Article Preview
TopSystem Design
The aim of this design is a low cost microcontroller based EMG system, which can record and store EMG signals. The system itself will consist of two main parts, hardware and software. The hardware part will be responsible to detect, digitize, and transmit the muscle signal to computer. The software part will be responsible for recording and displaying the digital data. We divide the prototype requirements into three groups. They are usage requirements, hardware requirements (Day), (De Lucca, 2003), and software requirements (Van Bemmel & Musen, 1997). Example of usage requirements are portability and safety. Hardware requirements are derived from the characteristic of muscle signal, it frequency, amplitudes, and signal noises that may detected during measurement.
Figure 1 is the simplified block diagram of the system. The two blocks, hardware and PC is connected by using serial RS232 protocol.
Figure 1. System’s simplified diagram block
The hardware part consists of two sub-parts: analog and digital. The signal amplification is done by using cascaded amplifier, with instrumentation op-amp as the first amplifier (Merletti & Hermens, 2004). We choose MAX4194 for this purpose, because it enable us to further uses the common signal as a feedback to reduce noise signal that may interfere during measurement (Mancini, 2002). This step is simply done by reading the signal from the from between two resistors that control MAX4194 gain. If we use MAX4196 or MAX4195, these resistors are located inside the IC. For the next stage amplification we choose TL072 op-amps for all op-amps other than instrumentation op-amps. We set the gain of instrumentation amplifier to 26. This includes the op-amps used in the filter. We set the filter so it will have unity gain (Bramble, 2002), although during testing, the gain is not as uniform as we expected.
Signal filtering will remove unwanted frequency. In this design we develop a band pass filter with the 4th order Butterworth-Salley Key topology. The band pass frequency is 20Hz to 400Hz. These setups are repeated for each channel. We decided not to put a 50Hz notch filter in these setups because 50Hz frequency is considered as one of the frequency where muscle signal can be strongly detected.
Since in this design we include 4 channels requirement, we need to further average the common signal from these 4 channels before we can use it as a feedback (Ott, 1976) to minimize noise. This technique is shown by Siriprayoonsak (2005). However we found that in our system, we are unable to use electrodes as the electrode for this feedback signal, so we use aluminum foil instead (Figure 2).
Figure 2. Bode plot from one of the channel