The second project undertaken for Jeff’s research was a 16-channel Electromyography Amplifier. At the time, Jeff was working on his PhD thesis at Queen’s University, in the lab of Dr. R. Mel Robertson.
Electromyography is recording the electrical signals that activate muscles. Jeff was interested in recording the the activity of the wing muscles of a locust in tethered flight. Many wing muscles are involved, and he wanted to record them all simultaneously. The lab had a number of Grass Model P15 amplifiers that were used for this type of work, but each one has only a single channel, and they were in short supply.
EMG/EEG/ECG are among the major applications of instrumentation amplifiers. The enormous input impedance and common-mode rejection ratio (CMRR) of an instrumentation amplifier allows them to extract usable signals from the poorest, noisiest, weakest sources. We had previously used an Analog Devices AD620 instrumentation amplifier chip for the Position Transducer project. We also used it for the Myoelectric Locust hack, an application which was basically just EMG. On that project, we learned a number of useful things about how to do EMG, which led directly into this project.
The most important lesson we learned that fed into this project was that a really large input impedance can cause its own problems. On M.E.L., we found that the circuit worked fine for a while, but after a time the output of the amplifier drifted all the way to the power rail and stayed there. The reason turned out to be that the huge input impedance of the AD620 tended to collect charge, until eventually the DC offset exceeded the range of the amplifier. The problem was solved by adding very large resistors (22MOhm) from each input to ground, to drain off the collected charge.
The amplifier is intended to be used inside a Faraday cage, to reduce the amount of 60Hz hum picked up. I made a separate +/- 12V DC power supply, which would remain outside the Faraday cage. I tried to filter the output as well as I could, to prevent 60Hz hum from getting into the cage. A shielded cable delivered the power to the amplifier chassis inside the cage. I ran all the output signals on a single cable to a rack-mountable BNC terminal panel outside the cage.
Each channel of the amplifier provides three switch-selectable gains (x10, x100 and x1000), and low-pass and high-pass filtering. To keep costs down, the corner frequencies of the low-pass and high-pass filters are not easily changed. They are determined by capacitors installed inside the unit on a screw-terminal strip. In addition, the amplifier has a monitor channel. Any or all of the channels can be mixed onto the monitor bus, which is further amplified and played through a speaker. The monitor allows an audible indication of the quality of the electrode implantation. Finally, the amplifier provides an accurate calibration voltage pulse of either 50 or 100 uV (derived from a precision voltage reference chip), which can be sent to any input to calibrate voltage levels accurately.
For electrodes, Jeff used very thin “magnet wire”. Thin as a hair, and with a lacquer insulation. To implant the electrodes, Jeff put the locusts in a refrigerator for a while to put them sleep. Then he would poke a hole in the exoskeleton just outside the muscle. The magnet wire was then inserted through the hole to the appropriate depth to reach the muscle of interest. He did not strip off the lacquer insulation, so that the only point that made electrical contact was the tip of the wire. The electrode would then be secured in place with a drop of melted wax.
Here is a picture showing EMG signals Jeff captured from 12 wing muscles simultaneously. Six muscles on the left, and the same six muscles on the right. The region labelled “stimulus” is where a simulation of a bat echo-location sound was played, to trigger the locust’s “evasive manoeuvers”. At that point, you can (barely) see that the muscle enervation pulses start double or triple firing. In the magnified region on the right, you can see the left and right sides become slightly asymmetric, which has the effect of causing a turn (or it would if the locust wasn’t glued to a stick and flying in a wind tunnel.) These results were published in a paper, J.W. Dawson, F.H. Leung, R.M. Roberson (2004) Acoustic startle/escape reactions in tethered flying locusts: motor patterns and wing kinematics underlying intentional steering. Journal of Comparative Physiology A (2004) 190: 581-600.
I wrote this user’s guide for the amplifier: EM-16 User’s Guide
The circuit board was fabricated using a postive photographic etching process. The photomask was printed on laser-printer-compatible transparency film at 1:1 scale. A photomask produced this way is not sufficiently opaque to produce good PCB results. If you hold it up to a bright light, you can see light through the toner. I found a decorative special-effect foil product that can be used with laser-printers to make metallic imagery. Normally, it’s used only to add gold or silver lettering at specific places on the page. I used it to put a metallic foil coating over the entire photomask.
The way the film normally works is you print out your page normally. Then you cut out small pieces of the metallic effect foil, and secure them over the specific parts of the page where you want the effect, using small adhesive dots that come with the film. Then you run the page through the printer again, printing out a blank image. The idea is to just run the page through the fuser of the laser-printer. The heat of the fuser will bond the metallic film onto the preexisting toner on the paper. Then you can peel off the plastic carrier, leaving the metallic effect on the paper.
When using the metallic film to make a photomask, I use the entire sheet at once. I found it tended to jam up in a laser-printer, so I actually use a laminator to fuse the film onto the toner. It took some experimentation to get the laminator feed rate set properly. I put the transparency photomask and metallic film together through the laminator. When I peel the metallic film carrier, the photomask has been covered with metallic flim, and is almost perfectly opaque.
Unfortunately, I find that with the transparency, a lot of the metallic film sticks to the open areas where there is no toner. So, I have to do a fair bit of cleaning up with an X-acto knife to scrape off metal where it shouldn’t be. In the end, I get a good quality positive photomask that can be used to fabricate many PCBs.
Of course these days, low-volume quick turnaround PCB shops are increasingly common, so my days of fabricating PCBs at home may be over. Good riddance, I say.
Front Panel Construction
Making nice professional-looking front panels for my electronics has been an obsession for some time. I still do not have a technique that gives me satisfactory results at a reasonable cost. I’ve tried many things along the way.
For this project, I used a homemade decal. I wanted white lettering. I approximated white, using the same silver effect foil that I used for PCB fabrication.