APPENDIX F

DYNAMIC RESPONSE
OF INSTRUMENT SYSTEMS

In our discussions in this text, any temporal variations in the measured quantity
have been treated as random variations that contribute to the random uncertainty
in the measurement. However, it is also necessary for us to consider additional
error sources due to the response of instruments to dynamic, or changing, inputs.
An instrument may produce an output with both amplitude and phase (time lag)
errors when a dynamic input is encountered.
These dynamic response errors are similar to the variable but deterministic bias
errors discussed in Appendix E, and they can be very important in the analysis
of a timewise, transient experiment. In the following sections we present the
fundamentals needed to estimate these amplitude and phase errors.

F-1 GENERAL INSTRUMENT RESPONSE

The traditional way to investigate the dynamic response of an instrument is to

consider the differential equation that describes the output. We assume that the
instrument response can be modeled using a linear ordinary differential equation
with constant coefficients [1]

d ny d n−1 y dy
an n
+ an−1 n−1 + · · · + a1 + a0 y = bx(t) (F.1)
dt dt dt
where y is the instrument output, x is the input, and n is the order of the
instrument.
Instrument response to three different inputs will be discussed: (1) a step
change, (2) a ramp input, and (3) a sinusoidal input. These are illustrated in
Experimentation, Validation, and Uncertainty Analysis for Engineers, Third Edition H. W. Coleman and W. G. Steele 299
© 2009 John Wiley & Sons, Inc. ISBN: 978-0-470-16888-2
300 DYNAMIC RESPONSE OF INSTRUMENT SYSTEMS

Figure F.1 Step change in input to an instrument.

Figure F.2 Ramp change in input to an instrument.

Figures F.1, F.2, and F.3. Mathematically, these inputs are described as follows:

1. Step change x=0 t <0

(F.2)
x = x0 t >0
RESPONSE OF ZERO-ORDER INSTRUMENTS 301

Figure F.3 Sinusoidally varying input to an instrument.

2. Ramp x=0 t <0

(F.3)
x = at t≥0
3. Sinusoidal x = X sin(ωt) t >0 (F.4)

The response of zero-, first-, and second-order instruments to these inputs are
considered next.

F-2 RESPONSE OF ZERO-ORDER INSTRUMENTS

Since n = 0 for a zero-order instrument, Eq. (F.1) reduces to an algebraic

equation:
y = Kx(t) (F.5)

where K (= b/a0 ) is called the static gain. Equation (F.5) shows that the output
is always proportional to the input, so there is no error in the output due to the
dynamic response. Of course, there will be static errors of the types we have
discussed previously.
An example of a zero-order instrument is an electrical resistance strain gauge.
The input strain  causes the gauge resistance to change by an incremental amount
R according to the relationship [2]

R = F R (F.6)

where F is the gauge factor and R is the resistance of the gauge wire in the
unstrained condition. Since the instrument itself, the gauge wire, is experiencing
the input strain directly, there is no dynamic response error in the output.
302 DYNAMIC RESPONSE OF INSTRUMENT SYSTEMS

F-3 RESPONSE OF FIRST-ORDER INSTRUMENTS

The response equation for first-order instruments is usually written in the form
dy
τ + y = Kx (F.7)
dt
where τ (= a1 /a0 ) is the time constant and K (= b/a0 ) is the static gain. The
definition of a first-order instrument is one that has a dynamic response behavior
that can be expressed in the form of Eq. (F.7) [3].
A first-order instrument experiences a time delay between its output and a
time-varying input. An example is a thermometer or thermocouple that must
undergo a heat transfer process for its reading to respond to a changing input
temperature.
The response of a first-order instrument to a step change is found by solving
Eq. (F.7) using Eq. (F.2) for x and the initial condition that y = 0 at t = 0. This
solution can be expressed in the form
y
= 1 − e−t/τ (F.8)
Kx0
which is plotted in Figure F.4. In one time constant, the response achieves 63.2%
of its final value. One must wait for four time constants (4τ ) before the response
y will be within 2% of the final value.
The response to a ramp input is found by solving Eq. (F.7) using Eq. (F.3)
for x and the initial condition that y = 0 at t = 0. This solution is

y = Ka[t − τ (1 − e−t/τ )] (F.9)

which can also be expressed as

y − Kat = −Kaτ (1 − e−t/τ ) (F.10)

Figure F.4 Response of a first-order instrument to a step change input versus nondi-
mensional time.
RESPONSE OF FIRST-ORDER INSTRUMENTS 303

Figure F.5 Response of a first-order instrument to a ramp input versus time.

Equation (F.9) is plotted in Figure F.5. For no dynamic response error, we would
obtain y = Kat and the right-hand side (RHS) of Eq. (F.10) would be zero.
The two terms on the RHS therefore represent the error in the response. The
exponential term (Kaτ e −t/τ ) dies out with time and is called the transient error.
The other term (−Kaτ ) is constant and proportional to τ . The smaller the time
constant is, the smaller this steady-state error will be. The effect of the steady-state
error is that the output does not correspond to the input at the current time but
to the input τ seconds before.
The response of a first-order instrument to a sinusoidal input is found by
solving (Eq. F.7) using Eq. (F.4) for x. This solution is
KX
y = Ce−t/τ + √ sin(ωt + φ) (F.11)
1 + ω2 τ 2
where

φ = tan−1 (−ωτ ) (F.12)

and C is the arbitrary constant of integration. The exponential term in Eq. (F.11)
is the transient error that dies out in a few time constants. The second term on
the RHS of Eq. (F.11) is the steady sinusoidal response of the instrument.
By comparing the steady response with the input [Eq. (F.4)], we see that
the √response has an amplitude error proportional to the amplitude coefficient
(1/ 1 + ω2 τ 2 ) and a phase error φ. These errors are shown in Figures F.6
and F.7. Each of these errors varies with the product of the time constant τ and
the frequency of the input signal ω. As ωτ increases, the amplitude coefficient
decreases and the deviation from a perfect response becomes greater and greater,
as seen in Figure F.6. A similar behavior is observed in Figure F.7 for the phase
error, which asymptotically approaches –90◦ as ωτ increases.
304 DYNAMIC RESPONSE OF INSTRUMENT SYSTEMS

Figure F.6 Amplitude response of a first-order instrument to a sinusoidal input of fre-

quency ω.

Figure F.7 Phase error in the response of a first-order instrument to a sinusoidal input
of frequency ω.

F-4 RESPONSE OF SECOND-ORDER INSTRUMENTS

With n = 2 in Eq. (F.1), the response of a second-order instrument becomes

d 2y dy
a2 2
+ a1 + a0 y = bx(t) (F.13)
dt dt
If this expression is divided through by a2 , it can be written in the form

d 2y dy
2
+ 2ζ ωn + ωn2 y = Kωn2 x(t) (F.14)
dt dt
RESPONSE OF SECOND-ORDER INSTRUMENTS 305

√
where K(= b/a0 ) √ is again the static gain, ζ(= a1 /2 a0 a2 ) is the damping
factor, and ωn (= a0 /a2 ) is the natural frequency. The definition of a
second-order instrument is one that has a dynamic response behavior that can be
expressed in the form of Eq. (F.14) [3]. Instruments that exhibit a spring–mass
type of behavior are second order. Examples are galvanometers, accelerometers,
diaphragm-type pressure transducers, and U-tube manometers [1].
The nature of the solutions to Eq. (F.14) is determined by the value of the
damping constant ζ . For ζ <1, the system is said to be underdamped and the
solution is oscillatory. For ζ = 1, the system is critically damped , and for ζ >1
the system is said to be overdamped .
The second-order instrument response to a step change is found by solving
Eq. (F.14) using Eq. (F.2) for x and the initial conditions that y = y  = 0 at t = 0.
The solution depends on the value of ζ and is given by:
ζ > 1:
  
 ζ 
−ζ ωn t
y = Kx0 1 − e cosh(ωn t ζ2 − 1) +  sinh(ωn t ζ2 − 1)
ζ2 − 1
(F.15)
ζ = 1:
y = Kx0 [1 − e−ωn t (1 + ωn t)] (F.16)

ζ < 1:   
1 
−ζ ωn t
y = Kx0 1 − e  sin(ωn t 1 − ζ2 + φ) (F.17)
1 − ζ2

where 
φ = sin−1 ( 1 − ζ 2 ) (F.18)

This response is shown in Figure F.8. Note that 1/ζ ωn is now the time
constant. The larger ζ ωn is, the more quickly the response approaches the
steady-state value. The form of the approach to the steady-state value is
determined by ζ . For ζ < 1, the response overshoots, then oscillates about the
final value while being damped.
Most instruments are designed with damping factors of about 0.7. The reason
for this can be seen in Figure F.8. If an overshoot of 5% is allowed, a damping
factor ζ  0.7 will result in a response that is within 5% of the steady-state value
in about half the time required by an instrument with ζ = 1 [1]. Note that the
steady-state solution for all values of ζ > 0 gives Kx 0 .
The response to a ramp input also contains a transient and steady-state portion.
The steady-state solution is
 
2ζ
y = Ka t − (F.19)
ωn
306 DYNAMIC RESPONSE OF INSTRUMENT SYSTEMS

Figure F.8 Response of a second-order instrument to a step change input for various
damping factors.

The response lags behind the input by a time equal to 2ζ /ωn . High values of ωn
and/or low values of ζ reduce this lag in the steady-state response.
The response of a second-order instrument to a sinusoidal input is (at steady
state) given by

KX
y= sin(ωt + φ) (F.20)
[(1 − + (2ζ ω/ωn )2 ]1/2
ω2 /ωn2 )2

where  
2ζ ω/ωn
φ = tan−1 − (F.21)
1 − ω2 /ωn2

As in the first-order system, the response contains both an amplitude error pro-
portional to an amplitude coefficient and a phase error. These errors are shown
in Figures F.9 and F.10.
From Figure F.9 we see that for no damping (ζ = 0) the amplitude of the
response approaches infinity as the input signal frequency ω approaches the
instrument natural frequency ωn . In general, this maximum amplitude, or res-
onance, will occur at

ω = ωn 1 − 2ζ 2 (F.22)

Note that for nonzero damping (which is always the case physically) the ampli-
tude at this resonant frequency will be finite. Also note that for ζ  0.6 to 0.7 and
RESPONSE OF SECOND-ORDER INSTRUMENTS 307

Figure F.9 Amplitude response of a second-order instrument to a sinusoidal input of

frequency ω.

Figure F.10 Phase error in the response of a second-order instrument to a sinusoidal

input of frequency ω.
308 DYNAMIC RESPONSE OF INSTRUMENT SYSTEMS

ω/ωn < 1, the amplitude error is minimized and the phase error is about lin-
ear, which is desirable because this produces minimum distortion of the input
signal [1].

F-5 SUMMARY

The dynamic response of zero-, first-, and second-order instruments has been pre-
sented for step, ramp, and sinusoidal inputs. In the case of a zero-order instrument,
it was found that there is no dynamic response error. For first- and second-order
instruments, there are time delays for step and ramp inputs and amplitude and
phase errors for sinusoidal inputs. By choosing or designing instruments with
appropriate values of time constant and natural frequency, the effects of these
errors can be minimized.
In a complete measurement system, different instruments will usually be con-
nected in the form of a transducer, a signal conditioning device, and a readout.
An example might be a thermocouple (a first-order instrument) connected to an
analog voltmeter (a second-order instrument). In such cases the dynamic output
of the first instrument can be determined and this value can be used as the input
to the second instrument. The dynamic response of the second instrument to this
input is then determined to obtain the dynamic response of the system. In most
cases the significant dynamic response error effect will occur with only one of
the instruments, usually the transducer.

REFERENCES

1. Schenck, H., Theories of Engineering Experimentation, 3rd ed., McGraw-Hill, New

York, 1979.
2. Holman, J. P., Experimental Methods for Engineers, 4th ed., McGraw-Hill, New York,
1984.
3. Doebelin, E. O., Measurement Systems Application and Design, 3rd ed., McGraw-Hill,
New York, 1983.