CHAPTER 7

Bioelectric Amplifiers

7-1 Objectives
1. Be able to state the requirements for a bioelec 4. Be able to state the principles of operation of
tric amplifer. isolation amplifiers.
2. Be able to describe the basic principles of op 5. Be able to describe the problems associated
erational amplifiers. with the acquisition of bioelectric phenomena.
3. Be able to draw several different bioelectric 6. Be able to draw the elements of a medical data
amplifier configurations. acquisition system.

7-2Self-evaluation questions 4. Find the voltage gain of an inverting follower

These questions test your prior knowledge of that has a 100-k2 feedback resistor and a
the material in this chapter. Look for the answers 5-k2 input resistor.
as you read the text. After you have finished 5. Describe an isolation amplifier. Why is it used
studying the chapter, try answering these ques as a bioelectric amplifier?
tions and those at the end of the chapter.
6. State the mathematical expression that gives
1. List the features required in a the gain of an noninverting follower.
bioelectric
amplifier.
2. List the basic properties of the 7-3 Bioelectric amplifiers
amplifer.
operational
Amplifiers used to process biopotentials are called
3. Draw the circuit for an bioelectric amplifers, but this designation applies
operational amplifier to a large number of different types of amplifiers.
integrator, and describe its operation. The gain of a bioelectric amplifier, for example,
144
 Bioelectric Amplifiers 145

may be low, medium, or high (i.e., X10, X100,

XI000, X10,000). Similarly, some bioelectric am- Medium-gain amplifiers are those that
provide
gain factors between X10and X1000 and are used
plifiersare ac-coupled, while others are dc-coupled. for the recording of ECG waveforms,
The frequency response of typical bioelectric muscle po
tentials, and so forth.
amplifiers may be from dc (or near-dc, i.e., 0.05
kHz. High-gain, or low-level, signal amplifiers have
Hz) up to 100 gain factors over X1000, with some having factors
De-coupling is required where the input signals as high as x1,000,000. This type of
are clearly de or change very slowly (some in vivo amplifier is
used in very sensitive measurements, such as the
0, levels change in mm Hg per minute or per
recording of brain potentials (EEG).
hour). But even at frequencies as low as 0.05 Hz. Two important parameters in bioelectric ampli
ac-coupling may be used instead of dc. The reason
fiers, especially those in the high- and medium
for this is to overcome electrode offset potentials. gain classes, are noise and drift. Drift is the (spuri
In the ECG amplifier, for example, frequency ous) change in output signal voltage caused by
components as low as 0.05 Hz might be processed. changes in operating temperature (rather than in
But the electrode-skin connection produces an
put signal changes). Noise, in this case, normally
electrode offset (dc) potential that will interfere is the thermal noise generated in
with the ECG signal. The amplifier, therefore, resistances and
semi conductor devices. Good design and prudent
must be ac-coupled to block the dc offset in the in
put signal, yet have a frequency response down to component selection reduce these problems to the
negligible level in modern equipmert.
0.05 Hz to faithfully reproduce the patient's ECG All three classes of bioelectric amplifiers must
waveform.
have a very high input impedance. This require
The high-frequency response is the frequency ment is the one commonality among all bioelectric
at which the gain drops 3 dB below its midfre
quency value. In some cases the -3 dB high amplifiers, because almost all bioelectric signal
sources exhibit a high source impedance. Most
frequency point will be a frequency as low as 30 bioelectric sources have an impedance between
Hz, but in most cases it is 10 kHz. Specialized 10° and 10' 2, and ordinary engineering design
models used to process specific waveforms may practices dictate an amplifier input impedance that
have a particular response. ECG amplifiers, for is at least an order of magnitude higher than the
example, usually have a frequency response of source impedance. Modern metal oxide semicon
0.05 to 100 Hz.
ductor field effect transistor (MOSFET) and junc
A few general-purpose amplifiers have ad tion field effect transistor (JFET) amplifier (opera
Justable frequency response and are thus usable tional amplifier) devices have input impedances on
Tor a wide range of applications. In
Wise to use only the minimum general, it is the order of 1 teraohm (10" ).
frequency response
heeded to ensure good reproduction The properties of the integrated circuit (1C) op
of
Waveform. This practice permits rejectiontheof input
high
erational amplifier make it especially well suited
as a bioelectric amplifier. A discussion of opera
frequency noise. tional amplifier theory follows. A greater under
Low-gain amplifiers are those with gain factors standing will be achieved as we discuss basic prin
Detween Xland X10, The unity-gain
fier is used (X1l) ciples and then practical circuits made from
mostly for isolation, buffering,ampli
POssibly impedance transformation between and commercially available components. These parts
help make biomedical
sOurce and readout device. signal instrumentation what it is
ae often used for the Low-gain amplifierS today. In addition, circuit applications are shown
measurement of action po
tentials and other relatively high--amplitude to demonstrate the versatility of operational ampli
electric events. bio- fiers and to stress their importance in biomedical
instrument circuits.
 146 CHAPTER 7

The power supply shown in Figure 7-]b bos

7-4 Operational amplifiers two batteries, but an
ac-mains-S-operated bipolar
supply will also work. Battery E, forms the Vce
The operational amplifier (op-amp) is a device
that behaves in a unique manner:The properties of supply and is positivewith respect to ground. Bat.
tery E, forms the Ve supply and is connected so
the circuit containing an operational amplifier are
determined by the properties of the negative feed that it is negative with respect to ground.
back loop. For the elementary voltage amplifier Note on the op-amp symbol shown in
Figure
circuit configurations, we require only the basic 7-la that there is no ground terminal on the on
properties of the device, Ohm's law, and Kirch amp. The only ground connection in this oner
hoff's law to derive the transfer equations. The au tional amplifier is formed at the junction of the
thor of one book on op-amp circuit design* was two power supplies.
reportedly tempted to title his book Ohm's Law The op-amp has two inputs: the inverting input
With Applications. There are more elegant, more and the noniverting input. These are indicated by
mathematical metheds for analyzing the behavior
(-) and (+)signs, respectively.
of the voltage amplifier configurations, but none
shows circuit action quite as vividly as the simple
method presented here, as follows. Vecl+)
The op-amp gets its name from the fact that it
was originally conceived to solve mathematical
operations in analog computers. Although analog
computers are no longer in widespread use, many Inverting
electronic instruments use operational amplifiers input
and are, in effect, scaled-down, dedicated, or sin Noninverting -o Output
gle-purpose analog computers. input
Commercial operational amplifiers have been
available since the early 1950s, and integrated
circuit op-amps since the mid-60s. Prices of
modern IC operational amplifier devices range Ved-)
from less than a dollar for
low-quality units to
dozens of dollars for high-grade
(a)

units. Many premium grade op-amps specialized

are priced -o Vcc+)
in the
$2 range.
less-than-$10 range, and some are in the
The circuit symbol for an
Figure 7-la. The amplifier mustop-amp is shown in
quadrants, so the output terminaloperate
must
in all four
be able to -oGround
swing either to the positive or to the
power supply (Figure 7-1b), negative. The
therefore,
bipolar; that is, it must consist of two must be
one positive to ground and a second supplies,
ground. negative to
oVegt-)
(6)
*John I. Smith, Modern Flgure 7-1
Operationa! Circuit Design, Wiley
Interscience (New York, 1971). The operational
amplifier. (a) Circuit symbol. (b)
Power supply configuration.
 Bioelectric Amplifiers 147
The inverting input produces an output signal
that is 180 degrees out of phase with
the input sig- 7-4-1 The properties of ideal
nal. This is called inversion of the signal. op-amps
The noninverting input produces an output sig-
that is in phase with the input signal. We can analyze the op-amp by
nal
There is
nophase inversion of the signal between input and lowing ideal properties: considering the fol
output. 1. Infinite open-loop (i.e., no feedback) voltage
Both inverting and noninverting inputs offer the gain (Avol = 0).
same gain, so we may conclude that these respec- 2. Zero output
tive inputs have equal but opposite phase effect at impedance (Z, = 0).
the output. 3. Infinite input impedance (Z, ).
The various signal voltages that can affect the 4.
op-amp output terminal are shown in Figure 7-2. Infinite frequency response.
E, is appliedto the inverting input, while E, is ap- 5. Zero noise contribution.
pliedto the noninverting input. As long as E, and
6. Both inputs follow each other in feedback
Ex are not equal and of the same polarity, the op circuits. That is, in a circuit with negative
erational amplifier will see a differential input
feedback, a voltage applied to one input allows
voltage consisting of b; - E. The output voltage us to treat the other input as if it were at the
will be proportional to the gain of the stage and the same potential.
diference between E, and E,.
Common-mode signal voltages are those that These six properties will be cited frequently
are common to both inputs, such as E, or where throughout this chapter.
E, and E, are of equal magnitude and have the Before continuing our analysis of the feedback
same polarity. In common-mode situations, the voltage amplifier configurations, let us consider
differential voltage between the inputs is zero, so first some implications of these properties.
the output is zero. Infinite open-loop voltage gain. The open-loop
The common-mode rejection ratio (CMRR) of voltage gain (Avo) of any amplifier circuit is the
an op-amp is an expression of how nearly any gain without any feedback. In the ideal op-amp,
given device approximates the ideal situation, in this gain is defined as infinite. An implication of
which a common-mode signal has no effect on the this property is that the closed-loop characteristics
output terminal voltage. of the circuit are determined entirely by the prop
erties of the feedback loop network and are inde
pendent of the amplifying device.
Zero output impedance, Z, =0, implies that the
output is an ideal voltage source.
Infinite input impedance, Z; = , tells us that
the input terminals neither sink or source any cur
Eou
rent, nor do they load any circuit to which they are
connected.
Property 6is crucial to our circuit analysis: The
Differential voltage :E- E1 inputs tend to followeach other: This means that
Common voltage : E3, same
E =E)
we treat both inputs as if they were at the
Figure 7-2 potential. If a given voltage is applied to, say,
treat the
Signal voltage sources. the noninverting input, then we must
148 CHAPTER 7

Summing R2
inverting input as if it were at the same potential. junction
In fact, if a voltage is applied to one input, a volt
meter would measure the same voltage at the other R A
input.
onerational amnli
 Bioelectric Amplifiers 159
RÊ R
7-7 Differential amplifiers
Gain Path 1 A differential amplifier
(diff-amp) produces an
output voltage that is proportional to the difference
Ei
between the voltage applied to the two input ter
minals. Since an operational amplifier has a pair
of differential input terminals, it
may be
Eout connected for use in a differential anmplifiereasily
con
Gain Path 2 figuration.
E, In the most elementary form of de differential
R4 amplifier (Figure 7-20) only a single ICopera
R2
E. =E, - E, Ay = R tional amplifier is required. In this particular cir
cuit, the voltage gain for differential signals is the
Eout =() (Ein) =En same as for inverting followers (i.e., A, = RJR),
If R, = R3= R = R4 provided that the ratio equality R/R RJRg is
maintained. It is standard practice to ensure this
Figure 7-20 equality by stipulating that R= Rg and R, = R4
Differential input amplifier basics (single-ended Then gain path 1 is equal to gain path 2.
output). Figure 7-21 shows differential input amplifier
circuit detail. Depicted is the external circuit and
some internal "guts" of the diff-amp, which is re
Circuits such as Figure 7-19 are used in many ally just an op-amp. All op-amps have plus and
medical instruments to compute a value (i.e., E) minus bias current b+ and lb-), output load cur
from several different input values (i.e., E through rent ,), internal common-mode and differential
Es). It is also possible to use the noninverting in resistance and capacitance, and output resistance,
put, in which case no polarity inversion takes Ro. Of course, the ground reference is actually the
place, and Equation 7-12 must be modified to re point between the power supplies.
flect the contribution of signals applied to the non Examining Figure 7-21 reveals that the diff
inverting input equation (Equation 7-11). amp is a combination of inverting and noninverting

R = 10 k2 Ry = 10 ks2

+Vcc
Ib

W
RC cm lo
Vos SRCdiff
RC cm2

Ib+
-Vcc

R4 = 10 k2
R3 = 10 k2

Figure 7-21
Diferential input amplifier detail.
 160 CHAPTER 7

op-amp configurations. The external components Figure 7-22 shows a commercially available
interact with each other and with the op-amp to one-chip differential amplifier. It has internal thin.
cause gain errors. For example, the external 50-62 flm resistors made of nichrome or nickel
Source resistance, R.. in the inverting path causes a chromium. These resistors have been trimmed .
0.5% gain error from the ideal of -1 V/V. Notice high precision by cutting with a laser beam, The
that the input resistance to the op-amp's inverting more the resistors are cut away by the automatic
input is 10 kh. The input resistance ratio on the laser trimming machine, the higher the resistance
noninverting input also has a 0.5% error. This is because there is less conductive material), When
why low input resistance diff-amps require very precise resistor ratio accuracy is reached, cutting
low source resistance to achieve high gain accu stops. In this component, the gain error is 0.005%
racy. The nice thing about the op-amp'spositive and the. common-mode rejection (CMR) is 100
(+) input is that it is high impedance (resistance dB. This means that the negative and positive gain
and capacitive reactance). It therefore does not paths are very nearly equal. Also, the gain and
load or cause an error on the R, - RA resistor di CMR temperature drift is very low-approxj
vider. We know that the internal op-amp resistance mately five parts per million per degree centigrade
causes very litle error, because Rem is usually (5 PPM/°C). It is really the resistor ratio that is
1000 MI. However, Cemy Usually < 5 pF, can changing slightly with temperature. This is much
cause a gain error at higher frequencies. better than could be achieved by using separate ex
This discussion on op-amp circuit detail con ternal discrete resistors. Here the advantage of
cludes that, although circuit gain is mostly depen modern monolithic circuits becomes evident. All
dent on external resistance, such as R1, R, R3, and resistors on the same chip tend to behave exactly
Ra, certain dc and ac errors are always present. like one another. Their ratio stays constant.
This information helps us understand how real The diff-amp is useful because it rejects com
world diff-amp circuits work. mon-mode voltages while amplifying the differen
tial signal of interest. For example,
suppose equal
60-Hz noise is present on each input, and one in
put is at 5 V dc and the other is at 2 V dc. The cir
2 RË INA105BM cuit in Figure 7-22 removes the noise and ampli
!5 fies the 3-V dc differential signal.
25 kS2 25 kS2 CMR of this commercially available Remember, the
very high because on-chip resistors have diff-amp is
been ra
6 tio-matched to make the + and gain paths
Out nearly equal. The unity gain diff-amp output is a
E, 3-V dc signal in which 60-Hz noise
31 R3 R interference is
greatly reduced. The differential amplification re
+In
25 k2 moves noise, because equal
is present on each input. Thecommon-mode noise
E 25 kS2

tracts the equal noise voltage todiff-amp

Ey = E-E, just sub
Gain Error = 0.005%
CMR = 100 dB give nearly zero
while amplifying the difference in the
Nonlinearity = 0.0002% nals present on its inputs. unequal sig
How low noise becomes
Figure 7-22 at the output depends on how
One-chip (monolithic) precision unity gain CMR is. high the diff-amp's
amplifier. (CourtesyCorporation)
1993 Burr-Brown of difference
Burr-Brown under copyright The circuit shown in Figures 7-20
suffers from the same through 7-22
ance as the inverting restrictions on input imped
follower, because the input
 Bioelectric Amplifiers 161

impedanceis limited by R, and Ra. If a high gain

and other low-output impedance differential signal
required, then a high Rg/R, ratio is required, yet SOurces at gains up to 100. If higher gains are re
is
practicalcircuit considerations limit the maximum
quired, then a premium-cost amplifier (allowing
values for theseresistors.
and mminimum feedback resistors greater than 10° 2) must be used,
The simple dc differential amplifier is used or an additional stage of amplification is required.
in circuits where the source impedance is An alternate solution, one that provides a much
mostly
low. Strain gauge transducers, for example, typi- higher input impedance, is to use the instrumenta
element resistances below 1k). Wheat- tion amplifier circuit described in section 7-7-1.
cally have
stone bridge strain gauges with these resistor val
eg have a resistance approximately equal to the
resistance of each element arm (assuming all arms 7-7-1 Instrumentation amplifiers
are equal). If the equivalent bridge resistance is as The solution to both high-gain and high-input im
bich as 1 k2, then the minimum value for R, and pedance problems is the instrumentation amplifier
R. in the dc amplifier is 10 times as great, or 10 (IA) (Figure 7-23). This circuit uses three opera
K0. Most low-cost IC operational amplifiers tional amplifiers, A, through A3. The two input
should not be used with feedback resistors greater amplifiers (i.e., A, and A) are connected in the
than 1MQ. These limitations of real (as opposed noninverting follower configuration, while the
to ideal) operational amplifiers limit the practical third amplifier is connected in the simple de dif
ferential amplifier circuit shown in Figure 7-20.
oain of a differential amplifier (Figure 7-20) to
10°/10", or 100. Let us initially simplify our circuit analysis by set
The circuit of Figure 7-20, then, can be used ting the gain of A, equal to unity (i.e., R4 = Rs =
with Wheatstone bridge strain gauge transducers R6 = R).

E
RA R
E4
A1
(E)

R,
R As
Ein
R3
Eout
|B
R,
(E)
R6
E3
E A, = D()
If R,= Rg
Ry =Rs
Rs =R,

Figure 7-23
Instrumentation amplifier.
162 CHAPTER 7

Let us also assume that E is applied to the non Erample 7-4

inverting input of amplifer A, and that E, is ap Find the gain of an IA (Figure 7-23) if the follow
plied to the noninverting input of amplifier Az. Ad ing resistor values are used: R, = 10 kh,R, = 500
ditionally, E, is the output of Az, and E, is the n,R, = 10 kn, and Rg = 100 kn.
output of A,. Voltages E, and E, are also shown at
the imverting inputs of A, and Az, respectively, Solution
again reflecting property 6.
There are two contributing sources E, and (7-17)
E4. In the case of E: (2 x 10 kn
0.5 k2
)-() (7-13)
A, = (40 + 1X10) = 410
and for E:
In ordinary practice the following equalitics are
observed: R,= Rs, Ry = R, Rs = R,. Interestingly
-Ex()-x() (7-14) enough, a mismatch of R, and R, has little effect
on the CMR ratio but does result in a differential
If we set R, = Rs (not essential, but it simplifies gain eTOr.
the analysis), and then combine Equations 7-13 IAs are used in biomedical applications be
and 7-14, we may write cause of several factors: ability to obtain high
gain with low resistor values, extremely high in
put impedance, and superior ejection of com
(G, -E) -(E- E ) mon-mode signas. Slight resistor mismatches in
the circuit of As can degrade CMR; therefore,
+ (E -) many designers use a potentiometer for R7. The
potentiometer is adjusted (while a high-level,
(E, - E,) = (E - E,)+1+) common-mode signal is applied) for minimam
R R,/ output signal.
The physical form taken by an IA might be a
(E- E) =(E;- E)1) (7-15) discrete op-amp circuit, such as Figure 7-23, in
which three IC operational amplifßer devices are
Therefore: used. It may also take the form of a hybrid
function module, in which chip-form IC opera
2R; tional amplifiers and the resistors are constructed
A,-+1
R,
(7-16) on a thin ceramic substrate and then potted in a
block of epoxy resin. The third form possible is a
The voltage gain of the AJA, section is given monolithic (1C) instrumentation amplifiet. In boch
by Equation 7-16, ut when the gain of A, is the hybrid and monolithic versions there may be a
nonzero, we must also include a term that ac pair of external teminals for setting gain. These
counts for this extra gain. The gain of an IA such terminals are for an externally connected R, in Fig
as the one in Figure 7-23 is given by the transfer ure 7-23. The gain equation is usüally a constant
function divided by the value of R, connected between the
two terminals.
(7-17) Modern LA design can be tailored to meet
specific biomedical applications. The circuit in
 or centimeter.

7-11 Isolation amplifiers

Some hospital patients are extraordinarily suscep
tible to electrical shock hazards. It is believed that
60-Hz ac currents small enough to be deemed
harmless ordinarily may be lethal to a patient un
der certain circumstances.
on.
To prevent accidental internal cardiac shock.
the manufacturers of modem bioelectric ampli
h fiers, especialy those used in ECG recording, use
isolation amplifiers (iso-amps) for the direct pa
tient connection. These amplifiers provide as much
as 10" of insulation (isolation) between the pa
tient connector and the ac power mains line cord.
The basic design of an iso-amp is shown in Fig
ure 7-41. It is usually composed of an input
ampli
fier, some type of modulator, an isolation barier, a
demodulator, and an output amplifier. Modulation
schemes include amplitude,
voltage-to-frequency,
 Bioelectric Amplifiers 177

Input
amplifier modulator
Electrical Output
input demodulator amplifier
LED
Electrical
output

Input common Isolation

barrier Output common
Flgure 7-41
Basic design of an isolation amplifier.
duty cycle, pulse width, flyback loading, and oth
ers, Barriers can be optical, magnetic transformer. voltage (IMV) referenced to the iso-amp's output
must go through the large barrier resistance before
capacitive, or even heat transfer. Notice that there it can mix with the input signal. Hence,
is an input common and an output common that most of
are electrically isolated from one another to the the interfering voltage or noise is dropped across
the barrier; very little adds to the input.
tune of millions of ohms. The iso-amp is really an Since the barrier is not an infinite impedance,
energy converter. Electrical energy on the modula some error is created. We say that an IMV, shown
tor side is converted to some "nonelectrically con as Vim across the barrier, causes an error to appear
ductive" energy in the barier and then converted in the output voltage, Voutr Figure 7-44 shows how
back to electrical energy on the demodulator side. this error is calculated. The measure of how well
That's all there is to it. But what function does an the iso-amp attenuates or rejects the IMV is called
iso-amp really perform? isolation-mode rejection (IMR). The isolation
A circuit diagram symbol that is often used to mode rejection ratio in VN is called IMRR.
represent the iso-amp is shown in Figure 7-42. IMRR in V/V = log (IMRa/20). This is the re
This symbol has not been standardized, so some verse of the equation shown in Figure 7-44, IMR
manufacturers use variations of their own on their in dB = 20 log o(IMRRyN). The errOr is gained
circuit diagrams. up just like the input signal and results in some dc
Isolation amplifiers actually operate on the or ac voltage that adds to the normal signal. For
principle of attenuation. The high barrier imped example, from a commercially available product
data sheet, if IMR were 120 dB, then IMRR
ance (>102 n in parallel with <10pF) acts in
would be 1,000,000 V/V. If Vim were 1,000 V dc,
series between input and output as shown in Fig
ure 7-43. Therefore, an interfering isolation-mode the error at the isO-amp's input would be 0.001 V
or 1mV dc. This would be 0.1% eror if the in
put signal were 1 Vdc. If the iso-amp had a gain
of 10, then the output signal would be 10 V dc
+Input D and the error would be 10 mV dc. The error
would still be 0.1%. It is the same, provided that
Floating all voltages are taken with respect to the input
(RTI) or with respect to the output (RTO). Note
ground VouT
-Input that the iso-amp rejects low-level dc or ac inter
Chassis fering voltages appearing across the barrier just as
ground
it rejects kilovolts. In fact, the iso-amp is just as
attenuating
good for attenuating noise as it is for
Figure 7-42 high voltages.
Symbol for an isolation amplifier.
 178 CHAPTER 7

High barier impedance

attenuates the noise

10 to 10!2 2

VNoise

s0Q

Flgure 7-43 can be thought of as

Reduction of interference in an IA is by cancellation. In an iso-amp, it
attenuation.

In contrast, an LA cancels or rejects common ally higher (125 dB) than CMR in an IA (105 dB).
mode (CM) voltage present on each of its two in It is just the way they are constructed that makes
put terminals. A comparison of CMR to IMR ap the difference.
pears in Figure 7-45. CMR in an IA is a measure Sometimes an iso-amp may have CM noise on
of how well it rejects interference or noise. It does its input as well as IM noise across its barrier. Fig
this by cancellation through its balanced gain ure 7-46 shows this. Vem is noise on the two inputs
paths. IMR in an iso-amp is also a measure of how with RTI common. Vim is noise on the input with
well it rejects interference or noise. It does this by RTO common. It is across the barrier. If an IA
attentuation through its high-barrier impedance. If were inside the iso-amp shown, then CM as well
one were clever enough to reconfigure the IA cir as IM interference would be rejected. Here,
cuit at the top to make the CM voltage in the IA Voutgain( Vsig t Vm/CMRR
appear like isolation-mode (IM) voltage in the iso
amp, then one would get better rejection of noise. + Vim/IMRR) (7-20)
Why isthis so? Because IMR in an iso-anmp is usu In summary, modern isolation amplifiers serve
three purposes: (1) They break ground loops to
permit incompatible circuits to be interfaced to
oVoUT
Input Output Instrumentatlon Ampltier
Common Common
0More residal noe

VV,x Gain +Ve

CMR= 105 dB
Common

Ibolatoa AmpBler
Vour=(Vt ))x Gain =(VN XGain) +(BX Gain)
Signal Brror
V Laput common
VV, Xx Gain +VM en
DMR = 12s dB
M DMOutput commo
where IMR=20 LOGo IMRR Coechaion: Lo amp does a betor job of rejecting Doie, becao DMRo » CMu
(dB) (VN Ratio)
Flgure 745
Figure 7-44 Comparison: CMR versus IMR.
IMR eror.
 Bioelectric Amplifers 179

any one of the other isolation techniques must be

used.
V
VsIG IN 2 VsIG IN I Residual
noise 7-11-2 Carrier
VCM Noise VoUT
Figure 7-47a shows an isolation amplifier using
carrier technique. The circuitry inside the dashed
VIM Noise line is isolated from the ac power mains and the
rest of the circuitry that is powered from the ac
Vour =Gain (VsIG VCM ViM mains. In most cases, the voltage gain of the iso
CMRR IMRR lated section is in the medium-gain range (e.g.,
Figure 7-46 X10to X500).
An iso-amp may have common-mode (CM) noise on The isolation is provided by separation of the
its inout as well as isolation-mode (IM) noise across ground, power, and signal paths in the two sections
its barrier. by transformers T and T. These transformers
have a core material that is very inefficient at 60
Hz but works well in the 20- to 250-kHz range.
This feature allows the transformers to easily pass
gether while reducing noise; (2) they amplify sig the carrier signal but impedes any 60-Hz energy
nals while passing only low leakage current to pre that might be present.
vent shock to people or damage to equipment; and Although most models use a carrier frequency
(3) they withstand high voltage to protect people, in the 50- to 60-kHz range, there are several types
circuits, and equipment. that use almost any frequency in the 20- to 250
Several approaches to the design of isolation kHz range.
amplifiers are used: battery powered, carrier, opti The carrier oscillator signal is coupled through
cally coupled, and current loading. transformer T, to the isolated stages. Part of the
energy from the secondary of T, goes to the mod
ular stage; the remainder is rectified and filtered
7-11-1 Battery-powered and then used as an isolated dc power supply. The
This approach is perhaps the simplest to imple dc output of this power supply is used to power the
ment, but it is not always the most suitable for the input amplifiers and modulator stages.
customer's convenience because of problems in An analog signal applied to the input is ampli
herent in battery maintenance. A few products ex fied by A, and is then applied to one input of the
1st, however, that use a battery-powered, front-end modulator stage. This stage modulates the ampli
amplifier, even though the remainder of the prod tude of the signal onto the carrier.
uct is ac mains-powered. For cardiac output com Transformer T couples the signal to the input
puters, this approach is almost universally used. of the demodulator stage on the nonisolated side
The bioelectric amplifier in this type of instru of the circuit. Either envelop or synchronous de
ment is exactly like that in an ac-powered model. modulation may be used, although the latter is
The sole difference is that it is powered from abat more common. Ordinary dc amplifiers following
tery pack. the demodulator complete the signal processing.
This type of instrument must be totally self An example of a synchronous demodulator cir
Contained. If any external instrument or device cuit is shown in Figure 7-47b. This type of circuit
(.g-, oscilloscope, strip-chart recorder, rate meter, is based on switching action. Although the exam
battery charger) is used, then a model employing ple shown uses PNP bipolar transistors as the
 -Inq Synchronous
demodulator
+ In o Modulator and A o Output
Jlow-pass filter
dc Amplifier
Vee' 50 kHz
carrier
Input Ve' 50 kHz,
signal Carrier

ground

Isolated 50 kHz
power carrier
supply oscillator
Isolated
Nonisolated To
nonisolated
dc-to-dc circuitry
converter
dc
power supply
Floating Chasis
ground ground

ac
power mains
|110VACI6OHz
(a)

T,
Analog RA R
signal Reference
input signal

R
Output
C

(b)
Flgure 7-47
(a) Carier-type isolation amplifier. (b)
Synchronous demodulator.
 electronic switches, others use CMOS
switches or FET transistors. analog IC optoisolators that contain the LED and photo
The signal from the transistor inside of a DIP ICpackage.
of 50 kHz (up to 250 modulator has a frequency There are actually several approaches to optical
kHz, or even 500 kHz) and
is coupling. Two very popular methods are the car
amplitude>modulated with the
isolated amplifier. This signal is signal from the rier and direct methods. The carrier method is the
emitters of transistors and (via applied to the same as discussed in section 7-11-2, except that an
T) in push optoisolator replaces transformer T;.
pull. On one half of the cycle, therefore, the emit
ter of , will be positive with respect to The carrier method is not the most widespread
ter of . On altermate half cycles, the the emit in optically coupled amplifiers because of fre
situation occurs: 2 is positive with respectopposite
to 1
quency response limitations of IC optoisolators.
Only recently have these problems been resolved.
The bases of , and , are also driven in push
A more common approach is shown in Figure
pull, but by the 50-kHz carrier signal. This action 7-48. This circuit uses the same dc-to-dc converter
causes Q and Q to switch on and off, but out of
to power the isolated states as was used in other
phase with each other.
On one half of the cycle, we will have the po designs. This will keep A, isolated from the ac
power mains but is not used in the signal coupling
larities shown in Figure 7-47b. Transistor is process.
turned on. In this condition point A on T is The LED in the optoisolator is driven by the
grounded. The voltage developed across load re output of isolated amplifier A. Transistor ,
sistor R4 is positive with respect to ground. serves asa series switch to vary the light output of
On the alternate half cycle, O, is turned on, so the LED proportional to the analog signal from A,.
point B is grounded. But the polarities have re Transistor , normally passes sufficient collector
versed, so the polarity of the voltage developed current to bias the LED into a linear portion of its
across R4 is still positive. This causes a full-wave operating curve. The output of the phototransistor
output waveform across R4, which, when filtered, is ac-coupled to the remaining amplifiers on the
becomes a de voltage level proportional to the am nonisolated side of the circuit, so that the offset
plitude of the input signal. This same description condition created by LED bias is eliminated.
of synchronous demodulators also applies to the
circuits used in some carrier amplifiers. 7-11-4 Current loading
Avariation on this circuit replaces the modula
tor with a voltage-controlled oscillator (VCO) that A current loading isolation technique was used by
allows the analog signal to frequency-modulate a Tektronix in their portable medical ECG monitors.
carrier signal generated by the VCO. The power A simplified schematic is shown in Figure 7-49.
supply carrier signal is still required, however. A Notice that there is no obvious coupling path for
phase detector, phase-locked loop (PLL), or pulse the signal between the isolated and nonisolated
counting detector on the nonisolated side recovers sides of the circuit.
the signal. The gain-of-24 isolated input amplifier in Fig
ure 7-49 consists ofadual JFET (Q)and an oper
ational amplifier. This circuit illustrates use of
7-13-3 Optically coupled JFETs to improve the input impedance of an oper
Electronic optocouplers (also called optoisolators) ational amplifier.
The output of A, is connected to the isolated
are sometimes used to provide the desired isola
tion. In early designs of this class a light-emitting V (i.e., -10V dc power supply through resis
tor R). This power supply is a dc-to-dc converter
diode (LED) was sandwiched with a photoresistor
or phototransistor. Modern designs, however,
use operating at 250 kHz. Transformer T, provides
 Bioelectric Amplifiers 193

allow modern biomedical systemsto do abetter job are made worse by the high-gain amplifiers needed
of measuring important
temperatures. to increase the very weak biopotentials to a read
The final example of isolation is one that does able level.

nott really
break a ground loop; it just adds a mod- Noise is generated by almost every part of the
resistance in series. We say that when a very patient's body,
recording apparatus, including the contribution
erate of
resistance is provided between two points in but the worst offender is the noise
large isolated, But this is a matter of the amplifier itself.
a circuit, they are Drift is the change in gain or dc offset
(i.e.,
degree. The circuits shown in Figure 7-61 make
a commercially available compo- baseline) caused by thermal effects on the ampli
the point. Here, measurement, but its refer fier components. Drift may be substantially re
nent takes a differential
ence points (pins I and 5) are only
about 400 kO duced through the use of large amounts of negative
typical
away from either input (pin 2or 3). Since V be feedback in an ac-coupled amplifier. The problem
low
semiconductors can handle around 10to 20 to be solved is to convert a dc (or near-dc,
will
fore breakdown, something special must be done frequency analog) signal to an ac signal that
towithstand 200 V. Here is what is done:
The cir pass through the amplifier. analog
common The solution is to sample, or chop, the
cuits in Figure 7-61 first attenuate the the ac
mode voltage on each input by a factor
of 20. Two signal at a frequency that will pass through
biomedical
hundred volts are reduced to an acceptable 10 V. coupled amplifier. Although most excitation signal
chopper amplifiers use a 400-Hz
Then the reduced voltage is gained up by a factor 60-, 100-,. or
for the chopper, some models use
of 20. The result is a unity gain differential ampli 1000-Hz chopper frequencies.
400 kQ of
fier with a unique characteristic. With amplifier is
input resistance, this circuit handles 10-V
differen An example of a simple chopper
a vibrator
shown in Figure 7-62a. The chopper is
V maximum.
tial signals riding on as much as 200 driven single-pole, double-throw
(SPDT) switch
Dothe circuits in Figure 7-6lreplace classical and output termi
isolation amplifiers? The answer is: sometimes.
that grounds the amplifier inputswitch.
to make nals on alternate swings of the
There is not enough isolation resistance The chopper vibrator coil is excited
by a 400
but
such circuits suitable for ECG applications,moni Hz ac carrier signal. Figure 7-62b
shows the ana
as
they are excellent for applications such
and -200-V
chopped
log waveforms for both the original and signal
toring battery cells. Here +200-V versions. Only the chopped version of the
power supplies are employed to charge, say, 3012
supplies in will pass through the ac amplifier.
Vlead-acid batteries. Uses range from stability
The chopper technique not only gains provides
portable X-ray equipment to emergency power from the ac-coupled amplifier but also by itself,
Battery cells
backup in hospitals, for example.breakdown diff low-noise operation. The sampling rate, externally
near ground do not need a 200-V tends to act as a low-pass filter for noise of
require
amp, but cells near the maximum supplies
multiplexer generated noise, although it puts some
fur
some similar type of amplifier. The its own on the system. Most
manufacturers
can select any channel, representing any battery ac-coupled
converter, digi ther limit the noise by making the only
cell, which can be sent to an A/D amplifier a band-pass amplifier that passes
tized, and stored in a computer. the narrow range of
frequencies around 400 Hz.
any given
The rms value of the noise signal in factors)
other
7-12 Chopper-stabilized amplifiers system is proportional to (amongbandwidth. By
the square root of the circuit
tries to record low the noise
lwo problems arise when one limiting the bandwidth, we also limit
(i.e., EEG recording of brain
level biopotentials amplitude.
problems
Waves): noise and dc drift. Both of these
 Bioelectric Amplifiers 195
ac amplifier
A
A
H
differential-input
in Figure 7-63. In this chopper amplifier is shown
circuit, the chopper is on the
Inpur Low-pass
filter input circuit only. Input
Output nected so that its center taptransformer T, is con
of the input becomes one terminal
connector, while the two winding
tremities are connected to the chopper. The ex
the chopper switch pole of
Chopper
vibrator
400-Hz
the input connector.
becomes the other terminal of
excitation
Most of the gain in this circuit is
ac-coupled amplifier A,. The signalprovided
remains
by
chopped version of the input waveform until it isa
(a) applied to the synchronous demodulator, where it
is detected and filtered to
waveform.
recover the original
The chopper amplifier technique of Figure 7-63
is typically used in EEG amplifiers and those uni
versal bioelectric amplifiers that have gains in the
range of X 1000 and over.

7-13 Input guarding

Original signal Physiological signals tend to have low-level ampli
tudes. The -80-mV action potential is a power
house compared to the 1 mV of the ECG wave
form and the 50 Vof the EEG waveforms.
In most cases, the physiological signal of inter
est is accompanied by large CM signals. Fre
quently, several hundred millivolts of 60-Hz signal
will be coupled into the input cables of a bioelec
tric amplifier designed to sense and amplify low
level physiological signals.
This situation is shown in Figure 7-64. The cir
cuit is shown in Figure 7-64a, in which a differen
tial amplifier is connected to a differential signal
source through a shielded cable. Both differential
Chopped version (E) and common-mode (Ecm) signals are present.
(b) If the usual practice of grounding the shield is fol
Figure 7-62 lowed, then the equivalent circuit of Figure 7-64b
Chopper amplifier. (a) Simplified circuit. (b) Compari is obtained. Resistances Rand R, are the sum of
Son of original and chopped waveforms. the cable resistances and the signal source output
impedance. Capacitors Cand C, represent the ca
pacitance of the shielded cable.
The networks R,/C, and R,/C, are of little con
sequence, provided that R, = R, and C, = C, but
if these equalities are not maintained, the inputs
196 CHAPTER 7

400 Hz excitation

ac - coupled dc output
amplifier amplifier
Synchronous
demodulator
and
Output
low-pass filter

Chopper
vibrator X1.000 -X1.000.000

o
400 Hz + In - In
input Signal
input
Figure 7-63
Differential chopper amplifier.