μblog: engineering from the trenches

The basic idea with this design is that a three electrode system is used, two electrodes for the ECG and one for grounding the person being recorded from. The common mode signal from the two recording electrodes is inverted and used to drive the ground electrode with the hope of trying to mitigate the corrupting common mode signal. This grounding is typically applied far from the recording site of interest, often the right leg.

For the design, I constructed three copper coated electrodes from U.S.A. pennies and coated them with conductive paste borrowed from the electroencephalogram area. The two recording electrodes were positioned over the left portion of the chest and the right side of the torso and fed into the inputs of an INA116 (three 0p-amp instrumentation amplifier). To get the common mode signal, I split the gain resistor of the INA116 (Rg) into two equal resistors and fed the center tap voltage into a buffer. The output of that was then fed into an inverting amplifier and connected to the third reference electrode. The optimal gain of this electrode is yet to be determined, but with a few experiments, I was able to get a pretty clean signal. The final step in the equation is to do some band-pass filtering. I am thinking of doing something along the lines of 0.5 to 30Hz.

The preliminary data is shown both on the scope and verified with an optical pulse-oxymetry unit. The shape of the ECG can be distorted due to improper electrode placement and the resting heart rate is a little high for my norm, possibly due to excessive straining to get the pictures without shaking the camera. On a final note, the safety of this system is yet to be fully evaluated. The power supply is grounded to mains ground and the drive voltage of the driven reference electrode cannot exceed 5V. I am thinking of adding a series resistor and some Zener diodes to make it more safe, but that will be included in the final design. I plan to play around with this idea for a week or two and then post full schematics and possibly a data set of my own ECG when it becomes available. As always, comments are very welcome.

del.icio.us | digg

One of my annoyances with the default settings in LabView is that it places all terminals on the front panel as icons in the block diagram view. These icons are large and tend to push stuff around when they are created automatically. Every time I have to re-install LabView, I figure out how to disable this feature only to promptly forget it. This time around I will record the simple fix to save myself time in the future. The setting can be disabled by going into the Tools -> Options screen in either the front panel or block diagram. Block Diagram is then to be selected from the left side and Place front panel terminals as icons must be unchecked from the right side. Now your block diagram is safe from bulky icons, until the next LabView upgrade that is.

del.icio.us | digg

For the first time in almost 170,000 miles, the Check Engine light on my 99 Jeep Grand Cherokee came on. I try to take really good care of the vehicle so it has only had minor problems so far. Fearing for the worst, I did a few web searches and confirmed that this model Jeep (and almost all of the later models) have a feature where the trouble code can be viewed on the digital odometer. By turning the ignition key on-off-on-off-on in a short period of time (~5s)  the Jeep is persuaded to display the error code(s) in the odometer window followed by a “done”. After checking the table in the Hayes auto repair manual, I confirmed that it was the throttle position sensor that was malfunctioning. Although the sensor was in an odd place, at the back of the throttle body, the repair was pretty easy and relatively quick. The potential lesson here is that if your check engine light comes on, there may be a simple way to retrieve the diagnostic code yourself leading you to either repair the problem yourself, if it is simple, or knowing what type of problem it is before taking it to a mechanic.

del.icio.us | digg

The chip here is the INA116 from TI (Burr Brown). This is an excellent instrumentation amplifier and a true workhorse here in the lab.  It is apparent from the layout that this is a BiCMOS and it should be noted how large the die is compared to other integrated circuits. There also many more resistors on this die than some of the other analog circuits with signs of laser trimming.

del.icio.us | digg

I am participating in a summer reading course on stochastic differential equations and subsequently ran across lecture notes from Dr. Evans entitled “An introduction to stochastic differential equations“.  They give a quick introduction to statistics and Brownian motion followed by stochastic integrals including the Ito formula. Finally stochastic differential equations are introduced and their applications are given. I have only looked over the first half of this in detail and found it to be pretty reasonable. Furthermore, Dr. Evans has a larger set of available publications which include lecture notes and surveys. The semi-official book for the course is “Elementary Stochastic Caculus with Finance in View” by Mikosch (typo is reproduced from inside the front cover). A review of the book will follow later when I read more of it.

Why pasta? It reminds me of a stochastic sample set. Image was found on Musable Gourmet.

( lawrence_evans_sdes.pdf )

del.icio.us | digg

I spent most of today working out how to perform basic cyclic voltammetry and control it using a NI USB-6009 acquisition board with LabView. The basic idea is that you place two electrodes, working electrode and counter electrode, into an electrolyte and sweep the potential between them at a constant rate, typically 50mV/sec. You then track the current going into the working electrode and plot that versus the applied voltage to determine electrode/electrolyte characteristics such as the electrode’s capacity to pass charge into the solution, mass transfer effects and redox potentials. For my purposes, I just want to create a method to have an untrained individual perform a simple test on electrochemically deposited iridium-oxide electrodes to see if they are fit to be used or should be trashed.

Since the USB device only outputs 0 and 5V power and has a 2.5V reference voltage, I designed the 2.5V to be effectively ground. Triangular waveform around 2.5V is written out on an analog-output channel and is then buffered by a voltage follower and connected to the working electrode. The counter electrode is held at 2.5V by way of a current follower, where the feedback resistor is either 1kOhm or 10kOhm resulting in the two gain settings (see schematic). Both of the op-amps used have a shutdown capacity, so the electrodes are not held at any fixed potential when the VI is not running providing that the digital IO output was in a low state before. The current signal is passed through a five-pole lowpass filter at about 10Hz. The resultant current is split between positive and negative phases and is integrated to give the anodic and cathodic charge transfer capacity (relative to the working electrode) respectively. Finally, the VI displays the graphical results and has the option to write ASCII log files.

This is the first version of this VI that was coded in a day so there may be errors and this should not be used for any important research. No quality or warantee is guaranteed or implied. The source code and the executable (LabView 8.2.1 runtime engine download required for executable to run on its own) are included in the ZIP file, so please enjoy them for non-commercial use. One known bug is that the executable runs automatically when opened which is undesired behavior, the executable should be stopped and then configured and then finally started again. Another wierdness is that the CV has a dip at the very begining, at the origin, probably due to the lowpass filter. Maybe there will be a version 2 sometime soon.

( elec-cv1.pdf ) ( elec-cv1.zip )

Update: I changed two small things in the original VI, mainly the CV now runs for two cycles and it records measurements from the second one only and the two feedback resistor values have been changed to 5.6kOhm and 56kOhm (from 1kOhm and 10kOhm).

( elec-cv101.zip )

del.icio.us | digg

This week, we have images of the PIC16F84A die. This is a pretty reasonable microcontroller from Microchip that finds use in many, many electronics projects. We can see a a pair of power buses around the outside of the chip with clusters of transistors available for latching the output high or low. We can see what looks like the FLASH and the RAM on the left and right of the chip. There are also a set of ten transistors that seem to be covered with plenty of metal, perhaps they are the security fuses.

As allways, feel free to comment to request a chip to be imaged or to request more pictures of a specific part of already-imaged dies. On the same note, I have been kicking around the idea of releasing one, two and three monitor wallpaper of either analog or digital circuits, so that may come in a week or two.

( pic16f84a.pdf )

del.icio.us | digg