ChainFire's Blog

Good news! Joe (earmuffs) and I had a meeting with the Director of Dining Services this morning. The goal of the meeting was to demonstrate our project and answer any questions. We presented our final hardware we created on february_2nd_2010. We explained how the IR LEDs and IR Receivers work to form the tripwire. And that even the bigger scale model would only involve longer wires but the model would not grow too much more in size. We reassured him that the wires would be out of the way of students, as safety is a main concern. The director looked at the benefits this project to the students as well as the Dining Services and approved our project to collect data at the dining halls.

We will start collecting the data at Russell Dining Hall. This hall has one exit and one entrance so the project does not need to be expanded too much more then its current state.

Implementation Plan:

• Initial test of two hours at Russel Hall.
• With a successful two hour test, we would collect data over a period of several weeks at Russel Dining Hall and publish it to the website.
• Then move towards collecting data at the other dining halls and publishing that data to the website.

We also agreed to give all of our data to the Dining Service Director so that they could use it to improve their customer services for their students. This project will be very beneficial and I am thrilled to be a part of it.

I discussed problems with the hardware in my last blog entry. The general hardware problem discussed in my last blog (january_31st_2010) entry was that we could not get our IR receivers to detect the IR emitters (no tripwire). We built the circuit several times and measured all the outputs which were all accurate… so we decided to wait until today to make further progress.

We found the problem this morning and fixed our hardware! In our attempt to increase the current going through our IR LED's we implemented a Driver chip which cycles between 5V (vcc), PWM and Ground to create a stronger signal. When we measured the voltage this morning with the oscilloscope we noticed that there was a lot of noise on the line. When we measured the frequency of this noise we realized that it was exactly 56.1kHz. Which means somehow our PWM was getting tied into our Vcc. This creates the problem because when the IR receiver receives the frequency through the Vcc it considers the wire as untripped. Even if we remove the IR LEDs or cut the signal the IR receiver would get the signal from the Vcc and consider the wire to be not tripped. We corrected this problem by redoing the wiring.

The hardware we have finalized puts out approximately 55mA-60mA to each IR LED. With stronger IR LED signal we are able to create the tripwire over a longer distance. The 60mA of current provides us with a distance of approximately two feet. The IR LED we are using has a max current rating of 200mA (IR LED datasheet) We have bumped the current to 170mA which should provide us with a distance of approximately two meters, which is enough to cover the length of the door.

The IR LEDs are harder to see in this design, they are located directly across from the black receivers. We cut the extended wires because it was difficult to keep the IR LED steady which made it a pain to keep them focused on the IR Receiver.

The schematic for the hardware can be found in the following PDF: IR Tripwire Schematic. The first page is the schematic of our current hardware which puts out approximately 55mA-60mA of current through each IR LED. The second page is the schematic of our expanded hardware which will put out approximately 170mA-180mA to each IR LED.

The code for the project was created and finalized yesterday. The code was done with broken hardware (discussed earlier in this blog). However we avoided IR receivers/IR emitters altogether by simulating wire the wire trips by connecting the input port to High (Vcc + resistor). The idea for connecting input ports to high was all earmuffs! The code was a success and keeps an accurate count of total people “inside”. The full commented code can be found on Earmuff's latest blog entry counting_code. A video of the complete small scale model can be viewed: Count Video

I have emailed the dining services department tonight to schedule a demo of our project. If they give us an approval we can expand our hardware and implement the project for the benefit of the University and for the benefit of the Students.

The hardware I discussed on january_22nd_2010 was problematic. We measured the current going into the IR LED's and the measurement was roughly 3-4mA. Which leads to us not getting over a 1.5”-2” range.

To increase the current going into the IR LED we used a ULN2001 (driver) chip. We plugged the PWM signal into the input section. The other side connects 5 volts to the IR LED through two 56ohm resistors in parallel to the output pins of the driver chip. The driver chip switches between the high current, PWM signal and ground. When this is done the result is the same 56.1k signal with more current. The new hardware picture can be found earmuff's blog.

The new hardware was tested today and the frequency remained the same through the driver chip. However the detection we had previously was lost. As of now we have unplugged the new parts added and reverted back to hardware discussed on January 22nd.

The code is getting to be frustrating. We are unsure of how to tie the three tripwires together. As of now the way we have the code the 1st tripwire counts up while the 3rd tripwire counts down. The three tripwires are connected by if statements and they should not count up or down by just tripping one of them.

We got an email response back from the Dining Services regarding our project being placed for actual testing. As of now they have denied us access for the purposes of their customer service until we demonstrate our project to them. Our goal this week has been to finish the small model of the project.

We are so very close to finishing this project. Over the last few days our goal has been to create the code to detect the direction of the person entering as well as keep an accurate count of the people inside the dining hall. This has been a very tricky process.

Problems: The micro controller polls the detection so quickly that even if we slide a small wire to engage the tripwire, the count goes all the way to twenty-two thousand. Also, when we tested the decrement portion the count goes all the way down to negative one thousand.

Solution: We will have to implement edge detection into our code. This would make it so the micro controller only detects the change in the signal from high to low and count that as a trip verses counting the entire duration of the trip. This will give us an actual real time count of a person walking in and out.

We ventured to Kent dining hall yesterday. We thought to ask for permission from the managers and make sure that taking pictures of the exit/entrance however it lead to discovering that we have to go through at least two other people before we can get pictures of the hall. We have their contact information. They have both been emailed about our project details and we requested permission to carry out our project.

The initial testing will still be done in a controlled environment, perhaps an empty hall way at night, just to make sure our equipment is working as intended. Also it would be much easier to fix any problems close to the lab vs running back and forth from the dining hall to the lab every time some small error occurs.

Today we got further on the hardware. We cleaned it up so the wires are not running everywhere and also expanded it to include three trip wires and not just one. The normal LED's function properly; when they IR receiver receives the signal the LED is off, when the signal is blocked from the receiver the receiver is turned off. This is our step in the right direction. For now we have detection. Tomorrow we take it one step further and code the direction of the detection and the accurate count keeping.

Before:

After:

Video: IR Tripwire