Weekly Update Michael Jorgensen Week of 2/25/11 3/3/11

Transcription

1 Weekly Update Michael Jorgensen Week of 2/25/11 3/3/11 I tested the circuit with my left masseter and temporalis muscles. I obtained a clean signal as shown in FIG. 1 (L masseter). It is evident that the gain of the circuit does not have to be adjusted. Figure 1. Left masseter signal tested on 2/25/2011 I also added some signal processing to the program, including RMS voltage and rectified. I have found in the literature that these signals are clinically useful. The rectified signal is shown in FIG. 2 and the RMS signal is shown in FIG. 3 each showing three clenches (test done by chewing gum). I contacted our client, Dr. Litt, to determine if he would like these signal processing tools to be added to the program. They will add more usability to the program, but has a tradeoff of additional complexity. Figure 2. Left masseter rectified signal Figure 3. Left masseter RMS signal I noticed some relatively small ECG artifact in the signal. It was noticeable in the unfiltered EMG signal and somewhat noticeable in the filtered EMG signal. It may be of interest to increase the lower cutoff frequency of the digital filter in LabVIEW to remove this artifact further. I placed a purchase order for a microcontroller for the purpose of multiplexing the four EMG signals into one output and for analog to digital conversion. I decided upon the ATmega644PA 8-bit MCU. I am currently using this MCU for my microprocessor applications lab, so I am familiar with the interface and the software components. In addition to the MCU, I ordered two crystal oscillators and a

2 tactile switch for the RESET interrupt. This MCU has everything that we need to interface with the Bluetooth module. I also configured the Bluetooth circuit. Since we do not have the laptop, we cannot transmit data. Thus, I just configured the power portion of the circuit. I used a 1N4001 diode to protect the Vcc pin of the module. The diode is connected to the 5 V output of the LM7805 voltage regulator, thus with a 0.7 V voltage drop, the Bluetooth module is powered by +4.3 V, which is a sufficient power supply for operation. An LED emits on the module, indicating proper connection. I observed that the current consumption for the bandpass filters was significantly high. Each muscle required a cascaded HPF and LPF to form a BPF. This created an excellent filter with a steep stop band; however, each muscle required one quad operational amplifier (TL074). Each quad op amp requires 1.4 ma supply current per amplifier. Thus, each muscle requires 4 amplifiers * 1.4 ma = 5.6 ma. For four muscles, the total current consumption for filtration is 5.6 ma * 4 muscles = 22.4 ma. This is a significant amount of current consumption when dealing with portable battery power. I decided to reduce the number of amplifiers for the bandpass filters to one. This results in a second order filter with relatively good characteristics. The bandpass filter design is shown in FIG. 4 and its characteristics are shown in FIG. 5. C2 1nF R2 330kΩ VDD -9V V2 120 Vrms 60 Hz 0 R1 82kΩ C1 0.1uF U1A 1 TL074ACD D5 1N4001GP Vout VCC 9V Figure 4. New bandpass filter design

3 Figure 5. New bandpass filter characteristics This design reduces the total current consumption for bandpass filtration by 75%, preserving the battery life of the two 9 V batteries. It also saves space on the PCB by reducing the number of ICs from four to one. Additionally, it reduces external resistor and capacitor count, thereby reducing the thermal noise attributed to each. I tested the new bandpass filter with an EMG signal and it exhibited an improved signal quality over the cascaded LPF and HPF. The new bandpass filter has a voltage gain of approximately 4, thus it further amplifies the EMG signal. The results of the EMG test are shown in FIG. 6. Figure 6. EMG test results for new bandpass filter

4 I started to design the PCB schematic in Multisim by using off-page connectors to connect different pages. I also created a general layout in Visio with accurate dimensions so that we may get an idea of how big our enclosure should be. This will be a big help to Kerry so that she may better design the hat with dimensions in mind. The general layout is shown in FIG. 7. The dimensions of the enclosure are 3 in x 4 in. The bottom of the enclosure will have a curved plastic design so that it may better conform to the patient s head. Figure 7. General PCB layout I also tested the reusable electrodes versus the surface electrodes. They work pretty well, and surprisingly, no electrolyte gel gave a better signal than with it, and they give almost the same results as the surface electrodes. There are some problems, however. They are difficult to keep attached to the skin, so it may be difficult to incorporate them into the hat and get them all to adhere. Also, they let in a lot of 60 Hz noise, but we can probably filter that out in LabVIEW. I'm going to run more tests on them on Friday so that we can get more data and show it to Dr. Litt to get his opinion. FIG. 8 shows a test with surface electrodes, FIG. 9 shows a test with reusable electrodes without electrolyte gel, and FIG. 10 shows a test with reusable electrodes with electrolyte gel.

5 Figure 8. EMG test surface electrodes

6 Figure 9. EMG test reusable electrodes no electrolyte gel

7 Figure 10. EMG test reusable electrodes electrolyte gel