Last week A few months ago, I wrote an article on connecting the Parallax RFID reader to the the Arduino. The reason I was revisiting that device was because I am working on a system for our hackerspace which will allow people to find out who is in the hackerspace at any given time. The overall system requires the Arduino to connect to the internet as a client and tell the Ruby on Rails application that person with RFID tag XXXXXXXXXX has walked into or left the space. The Rails application will be able to ‘publish’ this information to different mediums [web, twitter, facebook, IRC, etc] by providing a simple API for other programmers in the space to utilize.
The whole system is too complicated to explain right now and I am not finished. I plan on writing my next article about that. For now, I just wanted to document some of the issues I had with the Arduino Ethernet Shield and how I resolved them as it may be of help to others trying to figure it out.
Arduino Ethernet Shield
The ethernet shield, like the RFID reader, was not as simple to use as I hoped it would be. The information out there was decent, but nothing beyond explaining the examples. First I must explain my use cases to put context to the example code.
My goals for the code are to first send two pieces of information to the Rails application:
- A secret API key [to prevent fraud]
- The Tag code read by the RFID reader
Then I want to read the response and differentiate between four situations:
- The user has been marked as “Logged In”
- The user has been marked as “Logged Out”
- There is no user with the tag code that was sent
- The wrong API key was supplied
The ethernet shield allows your Arduino to operate as a Client or a Server. In this situation, I just want to use the client interface as I am connecting to a server to get information. Now let’s just look at the code required to make a simple HTTP GET request client with no parameters. I just modified this from the examples.
#include <Ethernet.h> // load ethernet SPI functions // a MAC address that you make up to identify the Arduino byte mac[] = { 0xDE, 0xAD, 0xBE, 0xEF, 0xFE, 0xED }; // this is the IP that will identify the Arduino byte ip[] = { 192, 168, 0, 160 }; // this is the IP of the server that we are trying to connect to byte server[] = { 192, 168, 0, 101 }; //create the client //wide area HTTP networks are on port 80 but my local rails app is on 3000 Client client(server, 3000); void setup() { //start ethernet Ethernet.begin(mac, ip); Serial.begin(9600); delay(1000); Serial.println("connecting..."); if (client.connect()) { Serial.println("connected"); //this constructs the GET request, it is the equivalent //going to the browser and entering http://192.168.0.101:3000/main/index client.println("GET /main/rfid HTTP/1.1"); client.println(); // print a newline char to tell the server we are done? } else { // this is called if conenction fails Serial.println("connection failed"); } } void loop() { if (client.available()) { // need to see if response has been read into the buffer yet char c = client.read(); //get next character and print it Serial.print(c); } if (!client.connected()) { //if we lose connection Serial.println(); Serial.println("disconnecting."); client.stop(); // disconnects from the server for(;;) ; } }
Some of this warrants a little more explanation for some…
Configuring the ethernet shield
First is the MAC address.
byte mac[] = { 0xDE, 0xAD, 0xBE, 0xEF, 0xFE, 0xED };
For those unfamiliar with basic networks, a MAC address is a unique number assigned to your network card that never changes. You don’t need to know much about it other than you need to make it up. You should be good if you use the one I have just used in this example but keep in mind that if you have multiple Arduinos connected to the same network, you may need to make up unique MAC addresses for each one.
Next is your IP.
byte ip[] = { 192, 168, 0, 160 };
This array refers to the IP that you want your Arduino to take. The Arduino Ethernet Shield currently ?doesn’t support DHCP?, so you have to choose an unused IP yourself. There are a variety of ways and steps to do this. First what you need to do is figure out what the first three bytes in this array need to be. This is determined by the local router you are connecting to. For instance, Linksys is usually 192.168.1.XXX and D-Link, in my case, is usually 192.168.0.XXX. You can find out what your situation is by checking the IP of a computer on that network. Then you need to choose an unused number for the last byte, usually between 100 and 255. If you are on a network with few computers, around 150 is usually a safe bet. If you are not sure or have a lot of computers on the network, you should find a way to log into your router’s admin interface and check the DHCP information. It will tell you all the IP addresses that are currently assigned and you can make one up an unused one based off that.
Last is the server’s IP.
byte server[] = { 192, 168, 0, 101 };
This is the IP of the server that you are trying to connect to. In my case, it is a laptop running my Rails application connected to the same network. If you want to connect to a server outside your network, on the internet, you need to find it’s IP. Your computer usually does this under the surface using DNS but you will have to do it manually here. One way you can do this is by pinging the domain name and seeing what IP it resolves to.
Sending information to a server
So, this example works out for many situations, but how can I send information to the server where my Rails app is being run? In my case, I need to send the API key and and the RFID tag I just read. What we need to do is build a query string. If you aren’t sure what that means, you will have to read that wikipedia article carefully. Let’s reexamine the code:
//I changed this line client.println("GET /main/rfid HTTP/1.1"); //to this line client.println("GET /main/rfid?apikey=123&tag=0123456789 HTTP/1.1");
When I run this code, I see this on the serial port:
connecting…
connected
HTTP/1.1 200 OK
Connection: close
Date: Mon, 26 Oct 2009 01:12:43 GMT
ETag: “8da44df37e592a5020e852d50fc31a64″
X-Runtime: 8
Cache-Control: private, max-age=0, must-revalidate
Content-Type: text/html; charset=utf-8
Content-Length: 8^=NOUSER
This is the HTTP response. The last bit:
^=NOUSER
Is what I have defined in the Rails application to tell the Arduino what has happened. Just for clarification, this is what the rfid function looks like in the Rails app:
#login via rfid service, has no view def rfid if params[:apikey] == $api_key @user = User.find_by_rfid_tag(params[:tag]) if @user @user.in_space = !@user.in_space @user.save if @user.in_space render :text => "^=IN" else render :text => "^=OUT" end else render :text => "^=NOUSER" end else render :text => "^=KEYFAILED" end end
I tested a few different query strings and made sure it worked for all conditions and it did. Now I need a way to dynamically send the tag that we read from the reader. As you might have read in my RFID post, I am storing the RFID tag in a 10 character array:
#define CODE_LEN 10 char tag[CODE_LEN];
I initially thought I could do this:
client.print("GET /main/rfid?apikey=123&tag="); client.print(tag); client.println(" HTTP/1.1"); client.println();
Notice that I am using a series of print()s to build a string and ending it with println() to close it out with a newline. The only problem is that this doesn’t work. My Mongrel server (the Rails server) was telling me that my HTTP request were invalid. It seems like a lot of other people were running into this problem as well. After temporarily giving up, I realized the problem. I was reading an essay by Brian Kernighan, one of the creators of the C programming language, about a simple regular expression parser that he uses to teach students about programming methodologies. Scanning through the code, I was reminded that a character array cannot be interpreted as a string without a terminal null character ‘\0′ as the last element. So, I handled this situation like this:
#define CODE_LEN 10 char tag[CODE_LEN + 1]; tag[CODE_LEN] = '\0';
This didn’t affect any of my other code and now the tag array can be interpreted as a legit string 
I love simple solutions.
Now I need a way to parse the response. The reason my responses had this format ‘^=XXXXXXX’ was so I could do something like this:
/** * Finds result that we are looking for in the returned HTTP response */ char getHTTPResult() { while (client.read() != '^') {} if (client.read() == '=') { return client.read(); } return 'x'; }
This would return the first character for the response I am looking for. For example, with the NOUSER message, I would get back the ‘N’ character. I implemented this but the problem is that it was really slow. Or at least, it was a lot slower than it could be. The problem is the amount of pointless HTTP data you have to read before you get to the actual message, or it could be that I am printing all that data? Doesn’t matter because I can make it faster. I could have configured my server to not spit out so much junk, or I could just place the message earlier in the response.
Reverse REST
The first thing you return is the status code so why not put it there? As I said before, there are only four states I am trying to identify and there are plenty of HTTP status codes to represent them. So I modified the Rails controller to look like this:
#login via rfid service, has no view def rfid if params[:apikey] == $api_key @user = User.find_by_rfid_tag(params[:tag]) if @user @user.in_space = !@user.in_space @user.save if @user.in_space render :status => 400 #IN else render :status => 401 #OUT end else render :status => 402 #NOUSER end else render :status => 403 #FAILED end end
I can’t help but wanting to call this ‘reverse REST’, LOL. Rerunning my simple “no user” test, I now get this:
HTTP/1.1 402 Payment Required
Connection: close
Date: Sat, 06 Feb 2010 02:45:19 GMT
X-Runtime: 12
Content-Type: text/html; charset=utf-8
Cache-Control: no-cache
Content-Length: 0
This is much quicker to parse. I probably could have used some nice code to make this more flexible, but just hardcoding the digit read process in is probably easier and just as reliable, if not more efficient:
char getHTTPResult() { while (client.read() != '4') {} if (client.read() == '0') { //just to make sure return client.read(); // return either '0' or '1' or '2' or '3' } return 'x'; // something bad happened }
At this point a simple switch allows us to define the behavior of the response:
void sendToServer() { client.flush(); //just to be sure if (client.connect()) { client.print("GET /main/rfid?apikey=123&tag="); client.print(tag); client.println(" HTTP/1.0"); client.println(); Serial.println(); switch (getHTTPResult()) { case '0': Serial.println("logged in"); break; case '1': Serial.println("logged out"); break; case '2': Serial.println("no user"); break; case '3': Serial.println("API key failed"); break; default: Serial.println("broke"); break; } client.flush(); //just to be sure, probably not needed here client.stop(); //should stop it } else { Serial.println("connection failed"); } }
A few more troubleshooting tips (will keep updated as they come in)
- Be sure to use flush() when appropriate. This ensures that your input buffer is clean and that you don’t have any left over remnants.
- I noticed that sometimes the ethernet shield would not power up when I plugged in my Arduino. You can tell by the group of lights on the top. I could only get it to start up when I had unplugged anything that was leaching power from my board (in my case the RFID reader which is leaching current off the 5V pin). I am not 100% sure why this is but my guess is that it has something to do with some components on the ethernet shield not getting enough current to start it up. If this happens to you, you will often find that client.connect() will return false every time. I will try to investigate this further.
Conclusion
Hopefully this isn’t too far out of context for your applications. Feel free to post questions about this device and I will try to help out.
Oh well.
DSO nano and synth circuit
A few days ago, a co-worker let me borrow his DSO nano to run a few tests. First thing I decided to do was throw a really chaotic waveform at it. I built this little synth with 2 square wave oscillators.
It is kind of comical to call this a synth or even a set of oscillators. It is basically a simple NOT gate arranged in a feedback loop. The capacitor fills up until it reaches the breakdown voltage of the internal diode in the NOT gate, at that point, the signal is inverted to GND and the capacitor discharges. When it reaches ground, the output signal inverts again and the process starts over. The result is a square wave. The potentiometer controls the amount of current flowing into the capacitor and thus alters the amount of time it takes to fill up. In effect, altering the frequency of the square wave. I built two of these and muxed the signals together with some resistors, kind of like a fixed mixer. The overall signal is pretty hectic, with weird beat frequencies and strange tiny oscillations in the individual frequencies themselves.
I found it pretty hard to get the DSO nano to get a good read on it but it was pretty hard with my regular scope as well. Probably not the best waveform to start with. So, I decided to try out some simpler waveforms. I programmed an Arduino to just spit out ‘a’ on the UART every second and used a rising slope trigger to freeze the waveform. It worked really well for that. It also worked really well for just a square wave generated on the Arduino. It was able to guess the frequency with accuracy. Unfortunately my phone had died at that point and I couldn’t get any pictures or video.
Overall I found it pretty simple to use and a surprising amount of functionality for it’s small size and price. I have to admit though, I wouldn’t find much use for it as I am almost always in front of a desk where I can use my USB scope and software. I think this would be more helpful for someone who works in ‘the field’ so to speak. It is pretty much as good as you are going to get in a pocket-sized package.