Pflfigers Archiv

Pfliigers Arch (1982) 395:251 - 252

EuropeanJournal of Physiology 9 Springer-Verlag1982

Instruments and Techniques

Digital Reproduction of Biopotential Waveforms for Neurophysiological Studies Sanjeev D. Nandedkar, Frank W. Ingle, Donald B. Sanders, and Yong I. Kim Division of Biomedical Engineering and Department of Neurology, University of Virginia School of Medicine, and the University of Virginia Jerry Lewis Neuromuscular Center, Charlottesville, Virginia 22908, USA

Abstract. A simple biological signal generator capable of reproducing complex biopotential waveforms is described. It is constructed by a combination of digital and analog circuit components and can be used under different experimental conditions, such as in calibration of biomedical instrumentation systems, or simply as a function generator providing voltage outputs of various waveforms. The biopotential waveform to be generated is sampled at a high frequency and the samples are stored sequentially in a programmable read only memory (PROM). The samples are then fed in the same sequence to a digital-to-analog (D/A) converter and the resulting output is amplified and a DC offset is added. External controls are provided to adjust the DC offset, amplitude and repetition rate of the signal generated. The reproduced voltage signals are stable and superior in quality to those produced by conventional biological signal generators. Key words: Biopotentials -- Biological signal generator Calibration Miniature end-plate potentials Programmable read only memory

Introduction Biological signal generators can be used for various purposes in the area of basic and clinical neurophysiology. In addition to generation of standard voltage outputs such as step, sinusoidal, square or triangular waves (Miliman and Halkias 1972; Sabah and Adbam 1974), modern biological signal generators are often required to produce complex irregular voltage waveforms similar to true bioelectric events. For instance, computer-based data acquisition systems often need calibration for accurate measurements. Normally a fixed step voltage, or a square or sinusoidal wave is used, these being quite different, in terms of frequency components, from the biological signal to be measured by the instrumentation system or the computer. Hence, a biological signal generator that can generate signals that are very similar to the signal under study should provide a superior calibration for a data acquisition system. Biological signals of complex and irregular shape can be generated electronically using a series of monostable multivibrators. Blazek et al. (1975) have described a simple function generator that uses a piecewise approximation for simulation. A sequence of pulses of different amplitude and duration is generated using an N-phase astable multivibrator Offprint requests to Y. I. Kim at the above address. D. B. Sanders' current

address is Division of Neurology, Duke University Medical Center, Durham, North Carolina 27710, USA

(Pratapareddy and Rajappan 1973), and the pulses are summed and integrated to produce the required analog signal. A major drawback of such a generator is that when the number of segments in the approximation is increased, the number of components required increases proportionately. It also requires manipulation of the amplitude and the duration of individual multivibrators to generate the desired signal. We have developed a signal generator that consists of an up to 256 step approximation of the signal to be simulated, thus ensuring high quality. The device is based on combined digital and analog circuitry and can be constructed with relative ease at a low cost.

Design and Construction of the Circuit If a signal is sampled at a sufficiently high rate, the samples stored sequentially and the digitized samples then fed to a digital-to-analog (D/A) converter in the same sequence, a faithful reproduction of the signal can be obtained. In the present application, the biological signals were sampled at a rate of 300-10,000 samples per second and the digitized samples were stored serially in an INTEL 1702-A programmable read only memory (PROM). An LSI-11 microcomputer equipped with analog-to-digital (A/D) conversion system was used to sample the signals and an IMSAI 8008 microprocessor was used to program the PROMs. The INTEL 1702-A PROM can store 256 bytes of data. If the signal is not long enough to contain 256 samples, the sequence of samples may be repeated so that two or more waveforms are stored in the PROM. The complete circuit diagram for the signal generator is shown in Fig. 1. Two 4 bit counters (SN 7493) are cascaded to make an 8 bit counter to address the INTEL 1702-A PROM containing the samples. The address is changed sequentially from 0 to 255 and then the cycle is repeated. The NE 555 timer is used as a clock generator for the counters. The 8 bit sampled data in the PROM is fed to the Datel IC BC8 digital-toanalog converter. The analog signal is then amplified and a DC offset is added to it. No low pass filter is used in the circuit to smooth the analog signal since the cutoff frequency required can be different for different biological signals. A suitable capacitor can be connected at the output to achieve the necessary filtering.

Controls External controls are provided for changing the frequency, amplitude and the DC offset of the signal generated. When 0031-6768/82/0395/0251/$01.00

252 +SV J

FREQUENCY CONTROL

o.,~, ~

22K

~1N4571

l

5.3 K

-I- 15V

5K ~ T R O L

--6.4v

A~ GND Vrc

_

A3 INTEL 1702-A

PROM

D~

=

D2 D5

:-

DI

1 5 ~

i o.f pf

DATEL

AMPLITUDE

5K

CONTROL

D4

2;

D5

iA6

D7

D6

A7 VGGVDD

=

, I_

:

2~_

~-

16

----"~

~

_~22K ~

'"

I ~" ~

B,OLOG,CAL SIGNAL

330 K

0.1 --~

-15V

Fig. 1. Circuit diagram of the biopotential waveform generator

Fig. 2A--C. Reproduced biopotential wa~eforms. (A) miniature endplate potentials (MEPPs), (B) electrocardiogram (ECG) and (C) a triphasic muscle motor unit potential recorded with the Macro EMG electrode. Calibration: horizontal - A: 1.5 ms, B: 2.5 s, C: 4 ms; Vertical

in the P R O M . The m o t o r unit potential recorded using a Macro E M G electrode (Stfilberg 1980; N a n d e d k a r et al. 1980) was sampled at 140 l*s intervals. By choosing the right P R O M a reproduction of the signal can be obtained for l a b o r a t o r y use or as a "live" demonstration. The signal generator described is relatively simple and inexpensive to build. It has been used as a M E P P simulator in the Neuromuscular Physiology L a b o r a t o r y of the University of Virginia and has proven to be useful as a voltage source for testing an analog or computerized neuromuscular data acquisition system. Relevant M E P P parameters such as amplitude, frequency (and rise time) and resting membrane potential can be changed individually by adjusting the amplifier gain, clock frequency and the D C bias, respectively. Using the same signal generator, a variety of other complex biopotential waveforms can be reproduced. They can be used as a voltage source in different neurophysiological studies or for testing biomedical instruments including electrocardiographs, electroencephalographs, and electromyographs. When voltage signals of irregular wave shape are required as a stimulus to the nervous system, the present device can be used with minor modification to generate such a stimulating voltage or current source. As a teaching aid, a library of P R O M s containing different signals can be developed to demonstrate different biological signals. Finally, since it provides signals similar to those actually measured by the computer, it is very useful as a high quality calibrator.

-- A, B, C: 1 V

the clock frequency is changed, the rate at which the samples are fed to the D / A converter is varied. Hence frequency control is used to alter the frequency or the repetition rate of the output signal. The amplitude control changes the gain of the amplifier to vary the amplitude of the signal. The bias control changes the D C level a d d e d to the summing amplifier to vary the D C offset. Results and Applications

Three biological signals reproduced by the signal generator are shown in Fig. 2. Each signal is stored in the P R O M in digitized form. The electrocardiogram (ECG) was sampled at 33ms intervals and one single record was stored in the P R O M . Miniature end-plate potentials (MEPPs) were sampled at 100 ~ts intervals. F o u r consecutive MEPPs are stored

References

Blazek V, Neelakantaswamy PS, Reddy VCVP (1975) Generation of complex waveforms for biomedical application. IEEE Trans Biomed Eng 22:535-536 Miliman J, Halkias CC (51972)Integrated electronics. McGraw-Hill, New York, NY Nandedkar SD, Kim YI, Sanders DB, Ann6 A (1980) Signal representation of macro EMG. IEEE Publishing Services, New York, NY IEEE Frontiers of Engineering in Health Care. pp. 296- 300 Pratapareddy VCV, Rajappan KP (1973) N-Phase astable multivibrator. Int J Electronics 35: 365-367 Sabah NH, Adham M (1974) A function generator for neurophysiological applications. IEEE Trans Biomed Eng 21 : 62- 63 Stfilberg E (1980) Macro EMG. A new recording technique. J Neurol Neurosurg Psychiat 43 : 475 - 482 Received October 14, 1981/Accepted August 28, 1982