EMG and EKG works

I decided to try the instrumentation amplifier again, doing an electromyograph, rather than an electrocardiogram.  The principle is the same, but you get the signal from a muscle other than the heart.  I chose to use my left biceps, because it can be conveniently close to the breadboard to the left of my laptop.

Electrode placement for electromyograph readings from the biceps. The purple lead on the elbow goes to the virtual ground, the green and yellow leads to the instrumentation amplifier inputs.

The circuit is a very simple one, taking advantage of the off-the-shelf ina126  instrumentation amplifier:

Schematic for the EMG circuit. The op amp is used to create a virtual ground halfway between the power rails. The instrumentation amplifier is set to have a gain of 103 using the Rgain resistor. [UPDATE 2012 July 14 21:04: I noticed a bug in the schematic—all capacitors are 4.7µF not 47µF (not that it matters much).]

With my arm relaxed, I get about 40mV of noise at the amplifier—mainly 60Hz and 1.080 MHz, the frequency of the strongest local AM radio station. I can get rid of the AM radio signal by putting a 4.7 µF capacitor on the output of the instrumentation amplifier, but this does not remove the 60Hz noise.

When I tense the biceps, I get low frequency signals—spikes about 10 msec long and about 0.2v high. Given that the gain is about a hundred, the signals from the electrodes must be about 2 mV.  I ran off of batteries initially, to avoid problems with noise from the wall wart, but after I got it working, I tried a 5v wall wart (with a 100Ω resistor and 470µF capacitor, to remove the worst of the ripple from the power supply), and it worked fine also.

Here we can see a couple of pulses with the muscle tensed. The scale is 0.2V per division vertically and 10msec per division horizontally, so the two pulses in the center are about 0.12 V high and 16msec apart.

Having had this success with EMG signals, I decides to try electrocardiogram signals also (ECG or EKG). Most of the info on the web is about 12-lead EKG systems, but I want to use only 3 wires. I found a nice reference intended for nurses that describes where to place electrodes for 3-lead, 5-lead, and 12-lead systems. I tried using the MCL₁ setup in Figure 4.3

Placement of EKG electrodes for modified central lead (MCL1), copied from Figure 4.3 of An ECG Primer by Nursecom Educational Technologies (without permission).  The text says that the ground electrode can be anywhere on the body, so I left it on my left elbow.

Interestingly, the ECG Primer refers to any pair of electrodes as a “lead”, which does not match usual electrical engineering usage (in which a “lead” is a wire and corresponds to a single electrode).

With the MCL1 placement of the electrodes (in the hollow below the left shoulder and between the ribs just to the right of the sternum), I was able to get EKG signals.  The depolarization pulses are about 50mV (which means about 500 µV at the electrodes), but the background noise is still bothersome.  I added a very simple RC low-pass filter between Vout and the oscilloscope (3.9kΩ and 4.7µF), and reduced the noise to less than 5mV, while still getting 40–50mV depolarization peaks. Replacing Rgain with a 100Ω resistor increase the spikes to about 300 mV.

With a little more amplification and a slightly lower-pass filter, this should be good enough to feed into an Arduino ADC, and test out the data logging program my son has been writing.