The main goal of our Biomedical Signal Processing Project is to design and implement an ECG amplifier from scratch, acquire an amplified and clean biosignal, sample this signal (i.e. digitalize it), and finally process it (i.e. filter, find patterns, provide diagnosis) using the low-cost OMAP L-138 Digital Signal Processor from Texas Instruments.
This is actually the big picture of this project and we are working to provide a series of tutorials to make these goals easily available for free! We wish to be able to provide full support to any student that is interested in the field to start conducting his/her own Biomedical Signal Processing and Analysis!
This is the first of a series of tutorials that will walk you through the design, implementation, and testing of our hardware and software solutions. Please enjoy it and feel free to post any suggestions in our comment box in the end of each post.
Now let’s talk about the actual science and design behind this ECG amplifier. First of all, a biosignal is characterized by high common mode signal, low amplitude differential signal, low frequency noise, drift, and cycle to cycle variability. All of these variables increase the difficulty to process these kind of signals using the average digital signal processor. So before starting with the filtering and actual digital signal processing of these signals, we have to step back and build an amplifier that will:
- amplify the the signal’s amplitude; and
- “clean” the signal
After having a clean and amplified signal, the next stage of this project will be to sample this signal — turning it to digital samples — and process it using our Texas Instruments OMAP L-138 Digital Signal Processor.
This is our next important milestone and shall be further discussed in a later post. Now, all we are interested about in on the ECG design!
We started with the design present in the Tompkins book (please refer to the bibliography in the end of this post).
(figure 2: ECG design from Tompkins book)
We assembled that design, which consists of 4 operational amplifiers, capacitors and resistors, and then changed the original design to satisfy our needs. The operational amplifiers used were the uA741, which are good for this application and are very low-cost chips.
Our final design was as follows:
The circuit above was first specified and tested using a simple sinusoid signal. As input, we had a 36.0 mV peak-to-peak sinusoid and the output of our ECG amplifier was a 7.20 V peak-to-peak sinusoid signal. This means that our ECG had a final gain of 200, and now we have a signal with a greater amplitude — meaning that we are one step closer to digitalize and process such signal.
In contrast, we had a very low amplitude common-mode signal gain — close to 0 V. This is desired in order to make sure our ECG circuit is working properly. In more detail, when a 60 Hz signal is the input of both buffers seen in figure 3, the output from our ECG circuit is an approximately constant wave form of 220 mV amplitude, or in other words, the circuit has no Common-mode Gain.
1. Tompkins, W.J. and Webster, J.G., 1981, Design of Microcomputer-Based Medical Instrumentation, Prentice Hall, Englewood Cliffs, NJ, 496p.