Archive | June 2012

Strange signal

I’ve been making good progress on tracking down and eliminating unwanted interference like this… we’ll see if my current solutions (mostly shielding) hold up longer term.

Here’s a signal the EMG picked up that I haven’t quite been able to figure out… note how much faster it is than a normal muscle impulse.



Simple High-Performance DIY EMG based on the Zoom H4N

What is an EMG?
Like an EEG and an EKG, it measures the electrical impulses that make the human body work. Whereas an EEG measures the brain and an EKG the heart, an EMG looks at electrical activity over the rest of the body — which pretty much means the muscles.
To get your muscles to contract, the nervous system sends electrical pulses into the muscle fibers, which amplify those pulses and send them on to other fibers until the whole muscle is moving.
Here’s what a normal EMG might look like:
What’s it good for?
Recording muscle activity! In my case, I wasn’t sleeping so well, and I wanted to figure out whether I was tossing and turning in bed as much as I thought I was. An EMG provides the means to figure out just when and how that was happening.
But people use EMGs for all sorts of physical therapy and sports applications.
The heart of the DIY EMG is a Zoom H4N audio recorder hooked up to three electrodes — two measurement electrodes and a common ground. The resulting WAV files (I can get over seven hours of recording on a 16GB SD card, just about enough to cover a whole night) are the EMG output, easy to analyze in any audio editor (I use Audacity).
The Zoom recorder is recording at the highest quality setting, 24bit/96kHz. This gives it a maximum bandwidth of almost 50kHz and, at least in theory, a dynamic range of 145dB. In other words, perfect for recording voltage spikes that might be very quick while ranging from tiny to huge.
The Electrodes
I went through four designs for the measurement electrodes. Most of them weren’t capable of picking up the EMG waveforms completely: the inductance and capacitance were too high. I settled on 5cm x 1cm strips of self-adhesive aluminum foil for the contact material. They’re stuck to the fanned-out copper strands from the center conductor of of RG174 coaxial cable… backed with black duct tape, of course.
Here’s what the build looked like:
(adding foil)
The other end of the RG174 is soldered to the center conductors of some standard 1/4″ mono plugs. The RG174’s shield braid is connected the plugs’ grounds, this helps reduce the amount of interference picked up by this very sensitive setup. (By using the Zoom H4N’s 1/4″ inputs, we get an input impedance of 500,000 ohms. We need a really high input impedance because even on a good day the electrodes don’t form a particularly good connection: on the order of 500k ohms as well.)
Each electrode then goes on the area I want to take readings from. You have to be quite close to the muscle to get readings from it, the signals don’t travel very far. If you’re far away, the spikes get ‘mushy’. Instead of sharp peaks, you see rounded humps. (The closer you are to the muscle, the sharper the peaks.)
For the ground electrode, I’m using a similar but much larger square (5x5cm) connected with a length of standard electronic hookup wire. I place it far away from the measurement electrodes to provide a good reference. The ground electrode gets connected to either of the two 1/4″ connectors’ grounds.
All the electrodes are held in place with standard medical tape. A dab of sugar or glucose syrup (available at your local pharmacy, corn syrup would also work) on the aluminum foil helps make a good electrical connection. And, after a few minutes, it’s sticky enough to help hold the electrodes in place.
The Results
Impressive! The Zoom captures a lot more detail than a normal clinical EMG:
Here’s zooming in a little on the waveforms… You can see the spike in the middle there occupies less than 1/290th of a second. Compared to a normal EMG that samples 50 times per second, that means the Zoom is capturing things you normally wouldn’t see.
Not bad for a few hundred bucks from an audio recorder and less than ten for the electrodes.