Lab #3: Electromyography (EMG)

The circuit we built for EKG can also be used to detect and amplify other biotentials, such as the EMG. Actually, you probably noticed that you can produce an EMG artifact on the EKG trace by contracting a muscle under the electrode. In today's lab we are going to try to eliminate the EKG and look at the EMG instead.

Recall that the function of a differential or instrumentation amplifier is to ignore or reject anything that is present on both input leads. Up till now, we have used this principle to reject the ambient 60Hz electrical noise. By repositioning the electrodes directly over a muscle we can also use the same CMR principle to reject the EKG too.

I used the tibialis anterior muscle on the shin because it usually gives a nice large and clear EMG signal, but you can try any muscle - e.g. biceps on your arm.

EMG of tibialis anterior muscle.

In order to reduce noise (both AC and EKG) to a minimum, it is best to attach the two electrodes as close together as possible. In fact you can even buy special EMG electrodes which come in pairs. The common (ground) electrode is best attached to an area where there is no muscle, e.g. over a bony prominence such as the knee cap or shin bone.MA-311 surface preamplifier

Bandstop & Bandpass filters

The frequency content of EMG is larger than the EKG: while there is very little EKG signal above about 40 Hz, the EMG goes up to 250 or even 500 Hz, so your filter cutoff will have to be adjusted upwards. Unfortunately, this means that the AC electrical noise will now not be stopped by the filter. For this reason, it is necessary to use a bandstop filter (or notch filter), which just eleimates a narrow range of frequencies:

The width of the notch obviously depends on the cutoff frequencies and orders of the filters. You can do this with purely passive components using a Twin-T circuit:

To block 60 Hz, R = 27kW and C = 0.1mF (fo = 1 / 2pRC)
In practice, it is sometimes difficult to get the component values exactly right, and it is difficult to tune this circuit, because all three resistors need to be adjusted simultaneously. The frequency of the notch of this alternative filter (from Horowitz) can be tuned with one preset, and its depth can then be tuned for maximum attenuation with the second preset:

There are three remaining limitations of this filter:

  1. the input impedance is low at high frequencies (approximately R/4)
  2. as with all passive filters, the source and load affect its properties
  3. the sharpness of the notch is limited
Active filters can be used to get round this problem, but in practice are difficult to make because the components need to be matched very closely. Because removal of AC power noise is such a popular operation, chips are available that are specially designed for this purpose. This one has a notch width of only 0.3 Hz, with 60 dB per decade rolloff.

The opposite of a bandstop filter, by the way, is called a bandpass filter, and is used whenever we want to just let one frequency through (e.g. in the EEG project).

Making the EMG signal resemble the muscle force

EMG is used primarily to indicate the timing and force of the underlying muscle contraction. In order to do this, it is important to understand the difference between the two variables, force and EMG.
 
Raw EMG is biphasic The force generated by a muscle is always contractile
EMG has a high frequency content (up to 250-500 Hz) Muscle force has a low frequency content (up to 3-5 Hz)

In order to convert the raw EMG into a signal resembling the muscle force, we need to do two things:

Detection

Detection, or rectification, means allowing only the positive half of a wave through. It is the basis of AM radio receivers, where it is used to separate out the lower frequency audio from the much higher (radio) frequency carrier wave on which it is modulated.

Modulated (high frequency) carrier wave Demodulated signal produced by detection

Detection is performed very simply by using a diode.Diode

Smoothing

The envelope of the resultant modulated wave is an analog of the modulating signal.It just needs smoothing out:

This is done by a very low pass filter. Note that this is a completely different efect from filtering before the detection, since the only frequency present at that time was the high frequency carrier wave.

For EMG, the frequency used to produce this linear envelope depends on the twitch time of the muscle. This varies a little depending on the type of muscle - in general smaller muscles (such as the muscles that move the eyeball) have faster twich times, whereas large muscles (e.g. gluteus maximus and quadriceps) have slow twitch times. For most of the muscles involved in walking, a low-pass filter of about 3 Hz has been found to produce an envelope similar in shape to the force pattern generated by the muscle.

Uses of the EMG signal 

Once transformed in this way, the EMG signal can be used not only as a means for estimating the muscle force, but also to control actuators such as prostheses. This is called myoelectric control.

Laboratory

We will use the AD620 Instrumentation Amplifier. This has several advantages:

Basic Circuit

The gain of the differential amplifier is set by RG.

The gain is a trade-off between CMRR and frequency response:

Since EMG has a very low amplitiude (about 0.5 mV), the total gain of the amplifier will need to be about 2,000. Although we could do all this with the AD620, it is probably better to add a second stage op-amp with a gain of about 20. So, the AD620 gain should be about (2,000/20) = 100.

Questions