Latch/EN (via the diode AND “gate”) (cannot be changed)

// -----------------------------------------------
//
//                         74HC595     (VCC)
//                       +----u----+     |          2.2nF
// (LCD D7)------------1-|QB    VCC|-16--+      +----||----(GND)
// (LCD D6)------------2-|QC     QA|-15         |
// (LCD D5)------------3-|QD    SER|-14---------+--[ Resistor ]--+
// (LCD D4)------------4-|QE    /OE|-13--(GND)         1.5k      |
// (BL Circuit)--------5-|QF    RCK|-12---------+                |
//                       |         |             \               |
// (LCD RS)------------6-|QG    SCK|-11-----------)--------------+--(Seri
//                     7-|QH   /CLR|-10--(VCC)   /               |
//                  +--8-|GND   Q'H|--9---|<|---+--[ Resistor ]--+
//                  |    +---------+     diode  |      1.5k
//                  |                           |
//                  |      0.1uF                |
//                (GND)-----||----(VCC)         +----||----(GND)
//                                              |   2.2nF
// (LCD EN)-------------------------------------+
// (LCD RW)--(GND)
//

The Elektuur/Elektor configuration has its own program, or here (both direct download link)

A capacitive soil humidity sensor: Part 3

Initially I imagined the probe as a sleek device with a wire coming out that one would stick in the soil, if necessary completely. This would mean that the circuit pcb would be an integral part of the PCB that  formed the plates, but as always, things go different in practice. For one thing , the piece of PCB I intended to use just wasnt long enough. Second, my PCB etching possibilites are temporarily impeded. Third, I wanted to add an LDR, meaning that I had to have some sort of  translucent cover.

I constructed the circuit on a small piece of stripboard:

IMG_20160225_163125 i2cprobe

With regard to the LDR, sure that is nonsense. If I want to add an LDR I  could just as well add it to my base station. But I am not adding it because I have to, I am adding it because I can and because I wanted to get some experience in programming an I2C slave. An LDR might not be so usefull but in future I may want another sensor, e.g.a sensor that reads if there is really water flowing from the irrigation tube.

Also I had decided on putting the capacitor plates back to back, but as I did not have double sided PCB I just used two pieces glued and  soldered(!) together.
So My BOM was rather simple:

  • 2 equal size pieces of PCB  Size depends on what you have, but do not make them too small. I used 12×3 cm.
  • 1 piece of 0.5-1.0 cm plastic for a baseboard. I used an old cutting board.
  • 1 clear/translucent cover, I used the lid of a whipped cream spray can.
  • 1 piece of  thin 4 wire cable, length depending on your need

pcbprobe2I glued the two piecesof pcb back to back. drilled a hole in all 4 corners and through soldered a piece of wire through each hole, thus anchoring the plates together. I removed some copper around the solder so it would become an island isolated from the rest of the plate. (See picture).

Soldering the plates together in the corners may not be necessary if you decide to electrically isolate them from the soil with e.g. shrink tube.

Eventually I will place an NTC on the plate as well after it is covered with Shrink tube. Esthetically it might be better to put the NTC under the shrink tube, but that could create an air pocket.

 

baseplateI made the base plate from a o.5 cm thick piece of soft plastic. Cut out a round shape  with a 5.5 cm diameter to fit the base of my  clear dome and  made a slit 3 x0.3 cm into that in which the PCB  fits snug. made a round hole for the connecting cable.

IMG_20160226_133820

 

 

I soldered two wires on the top of the PCB, one on each side. Soldered wires on an NTC, insulated those, attached the NTC to the bottom of the PCB with the wires leading to the top and then covered the PCB with shrink wrap.

probeEventually the probe looks like this (picture)probe3

 

A capacitive soil humidity sensor: Part 2

capacitive74HC14_attinyIn a previous article I presented a simple way to read a capacitive moisture sensor with a simple RC generator.
In this post I will present a sensor with some added functionality that can be read through I2C. The circuit (figure) doesnt need much explanation: the RC generator we saw in the previous article and the two variable resistors in a voltage divide read by analogue inputs. The Attiny 45 (NOTE, it is NOT an Attiny25) is the heart, or rather the brains of the sensor. As the Attiny  functions as an I2C slave we will need the TinyWireS library. The library comes with some examples and one example was quite easy to rework to what I needed. The code is as follows.

#define I2C_SLAVE_ADDRES 0x4
#include <TinyWireS.h>//https://github.com/rambo/TinyWire
#ifndef TWI_RX_BUFFER_SIZE
#define TWI_RX_BUFFER_SIZE ( 16 )
#endif

volatile uint8_t i2c_regs[] =
{
	0x00,
	0x00,
	0x00,
	0x00,
};
// Tracks the current register pointer position
volatile byte reg_position;
const byte reg_size = sizeof(i2c_regs);

byte LDRvalue;
byte NTCvalue;
int MoistPulse;

/**
 * This is called for each read request we receive, never put more than one byte of data (with TinyWireS.send) to the
 * send-buffer when using this callback
 */
void requestEvent()
{
	TinyWireS.send(i2c_regs[reg_position]);
	// Increment the reg position on each read, and loop back to zero
	reg_position++;
	if (reg_position >= reg_size)
	{
		reg_position = 0;
	}
}

/**
 * The I2C data received -handler
 *
 * This needs to complete before the next incoming transaction (start, data, restart/stop) on the bus does
 * so be quick, set flags for long running tasks to be called from the mainloop instead of running them directly,
 */
void receiveEvent(uint8_t howMany)
{
	if (howMany < 1) { // Sanity-check return; } if (howMany > TWI_RX_BUFFER_SIZE)
	{
		// Also insane number
		return;
	}

	reg_position = TinyWireS.receive();
	howMany--;
	if (!howMany)
	{
		// This write was only to set the buffer for next read
		return;
	}
	while(howMany--)
	{
		i2c_regs[reg_position] = TinyWireS.receive();
		reg_position++;
		if (reg_position >= reg_size)
		{
			reg_position = 0;
		}
	}
}


void setup()
{
	pinMode(1, INPUT);
	TinyWireS.begin(I2C_SLAVE_ADDRESS);
	TinyWireS.onReceive(receiveEvent);
	TinyWireS.onRequest(requestEvent);
}

void loop()
{
	readSensors();
	/**
	 * This is the only way we can detect stop condition (http://www.avrfreaks.net/index.php?name=PNphpBB2&file=viewtopic&p=984716&sid=82e9dc7299a8243b86cf7969dd41b5b5#984716)
	 * it needs to be called in a very tight loop in order not to miss any (REMINDER: Do *not* use delay() anywhere, use tws_delay() instead).
	 * It will call the function registered via TinyWireS.onReceive(); if there is data in the buffer on stop.
	 */
	TinyWireS_stop_check();
}

void readSensors()
{
	LDRvalue = (analogRead(A2)) >>2; // max value is 1023/4=255 is 1 byte (FF) physical pin3=PB4=A2
        i2c_regs[0] = LDRvalue;
	NTCvalue = (analogRead(A3)) >>2; // max value is 1023/4=255 is 1 byte (FF) pin2=PB3=A3
	i2c_regs[1] = NTCvalue;
	//pulsepin= pin6=PB1
	MoistPulse = pulseIn(1, HIGH);
	i2c_regs[2] = highByte(MoistPulse); // or use = MoistPulse >> 8;
	i2c_regs[3] = lowByte(MoistPulse); // or use = MoistPulse & 0xFF;
}

I have called the pulseIn function with an integer rather than with a long. This means that you will have to choose a value for R3 that gives a reasonable range for the type of soil that you use. A range of say 0-200 uS is very reasonable. Once you have done that you can also add a timeout to the pulseIn function. This should be abt 2 times the pulselength you expect (but depends on the duty cycle). With regard to the two variable resistors, they are in pull up so their value is Rntc = Rseries/((1023/ADC) – 1));  For the NTC this could be substituted in a Steinhart Hart approximation

The code to call the values (and that is loaded onto the master arduino) is even simpler:

#include <Wire.h>
void setup() {
  Wire.begin();        // join i2c bus (address optional for master)
  Serial.begin(9600);  // start serial for output
}

void loop() {
  for (byte i=0;i<4;++i){
  Serial.print("0x");
  Serial.println(readRegister(i),HEX); // print the character
}
Serial.println(" ");
    delay(1000);
  }

uint8_t readRegister(uint8_t regaddress)
{
	Wire.beginTransmission(4);
	Wire.write((byte)regaddress);
	Wire.endTransmission();
	Wire.requestFrom(4, 1);
	return Wire.read();


 delay(500);
}

This code only prints out the values, you still need to combine the LSB and MSB from the cycle time and e.g. switch a pump based on the value. You could do that with this function:

int combine (byte lsbv, byte msbv)
{
	int value = msbv << 8;
	value = value | lsbv;
	//value= msbv<<8 | lsvb;//if you want to do it in one go
	return value;
}

Don’ t forget that the I2C lines need pull-up resistors of 4.7-10k.

In a next article I will present the construction of the probe itself.

A capacitive soil humidity sensor: Part 1

An automated plant/garden watering system is a popular application for  the Arduino and other microcontrollers. The humidity sensor that is often used is  often a resistance meter: 2 probes in the soil form a resistor and as part of a voltage divider that gives info on the amount of water in the soil.
The major shortcoming of these sensors is that since there is a current flowing, the probe is sensitive to electrolytic corrosion. Another shortcoming is that it doesnt really give information on the soil humidity, but more on the  ion concentration in the soil. Remember: pure water is a bad conductor, It are the ions in the water that make it a conductor.

To get around the corrosion, people have started to feed the probe with AC rather than DC current, but at best the circuits for this supply a pulsating DC current. Another  possibility to minimize corrosion is to  switch off  the current to the probe and only switch it on when a measurement is taken, say every 5  minutes.

All these measures help but even then, just due to the contact with moist soil, the metal on the probe will corrode and weather.
Another method is to encase the probes in plaster. Some people swear by it but I  think it is bothersome. Also the plaster encased probe has a lag as it will retain water  for some time after the  soil is already dry and it will still be dry for some time after the soil is already moist.

Capacitive measuring is a way to avoid these problems. With capacitive measurement the ‘plates’ of the capacitor can be electrically isolated from the soil and the soil in fact forms the dielectrum of the capacitor. Water makes a good dielectrum, whereas ions dont. Capacitive measuring therefore will have a better correlation with the actual amount of water in the soil than resistive measuring and, as said, ideally there will not be any corrosion.

Although the Arduino can measure capacity, it needs 3 pins for that. Also, it isnt really practical to have a long wire as part of your soil probe capacitor go to your Arduino as the wire and whether it is  straight or curly or in a loop, will act as a parasitic capacity and affect the measurement.
Putting your Arduino right onto your plantbed might also not be the best solution.

capacitive74hc14
RC Oscillator With 74HCT14

A better way to do this is to introduce an RC oscillator in which the soil functions as a capacitor and the Arduino measures the frequency that comes out of the oscillator. More water  will increase the value of the ‘capacitor’ which usually will lower the frequency (or increase the cycle time) and that can be measured with the arduino.

An easy RC oscillator can be seen in the figure to the right. A 74HC14 or 74HCT14 Inverter Schmitt-trigger is all you need. For the HC version the Frequency is:
f=1/T=1/(0.8*R*C)
for the HCT version it is:
f=1/T=1/(0.67*R*C)

But in fact the frequency isn’t really that important because we are not making a frequency meter. We are only interested in changes of the frequency that relate to  dry, moist or wet soil. The workings of this type of Oscillator are explained here. and here. The theory behind the frequency calculation is explained here.

leydenjarI tested my set-up initially with a sort of Leyden Jar as a capacitor. Simply put I taped two pieces of alufoil on the outside of a glass jar, attached wires to it and tried if I could measure a change between an empty and a full jar. I started out using a 100k resistor, but only got a decent range after using a 2M2 resistor: an empty jar gave me 1uSec (maybe less, but I guess that was my lower limit) a half full jar around 50uS and a full jar about 100uS (measured with PulseIn). From using a known capacitor I knew I had to use the formula f=1/T=1/(0.67*R*C), but as said, neither the frequency or the actual capacitor value are of much importance as we are just looking for  changes.

Nevertheless the capacity of this jar (when full) could be calculated from
100*10⁻⁶=0.67*2.2*10⁶*C
C=(100*10⁻⁶)/(1.47*10⁶)
C=68*10⁻¹²
C=68pF
Now ofcourse this isn’t entirely correct as with PulseIn We only measured half a cycle, so in fact the Capacity is more likely to be 136pF, provided the dutycycle is 50% (some sources say it is 50% others say it is 33% with the space  2*mark).

byte pin = 8;
unsigned long duration;

void setup()
{
  pinMode(pin, INPUT);
  Serial.begin(115200);
}

void loop()
{
  duration = pulseIn(pin, HIGH);
  Serial.print("Time ");
  Serial.print(duration);
  Serial.print(" usec ");
  Serial.print(500/duration);
  Serial.print(" kHz ");
  Serial.print(500000/duration);
  Serial.println(" Hz");
  delay(500);
}

Then it was time to do some garden field tests. There are several ways to construct a probe: two pieces of PCB at a distance from eachother, One piece of PCB with two plates etched onto it, or a piece of double sided PCB. With regerd to the latter, we tend to think of a capacitor as two plates with a dielectrum in between, but in fact it can also be two  plates with a dielectrum around it.

capacitive probe-textMy garden field test, using two insulated plates of PCB (picture) kinda gave similar results as with the Leyden Jar so I knew I had a decent design.

Yet, two plates is not really very handy: you have a loose wire connecting the plates and as that forms part of the capacitor it introduces stray capacitance, so I decided to go for either a double sided PCB, or a single sided PCB with two plates etched onto it and then place the oscillator on that PCB as well. Frankly, that idea isnt new as many of the (semi)DIY (e.g. the ‘Chirp’) or commercial capacitive probes follow that design as well.

Now it is always good  that before you make a PCB to think through what you want. As I was using only 1/6th of the 74HCT14, I was thinking maybe I could use the other gates as RC oscillators as well  for e.g. an LDR or NTC. Now ofcourse I immediately dismissed that again as I could just as well use an analog read of those and I would have to run extra cables that would all carry a frequency signal, no doubt those would interfere, but it got me thinking. The ‘Chirp’ has a big brother that has an I2C interface, if I would add that, I could add some other sensors (light, temperature)  to my probe. With just the capacitive sensor I would have to run 3 wires (+Vcc, signal, ground). With I2C 4 wires (+Vcc, SDA, SCL, Ground) would be enough for a number of sensors on my probe. Also, as I have various plantbeds, having a number of I2C probes wouldnt require  more inputs on my Arduino.

Obviously, adding I2C meant adding a microcontroller, preferably a cheap one, which easily brings one to the attiny25/45/85 series. If you have good eye sight and a steady hand, you could consider the soic version that is available from 30-70 cts. If you go for a DIP version, that is around 0.9-1 euro. Considering one can get an entire Pro Mini for about 1.20, obviously that is a narrow gap and I (or you) could consider giving each plantbed a totally dedicated  Pro Mini microprocessor that could take care of the irrigation as well.
Having said that, in the past, I have made a self contained system for plantbeds, just using 1 Opamp, so maybe I should just consider this I2C excercise a ‘learning excercise’

capacitive74HC14_attinyI came up with the following circuit. It is easy to recognise the RC oscillator. The cycle signal is fed to PB1 whereas the analog inputs on PB3 and PB4
A3= PB3=physical pin2;A2=PB4=physical pin3.
Pin PB2 is used as SCL signal and PB0 as SDA signal for the I2C interface.

If for any reason you do not have a 74HCx14 in your toolbox, there are other ways of constructing an RC generator. E.g. a 555 or a 74HC00. Even a single transistor could be used. With this type of oscillator the dutycycle approaches 50%
rc-osc

555-oscillator-probe

In a next article I will discuss how to implement the I2C software.