Figure 18 represents how each of our force sensitive resistors (FSR's) will be used in the Wheatstone bridge to create an output voltage (Vg).
Figure 18 : Wheatstone bridge used for voltage calculations
Equations 2.1 and 2.2 give the voltage through the FSR (represented by Rg).
(2.1)
(2.2)
Combining the previous two equations allows us to solve for the current traveling through FSR.
(2.3)
(2.4)
Equations 2.3 and 2.4 were used in an Excel spreadsheet to determine the output voltage and current. Each size FSR has a maximum current that must not be exceeded. For each size FSR different resistor values were used to ensure that this current was not exceeded:
Max allowable current = 0.20 mA
R 1 = 60,000 ohms
R 2 = 2,000 ohms
R 3 = 10,000 ohms
Range of V g : 1.88V – 7.46V
Range of i g : 0.06mA – 0.15mA
Max allowable current = 1.27 mA
R 1 = 20,000 ohms
R 2 = 2,000 ohms
R 3 = 11,000 ohms
Range of V g : 0.12V – 7.50V
Range of i g : 0.08mA – 0.44mA
Max allowable current = 14.51 mA
R 1 = 8,000 ohms
R 2 = 750 ohms
R 3 = 10,000 ohms
Range of V g : 0.04V – 8.10V
Range of i g : 0.08mA – 1.09 mA
• for all calculations, a V in of 9V was used
• the range of resistances for the FSR's was 250 ohms to 100,000 ohms (as specified by the FSR manufacturer)
As discussed earlier in the report a Wheatstone bridge was used to connect the FSR to the Lab View hardware. Everyone in the group had very little experience using a circuit board so it was a challenge. The group decided to make the circuit using a solder less circuit board in case we made any mistakes while wiring it. We first looked up different wiring configurations on the internet. Doing this we built our Wheatstone bridge on the circuit board wrong (Figure 19).
Figure 19 :Original Wheatstone Bridge (Gray = resistor, Red = jumper)
We soon realized that there was something wrong with the circuit because the readings that we were getting did not make sense. The range of the resistance we were getting would get a sign wave that ranged from 2 E 6 all the way to 3 E 3 . These results were obviously not correct. When we did get a result that made sense however, we could not repeat this result. We came to the conclusion that our circuit must be wired wrong in the board. We then went to the electrical engineering department to get help; we were directed to Nishant Kumar who agreed to help us. We showed him what we wanted to do and he drew out on a piece of paper what we needed to do. Once we had the right configuration we wired the circuit board in a new configuration (Figure 20).
Figure 20 : Second Wheatstone Bridge Configuration ( Gray=resistor, Red = FSR)
At this point we knew that we had the right wiring on the board so we could start testing the sensor itself. To test the sensors we needed some known weights to place on the sensors. We got weights from engineering services ranging from 100g to 2000g. These were placed on the sensors so that we could test if we were getting the same voltage readings as were shown in the FSR user manual. We again struggled to get consistent readings. We would get resistance values that were close to what we had expected, and the next test would be a sin wave with huge amplitude. We testes all of the weights that we had and still could not figure out why our resistance values were not correct or even consistent. We decided that we needed to go back to Nishant Kumar to double check that our circuit was indeed correct.
This time we took the sensors and circuit board with us. At this meeting it was decided that the sensors needed amplification to get good readings. The engineer at the company that sold us the FSR's told us that amplification is not needed for the sensors but in the user's manual there was a set up that showed the FSR using an op-amp configuration. The op-amp circuit was added to the configuration showed in Figure 20. We decided to go ahead on try using the op-amp which made the circuit quite a bit more complicated (Figure 21).
Figure 21 : Wheatstone bridge with Op-Amp configuration
This configuration was more complicated as well as larger. We knew from the start that we were not going to be able to fit all seven circuits on one bread board if we wanted to use this configuration. The wiring on the op-amp was more complicated then that on the Wheatstone bridge so we decided to get our wiring checked before we started testing the system with the op-amp. This time when we went to the lab we decided to go ahead and test the sensors there so that if there was a wiring question we could get it answered right away rather then waiting another day to figure out what was wrong with it. After some tweaking with the wiring on the board we got it to a configuration that worked. Once we got the circuit working correctly we realized that we in fact did not need the op-amp. We tested the sensors by plugging it directly into the board and measure the voltage out of the op-amp while someone pressed on the sensor. It was clear that the sensor was plenty sensitive so we changed the op-amp to have a gain of 1, which is equivalent to it not even being there.
We finally had the circuit done correctly. In our last trip to the lab we also figure out an easier way to wire the Wheatstone bridge (Figure 22).
Figure 22 : Final Wheatstone bridge configuration ( Gray = resistor, Red = FSR)
Now that we finally had the circuit built correctly we could go ahead and start testing the sensors with some meaningful results. We were now able to push on the resistor and have the voltage rise steadily as we pressed harder on the sensor.
With the circuit working properly we could test the accuracy of the FSR. We using weights ranging from 100-2000kg we placed a weight on the sensor then observed the voltage, resistance, and the load. We were able to see right away that we had a huge problem. Both the resistance and the load calculation are based on the voltage out of the Wheatstone bridge. The voltage increases as the load on the sensor increases. The problem was that the voltage was rising too quickly. The voltage rose from zero to 8V, which is the maximum voltage output that the 9V battery gave, with only about 5 lbs. on it. This meant that the voltage of the sensor saturated at around 5 lbs, so any weight higher then that would not give a correct reading. This is a major problem if we want to measure loads on the human foot with maximum calculated loads of about 300lbs on one sensor. It was obvious that we needed to find a way for the voltage out of the circuit to rise at a slower pace.
Our first idea was to use the op-amp to reduce the signal. This was an idea that the group had but was not sure would work. An op-amp is made to amplify a voltage signal with a ratio of R1/R2, with R2 and R2 being resistors of our choice. We thought that if it is used to amplify the signal, why can't we use it to de-amplify the signal. If we made the ratio smaller then one then it would make the voltage smaller, and allow it to read higher weights. We soon realized that this idea was not going to work, we tested for 4 different combinations of resistors and got little to no reduction in the voltage readings. We now know that an op-amp is only meant to amplify the signal and using it in the opposite way, like we tried to do, does not work. Being that no one in the group is an electrical engineer we did not know that, theoretically the idea sounded good but it did not work.
The next step was to change the values of the resistors in the Wheatstone bridge. As discussed earlier in this paper the resistance values for each of the FSR's was calculated. This calculation was done using the current as the constraint; we just had to make sure that the current going into the resistor was at an acceptable level. This means that there could be many possible combinations of resistors that would work for the bridge. The first sensor that we tested was the large resistor. We originally calculated the resistors for the Wheatstone bridge as: R1 = 8,000?, R2 = 750?, R3 = 10.000?. Looking back at equation 2.4 we figured out how to lower the voltage out of the circuit. We could see that if we made R3 large and R1 and R2 small we could make the voltage smaller as the resistance of the FSR got smaller. In order to test the loading capability of the sensor we went to the schools varsity weight room. In the gym they had dumbbells ranging from 2.5lb – 95lb. We took the entire system to the gym so that we could test the sensors with a wide range of loads. After days of testing we came up with a set of resistance values that worked. The values that we used in the shoe for the large resistor are: R = 51?, R2 = 51?, and R3 = 5.1M?. With this configuration we loaded about 280 lbs onto the sensor and it gave us a reading of round 7V. This meant that the resistor could now withstand enough load for the purposes of the project.
Getting the sensor to handle the loads was a big step in the process, but we still had to make sure that our Lab View graph recorded an accurate load. This also took quite a bit of testing, because the load calculations in the program could be changed depending on the resistance readings. A more in depth explanation of the Lab View programming can be seen in that portion of the report. After quite a bit of trial and error we were able to get the software to record loads within the 25% accuracy range of the sensor. For the project the sensors will be used for quantitative measurements so it is acceptable for the sensors to be within a range rather then being precise.
After testing the large resistor we then moved on to the medium and small resistors. We realized that the programming that calculates the loads on the sensor needed to be changed for each different sized sensor. The resistance values also had to be changed for each sensor so that the sensitivity of the voltage could be adjusted accordingly. The final values for the resistors can be seen in figure 23.
Final Resistance Values
Small:
R1 = 20K?
R2 = 20K?
R3 = 750K?
Medium:
R1 = 2K?
R2 = 2K?
R3 = 750K?
Large:
R1 = 51?
R2 = 51?
R3 = 5.1M?
Even though we had all of the sensors working correctly we still had to test a few things. The first thing we tested is how the sensor reacts if a load is placed in the middle compared to if it is placed on the corner of the sensor. The sensor responded very well in this situation. If the load was on the side of the resistor, but the full load was still on the active area, it would show the same load as if it was on the middle. If the load was hanging off of the active area then the sensor showed a less, but we felt that it was about what portion of the load was on the sensor.
All of the testing up to this point was done using only one sensor at a time. Now that we have calibrated each size sensor will needed to build the entire circuit integrating all eight sensors. The first step in this process was to get a new breadboard. Throughout the testing process we were using a breadboard from one of the labs here on campus. These breadboards are old and have been used many times for different applications. This caused the breadboard to have issues with the connections. Many times in the testing process we set up the entire circuit and ran the Lab View program to begin testing, rather then getting zero volts we got a reading of 10.5V this meant that one of the connections in the circuit was bad. After discussing this problem with a few other students who have used the breadboards we found that it was a common problem among the schools boards. We decided to go to radio shack to buy a new board in order to ensure that all of the connections in the system were good. After buying the new board we went ahead and built all of the circuits in series. All this took was setting up identical Wheatstone bridges with the proper resistors to correlate with the sensor size. A 9V battery was used to power all eight bridges, through the testing we observed that this was enough power to run all seven sensors. Now that all of the electrical work is done we are ready to build the physical system. The final circuit board can be seen in Figure 24.
Figure 24 : Final Circuit Board Design