Step 4: Analog to Digital Conversion

Most of you will want to skip this step.

If you are curious, or having analog problems, or looking to design some analog circuits, this may be helpful.

The ATmega328 has six analog input pins (see pin mapping). I am assuming the chip has only one A/D (analog to digital) converter chip in it. It is called a ten bit converter because it uses ten bits, 210 or 1024. The next drawing is a pretty good representation of how the converter works. The description is pretty technical and not that important for most Arduino users. But look at the analog signal represented by the smoothly curving signal on the top left. The stair step picture on the bottom left is representative of the digital output. You should note that instead of a smoothly curving line like the analog signal, the digital output is very jagged, often called a stair step. For the Arduino, the digital input goes from 0000000000 to 1111111111 which is 1023 steps in decimal.

This digital output doesn’t come out of the ATmega directly. It is only used internally by the software.
For the standard Arduino if you have 5V on the analog pin, the analogRead() will be 1023. Since there are only 1023 different possibilities that means each step is 0.004888 V. What that means is if you have analogRead() of 1023, it might mean 5.000V or 4.996V or somewhere in between. So each stair step means that the actual voltage can be off by 0.004888 V.

Practically speaking: So many of you will say who cares and in most cases you are correct. The accuracy of this reading is dependent on the accuracy of Vcc. If the Arduino is using a 7805 voltage regulator, then the 5V is +/- 1.5% which is 0.075V. The actual voltage could be anywhere for 5.075V to 4.925V which is much greater that the tolerance of a step. Another factor is noise on Vcc and on the analog signal. With a PC USB, the 5V tolerance is usually worse. Mine measured 4.84V.

Practically speaking with the standard Arduino, your Analog accuracy is limited to +/- 1.5% of the 7805 anyway.

NOTE: I am not 100% sure the following is accurate.
Advanced information: The 5V equals 1023 is not always true for the Arduino. The 1023 value is the voltage on the AVcc pin (pin 20 of the ATmega) under most conditions. Usually AVcc is tied to Vcc and is 5V. However, if the sketch contains the function AnalogReference(INTERNAL), then an internal reference of 1.1V (for the ATmega 168 and 328) is used. I am fairly certain what this means is that if you have a 1.1V signal on an analog pin, then the analogRead will be 1023. What this means is each step equals 0.001075V which means you can get more accuracy but it also means your maximum analog input can only be 1.1V. Now AnalogReference() can also be set to EXTERNAL. What this means is the analog reference is equal to the voltage you put on the AREF pin.
WARINING: Be careful if you use EXTERNAL and an external voltage source; this could interfere with and damage the ATmega. If you run a different sketch, that doesn’t have AnalogReference(EXTERNAL) set, then the external voltage reference is going to be fighting the default voltage on AVcc. It’s like connecting two batteries of different voltages together. You have a complete circuit with voltage but zero resistance so lots of current. I think there is a way around this by connecting it through a resistor but am not sure.

The reason I mentioned AVcc and analog reference is that I was setting my Arduino-clone up to use the Nokia LCD. I noticed that LCD worked fairly well when I was using the USB power but it was very dark when I plugged in an Adafruit 9 volt battery adapter. So I grabbed my trusty little DMM and did some troubleshooting. With the USB power, the Vcc and the Aref pin on the ATmega were at 4.84V. With the 9 volt battery connected to the 7805 regulator the two pins were at 5.03V. I also had a temperature sensor attached and further testing verified that the temperature reading varied depending on what the AVcc voltage was.

So if you want to improve upon the accuracy of the ‘standard’ Arduino, there are ways to do it.
If you set AnalogReference(INTERNAL), the ATmega will use 1.1V as a reference, adjust your calculations accordingly.
TIP: Any time you change AnalogReference(), discard the first analog reading as it will be meaningless.

If you set AnalogReference(EXTERNAL), you can provide you own reference, assumed it has a tighter tolerance.
See WARNING above.
TIP: Any time you change AnalogReference(), discard the first analog reading as it will be meaningless.

Those same six analog pins use a different method for output. 
D/A (Digital to Analog): The Arduino does not use a D/A (Digital to Analog) converter. Instead it uses, something called:
Pulse Width Modulation (PWM). PWM is not an analog signal. It is a pulsed digital signal. The drawing is from http://arduino.cc/en/Tutorial/PWM.
What the Arduino sends out is a digital signal that pulses off and on at different rates.

Instruction               Output
analogWrite(0)         0 volts out
analogWrite(64)       see drawing
analogWrite(127)     see drawing
analogWrite(255)     5 volts out

The drawing depicts the output over a period of time. As the drawing shows as the analogWrite value goes farther above 0, the width of the positive pulse gets wider. In other words, the time the output stays at 5 volts is longer. Now this is hard to explain but if you think in terms of average voltage when analogWrite is 0 then the average voltage is 0 and when it is 255, then the average voltage is 5V. Now when it is 127, you can see that it is on ½ the time and off ½ the time so the average voltage is 2.5 volts. Now as the value goes larger than 127, you should see that the average voltage goes up.

For those of you who like real world “proof”, here is a way to demonstrate this. A capacitor smooths out DC voltages. That is why you will always see them on Vcc and power supply outputs. Now if you take a capacitor (I’m guessing a 1ufd or maybe .5ufd) from an analog pin to ground. If you write your sketch to set up the analog pin as output and send an analogWrite(127), your DMM should show the voltage at 2.5V. By the way, the mathematics of PWM is beyond me but I know it gets into AC circuitry which is way beyond what this Lazy Old Geek wants to get into.

To put it simply, PWM puts out a signal that simulates an analog signal for most purposes.
TIP: One thing to remember, while the analogRead goes from 0 to 1023, the analogWrite only goes from 0 to 255, a byte by the way.

Non-essential Info: Most adequately designed Arduinos and clones have a 0.01ufd capacitor from Vcc to GND and next to the ATmega chip. The reason is to filter noise from the digital ATmega. When running a sketch, most of the internal circuitry is trying to switch from 0 to 5 volts almost instantly. This produces a lot of noise, a lot of which is high frequency. The capacitor will absorb most of this and smooth it out, just like it did in our little experiment.
Great 'ible, very comprehensive! <br>Just one question; <br>Why does referencing 1.1v make the analog read more accurate? <br>Does it not also have 1.5% error? <br>Thanks
You are correct, there would still be a 1.5% error. But 1.5% of 1.1V is a smaller voltage than 1.5% of 5V. However, this is only speaking theoretically due to the inaccuracy of the A to D conversion. <br>One of the factors that will affect 'accuracy' is the tolerance of the 1.1V reference which is between 1.0 and 1.2V. <br> <br>However, for real applications, this is pretty meaningless. Generally, I choose a reference voltage based on the sensors connected to the Arduino. If the maximum analog voltage is going to be 1.1V or less than choosing the internal 1.1V reference would probably be best. <br>I usually like to use 3.3V reference as shield compatible Arduinos have a 3.3V source and so do most USB to Arduino converters. These are usually much more accurate then the USB 5V or (I think) the 7805 regulators. <br> <br>LOG
I am trying to hook up a nokia 5110 that is supposed to be the same pin mapping. Is this the pin mapping that the shield uses?<br><br>#ifdef PB1<br>#define LCD_RST PB1<br>#define SPI_CS PB2<br>#define SPI_MOSI PB3<br>#define SPI_SCK PB5<br>#define LCD_DC PB0<br>#define LCD_BL PD7<br>#else <br>#define LCD_RST PORTB1<br>#define SPI_CS PORTB2<br>#define SPI_MOSI PORTB3<br>#define SPI_SCK PORTB5<br>#define LCD_DC PORTB0<br>#define LCD_BL PORTD7<br>#endif<br><br>which when translated for the arduino pins this is how I believe it should be set<br>LCD_RST digital pin 1<br>SPI_CS digital pin 2<br> SPI_MOSI digital pin 3<br> SPI_SCK digital pin B5<br> LCD_DC digital pin 0<br>LCD_BL Pdigital pin 7<br><br>am I missing something?<br><br>
Sorry, my ISP won't connect to Instructables.com.<br><br>I tried to design my own interface to a Nokia LCD but couldn't figure it out so that's why I bought one with software. So I can't really help with the code. I don't see anything obviously wrong with your code but don't know.<br><br>One thing I do remember was that the LCDs were not 5Vdc so a lot of the hacks had ways to convert the signals. I think the LCD expects 3Vdc signals.<br><br>Good luck.
Very well written! Nice and detailed!
Thanks. I get frustrated with internet articles that don't quite tell you what you're looking for so I try to make mine as complete as possible.<br><br>Lazy Old Geek
Do you have any experience getting this display shield to share the SPI bus with other shields? I'm trying to get it to play nicely with a Freetronics ethernet shield (which unlike most w5100 based shields, honors the cs line).<br><br> -- Mitch
Sorry, I haven't. This is my only experience so far.<br><br>LOG
I can tell that you are a very organized, grammatically correct, and SMART, lazy old man. You speak my language and I love how you actually got the results that you desired, good work, very resourceful and patient. Paying attention to the smaller details was always a key highlight in engineering. <br><br>To add furthermore on to your very eloquent and detailed instructable that you have created I would also like to add one small tidbit to your comment on discarding the first reading from the modified AnalogReference() id. <br><br>Your tip that explains not to include the first reading ( and I am speculating, I have never used an arduino or any of the devices listed here ) most likely is due to two consequences. Firstly, when you plug a device into any battery or power source, there will be an initial &quot;pulse&quot; high energy output that comes from the power source. One can interpret this as the phenomenon of kinetic energy from physics, i.e. - there are two <i>different</i> equations for the initial (or static) state of energy and when the energy is in motion. Molecules are put into a state of havoc then ultimately into a more steady state of flow, thus explaining the two different equations. Different batteries and alternative power sources that I know of have different &quot;pulse&quot; currents that initially &quot;pulse&quot; out, just like the Silver Surfer character in the second &quot;Fantastic 4&quot; comic movie. Secondly, the initial reading would be tossed out ( again here, speculating ) because the unit, when modified, might not use the same algorithm as the normal, debugged input. It might not be compensating for the initial &quot;bounce effect&quot;, i.e. basketball effect of when a button is pushed. Digitally speaking, this means that whenever you are typing on your keyboard or pushing a button, there is a microcircuit with an algorithm automatically compensating for the impossibly small bounce that sends multiple &quot;1's&quot; to your next device. To fix this, the algorithm simply is written to ignore the initial bounce gap interval of &quot;1's&quot; and set the gap interval of &quot;1's&quot; to just plain and simply &quot;1&quot;. That's it, and everything runs smoothly.<br><br>Did my comment overcompensate the &quot;geek&quot; effect in this article? ;)<br><br>I hope someone here understands it and appreciates it.<br><br>-Keith
Thank you for your comments. I do try to pay attention to details. Unfortunately, as I get older, it is harder to do. In my case, I think this comes from a couple of factors. One, I spent many years in electronics, mostly developing test systems where missing details could develop serious design flaws. Two, in this work, i had to write a lot of test programs. In debugging and troubleshooting programs you learn to pay attention to the littlest details that won't allow the program to run or won't run correctly.<br><br>I find your comments interesting. I do not fully understand them but definitely see and engineering background and probably highly educated. <br><br>Some of the initial power states, I am aware of as are most designers. I don't know how familiar you are with microcontrollers but at least in the earlier ones I used to work with there was always a resistor and capacitor attached to the the reset pin so there was always a delay that held the microcontroller idle until power was stable. As for the ATmega chip, apparently, there is a watchdog circuit that keeps the microcontroller idle until power is stable. I am not sure if this is what you were referring to or not. <br><br>Here is the data sheet for this microcontroller all 448 pages of it.<br>http://www.atmel.com/dyn/resources/prod_documents/doc8161.pdf<br><br>My suggestion to discard the first sample only refers to conditions where the Analog Reference type is changed and not to initial power up conditions.<br>From the datasheet:<br>'The first ADC conversion result after switching reference voltage source may<br>be inaccurate, and the user is advised to discard this result.'<br>Now, i am not sure but I believe it is using the same algorithm but the hardware comparator reference is changing. Perhaps the reference has residuals of the old reference until after the first conversion?<br><br>I am also familiar with the 'bounce effect' and there are hardware circuits to 'debounce' switches. In the Arduino world, though mostly what you see are little tiny buttons connected to an ATmega digital pin with a pullup resistor. So 'debounce' is usually done in software, if at all. If you do an internet search for Arduino debounce, you will find lots of suggestions and software that didn't work and fixes but it is very hard to find any information on theory and how to make it work if it doesn't. <br><br>After perusing the Arduino forum and playground areas for analog reference information, I was totally frustrated by the lack of understandable information that I could trust. It was probably there, but I have no good way of finding it.<br><br>That is one of the reasons why I chose Instructables to write about what I know, explain some of the theory and put the theory to work. <br><br>So, yes, I really appreciate your comments. It's good to know that someone is actually paying attention.<br><br>LOG
Well, it would appear that you are still as sharp as a tac! I get what you are saying by the initial &quot;startup&quot; states in your pdf now that you explained it. <br><br>I took 4 years of physics and 2 1/2 of electrical engineering, along with another 2 1/2 of advanced math and statistics/finance/economics courses. Let's just say I wanted to get the most for my money out of college ^^ :). <br><br>As far as the &quot;not understanding&quot; part, I was merely referring to patterns I saw while reading your statements in this instructable. There are many equations and laws of physics which can be referenced and applied to the problems that you were having, and apparently already knew of. I just found it fascinating that someone would actually put in as much work as you did in writing an article - and yes, I do appreciate it very much! I learned a lot from your article and it brought back memories from when I used microcontrollers and programs to implement circuit functions (and the headaches!!). But as you say, it is <b><i>very</i></b> hard work. I certainly wish I had you as my teacher instead of foreigners with thick accents that I could not even understand!<br><br>The capacitors are there to compensate for delay like you said and to provide open/close switch circuit operations for different &quot;states&quot; of digital electrical &quot;time frames&quot; - so to speak. For any given output at any given time, there must be a programmed circuit (both hardware and software) to fulfill all of the different combinatorial digital circuit logic functions that one may want, therefor, one must acquire capacitors and transistors to provide for those exact functional outputs. It is indeed very complicated and I did not want to pursue that stressful environment. The capacitors are almost always coupled with transistors in any circuit due to our &quot;digital age&quot;. <br><br>ON....OFF....ON....OFF....etc...<br><br>The capacitor takes care of one part and the transistor takes care of the other. They pretty much work hand in hand, different properties and such with AC and DC current. I'm a sucker when it comes to &quot;why&quot; these things work the way they do. Technology is unbelievably mind boggling, it is very cool to see how you &quot;hacked&quot; the device to work for you and the very hard work you put into it to make it precisely the way you wanted it to be. I still could look at a circuit diagram drawing now and get lost in less than a second, and I'm only 26!! lol.<br><br>I make LED toys now ;). I've been working on new designs for over a year now and I am trying to start up a small business. I'm a geek too :).<br><br>Do you have any other instructables online?
Ha! Yes, I suspected you had a strong physics background!<br>My education was mostly practical, starting with training in the USAF. It was a rather ominous start. I was taking classes during the graveyard shift and all I remember was sitting around playing Battleships. They did get rid of that instructor. Like you I was a sucker for the &lsquo;why&rsquo;, so when I wasn&rsquo;t drunk or sleeping (graveyard shift again) I probably applied my training better than most.<br>After getting out, I later took a correspondence course in microcomputers. Most of my experience came from working in electronics test and measurement for over 20 years. The Internet became such a significant factor in my life, instructor, reference, guru.<br><br>In some ways doing this project wasn&rsquo;t that hard for me. It was mostly just applying some theory and a lot of experience. Probably the harder part was formatting the Instructable, trying to keep it interesting but informative and complete. I don&rsquo;t think I&rsquo;m ever satisfied with them but just decided to publish and hope I didn&rsquo;t make any major mistakes.<br><br>This is actually my fifth:<br> Making a bedframe with lots of storage<br> Easy Laptop stand<br> Arduino-No Blinky<br> A 67 cent postal scale<br><br>You might like Arduino-No Blinky more than most. I don&rsquo;t think it went over that well. <br>I actually was working on one called Arduino Basic Electronics but decided it was too theoretical for the more practical Instructable audience. I incorporated some of the material in the two Arduino Instructables. <br>I am currently working on one about GPS. Like you, I want to know how it works. <br><br>So good luck in your business. Let me know if I can help. I guess you can send me a Private Message.<br><br>LOG
Very cool btw, this must have taken you F-O-R-E-V-E-R! I appreciate the time that you spent to put together the grammar, organization, detailed information, and dummied down explanations. Your perspective writing is very well written and you have a keen sense of situational awareness. Very refreshing.
Thanks again. For a few years, I managed an electronics test group and had to do some instructing. More recently, I was an instructor in electronics and programming at a technical school. One of the things I learned was to know your audience and teach to them. Now the Instructable 'audience' varies all over the place. <br><br> Especially, in the Arduino Instructables, I notice a lot of authors, doing incredible, amazing projects with them and many seem to have a limited understanding of electronics and programming. Those are the ones who inspired me to do some of my Instructables. I hope to return the favor by helping them in future projects and giving them a better understanding of how it works. <br><br>Yes, these do take a while. But teaching is a learning experience. I'm always learning new stuff when I write one of these. It's good to keep the OLD brain cells active.<br><br>LOG

About This Instructable




Bio: Lazy Old Geek
More by msuzuki777:Senior Moments Weather Station 5 IP Time Clock Part 2 
Add instructable to: