A technique for measuring frequency response of

pressure, volume, and flow transducers

ANDREW C. JACKSON AND ALLEN VINEGAR

Department of Medicine, Children’s Hospital Medical Center, Department of Physiology,
Harvard Medical School, and Department of Physiology,
Harvard School of Public Health, Boston, Massachusetts 02115

JACKSON, ANDREW C., AND ALLEN VINEGAR. A technique is often expressed in polar coordinates, or as the ampli-
for measuring frequency response of pressure, volume, and tude ratio and the corresponding phase shift, as a func-
flow transducers. J. Appl. Physiol.: Respirat. Environ. Exercise tion of frequency.
Physiol. 47(2): 462-467, 1979.-A device and methodology is Techniques for estimating the frequency response of
presented for testing the frequency response of pressure, vol- pressure (4, 12) and flow transducers (2, 3, 10) have been
ume, or flow transducers. Also reported are responses of se- described. However, none of these devices can be used as
lected transducers of all three types over the range of 2-120 Hz.
Several pressure transducers tested had good frequency re- a general-purpose device to determine the response of
sponse when connected to the test system with a minimum of pressure, flow, and volume transducers. Being contin-
interconnecting fittings; others did not. Use of additional con- ually faced with the problem of determining frequency
nectors degraded the response as did the addition of air-filled responses of all three types of transducers we have de-
catheters. The frequency response of the pneumotachometers veloped a method for accomplishing this. The purpose of
tested were influenced largely by the response characteristics this report is to describe the equipment and the associ-
of the associated pressure transducer and interconnecting fit- ated methodology and to report examples of the fre-
tings. These results emphasize the need to test the response quency response of selected pressure, flow, and volume
characteristics of any transducer with specific connectors and transducers as determined by this technique.
fittings that are to be used to make the actual measurements of
pressure, volume, or flow.
METHODS
dynamic response; pneumotachometers; plethysmographs The apparatus (Fig. 1) includes the unique feature of
an enclosed speaker chamber. This unit consists of a rigid
wall container divided into two chambers separated by a
THE MECHANICAL BEHAVIOR ofthe respiratory system is loudspeaker. The chamber on one side of the speaker
generally described by relationships between applied (chamber 1) is sealed. The chamber on the other side of
pressures and resulting flows and/or changes in volume. the speaker (chamber 2) is also sealed except for a single
In the course of making such measurements it is often port. The pressure in chamber 1 (pl) is a function of the
useful to have an electrical signal proportional to one of volume displacement (Av,) of the speaker cone.’ From
these variables. To generate such a signal a transducer the ideal gas law the pressure pl can be expressed as
of the appropriate type is often used. It is of primary
importance that the electrical output of these transducers p1 = -p(Patm/V,)Av, (1)
should be uniquely and linearly related to the magnitude where Patm is the atmospheric pressure, VI is the volume
of the variable being measured. Also, since the respira- of chamber 1, and p equals unity when compressions are
tory system is a dynamic one (i.e., pressure, flow, and isothermal and in air equals 1.4, the ratio of specific
volume are usually changing with time) it is equally heats, when compressions are adiabatic. Thus, by mea-
important that a transducer be capable of responding suring the pressure in chamber 1, one can estimate the
fast enough to follow these changes. A convenient volume displacement produced by the speaker cone from
method of describing a transducer’s ability to follow the relationship
dynamic events is to examine its response to sinusoidal
--_ __---
variations in the variable over the appropriate frequency
range. This so-called frequency response can be defined ’ Throughout this paper the following notation and units will be
used: P, mean pressure, in cmH,O; p, acoustic (gauge) pressure, in
as the complex ratio of the resulting estimation of the cmHz0; V, mean volume, in liters; Av, volume change from mean
variable (i.e., the output of the transducer appropriately volume, in liters; and J?, time derivative of volume, or volume flow, in
scaled) divided by the actual variable. This complex ratio l/s.
462 0161-7567/79/0000-0000$01.25 Copyright 0 1979 the American Physiological Society
FREQUENCY RESPONSE OF TRANSDUCERS
I 1
Power I f
_ Amp
Pressure .. -
Transducers/U
FIG. 1. Schematic diagram of experimental
Avl = -(vdPPaWp,
The motion of the speaker cone will produce equal, but
opposite, volume changes in the two chambers: Avl =
-Avz. Since chamber 1 is sealed, speaker motion will only
change its pressure. However, in chamber 2 speaker
motion can raise the pressure, ~2, and/or move gas
through the port. If Av, is the volume change associated
with gas compression in chamber 2 and Av, is the volume
of gas that is moved through the port then
The amount of volume change associated
compression in chamber 2 is given by
Av, = -(V2/~Phn)p2
From Eqs. 2-4, we can derive the following expression
for the volume displaced through the port in terms of the
pressures in the two chambers
The derivative of v, with respect to time is just
volume flow rate through the port, or
VP = dAv,/dt = (Vldpl/dt
The analog electrical circuit of this mechanical system
is given in Fig. 2 where voltage is the analog of pressure
and current is the analog of flow. If the induced volume
changes are sinusiodal the current (flow) through the
branch representing the port is described in terms of the
voltage (pressure) at nodes 1 and 2 by
ir,, =jc3(Cglpl
where c3 is the angular frequency, in rad/s, j = J-1, Cgl
is the gas compression compliance in chamber 1,
VJ(/?Patm), and Cg2is the gas compression compliance
in chamber 2, VJ(PPatm). Thus, measures of the pres-
sures in chambers 1 and 2 provide enough information so
that either the volume (from Eq. 5) or the flow (from
Eq. 6) through the port can be estimated.
Frequency response ofpressure transducers. To meas-
ure the frequency response of a pressure transducer it is
connected to the rear chamber and its output compared
to the output of a reference pressure transducer, also in
chamber 1. The reference pressure transducer must be
one whose amplitude and phase response are within
acceptable limits and can thus be assumed to give a
463
I
Test rL
Pressure
Flow
Vot)Lme
C 91
Transducer
-
FIG. 2. Analogous electrical circuit of speaker chamber shown in
Fig. 1. See text.
apparatus. See text.
“true” pressure signal. For this purpose we used a Setra
0 model 239 pressure transducer (Setra Systems, Natick,
MA). The natural frequency of this transducer is re-
ported to be 2,000 Hz. The actual frequency response of
this transducer was determined between 20 and 120 Hz
by comparing its output to the output of a microphone
that had been calibrated against a secondary standard
between 20 and 20,000 Hz. The amplitude ratio of the
two outputs was within 5% and the phase difference was
within 5” over the range tested. Unlike the microphone
the Setra transducer is designed to operate down to 0 Hz.
Since transducers such as the Setra usually have good
frequency response characteristics at low frequencies and
with gas poor characteristics only at high frequencies we assume
that its response is acceptable at frequencies below 20
(4 Hz .
Frequency response of volume transducers. The fre-
quency response of a volume transducer can be measured
by connecting it to the port in chamber 2, oscillating the
loudspeaker at the desired frequencies, and comparing
the output of the transducer to the volume computed
from Eq. 5. This necessitates making pressure measure-
the ments in chambers 1 and 2 that would ordinarily require
two reference pressure transducers. However, if only one
+ Vndpndt)/(PPatm) (6) such transducer is available, as was the casein this study,
the following procedure can be used. Two transducers,
one of which is the reference transducer, are initially
connected to chamber 1 and their responsescompared at
all frequencies that will be ultimately used. At each of
these frequencies an appropriate compensation factor is
computed for the transducer with the poor response.
This transducer is then removed from chamber 1 and
connected to chamber 2 and all subsequent recordings
+ Cg2p2) from it are multiplied by the appropriate factor at any
particular frequency. Using this technique we were able
to compensate a Validyne model MP45 2-cmHa0 trans-
ducer so that its response, compared to the reference
transducer, was within 1% and 1” at frequencies between
2 and 120 Hz.
Frequency response of pneumotachometers. To meas-
ure the frequency response of a pneumotachometer it is
connected to the port in chamber 2 and the loudspeaker
is oscillated at the desired frequencies. The electrical
output of the transducer is compared to the flow com-
puted from Eq. 7.
Data acquisition. The signals from the transducers
monitoring pl (Setra model 239), p2 (Validyne model
MP45), and the transducer under test were amplified
464 A. C. JACKSON AND A. VINEGAR

(Ithaca model 451)) digitized (Digital Equipment model thermal time constant of the entire chamber (with the
ADll-K), and stored for subsequent analysis on a PDP speaker removed and port closed) by rapidly injecting
U/34 computer. The computer was also used to generate into the chamber a volume of gas with a syringe and
sine waves at the desired frequencies. This signal was recording the resulting pressure. The pressure rapidly
generated by a digital-to-analog converter (DEC model increased due to the initial volume change, then gradu-
AAll-K), amplified (Crown model D-150A, power am- ally decreased due to heat transfer from the system. The
plifier) and used to drive the loudspeaker (Trusonic time constant for this heat transfer was measured to be
model BOA 80). In these studies the response was deter- in the order of minutes, giving evidence that compres-
mined by sweeping through frequencies from 2 to 120 Hz sions in the entire container were adiabatic at all fre-
in ~-HZ increments. quencies we tested (2-120 Hz). Compressions in the
The signals p1 and p2 as well as the signal coming from larger chamber (chamber 1) are probably adiabatic as
the transducer being tested are of the form well since its volume and surface area are not substan-
tially different from those in the entire chamber (VI is
f;:(t) = Aisin(ot + &) + B;cos(cclt + 0;) approx 80% of total volume). In the second test we
where i = 1 for ~1, i = 2 for ~2, and i = 3 for the test compared the gas compression compliances at frequen-
transducer signal. The sine and cosine components of cies between 2 and 120 Hz. Note that if the port in
such signals can be computed by the relationships chamber 2 is closed, then (from Fig. 2 and circuit theory)
27T Avl = -Av2
Ai = 1/r sin(ot)f;:(t)dt (8) and
Jo
jaCglpl = -jwCs2p2
and
25-l

B i = Y77 cos(wt)f;:(t)dt (9) PA?2 = -c,,/c,, = -v,/v, (12)

Jo Thus, if compressions are adiabatic in both chambers
Discretized forms of integral Eqs. 8 and 9 are given by over the frequencies of interest, the ratio of their pres-
sures should be inversely related to the ratio of their
volumes (Eq. 12). Further, the ratio of PI/P2 should be
Ai = 2/N $ sin(2nr/N)f;:(n) (10)
n=l constant with frequency and its phase equal to 180”. We
found that this was in fact the case to within 5% and 5”,
respectively. The frequencies at which chambers of this

Bi = 2/N E cos(2nn/N)f;:(n)
n=l
Statham PMlSTC

where N is the total number of samples per cycle. A

each frequency at least nine cycles were generated. The
response during the first cycle was ignored because of
possible transients due to changing the frequency. During
the next eight cycles the three input signals were sampled
at 10 equally spaced times per cycle (thus, in Eqs. 10 and
11, n = 10). The 8 sets of 10 data points from each
separate channel at each frequency were then ensemble
averaged (6). The sine and cosine components of each
channel were estimated using Eqs. 10 and 11, where h(n)
(for n = 1, 10) is the lo-point ensemble average for each
channel. The frequency response of a pressure transducer HP 270 -T
PMlSTC
PM5 --
---,
4
is then computed by the complex ratio, (A3 + jB3)/
/
(A1 + jB,). The response of a volume transducer is
computed from (A3 + jBa)/Avp, where Av, is computed 4)
H
from Eq. 5 with pl = A1 + jB1 and pz = A2 + jB2. The : 0
response of a pneumotachometer is computed from (A3 e
+ jBs)/J&, where VP is computed from Eq. 6 with pl = A1
+ jB, and p2 = AZ + jB2.
180 - l

RESULTS 0 60 120
Frequency, Hz
Before using Eqs. 5 and 6 to compute the frequency
FIG. 3. Frequency response, amplitude (upper panel) and phase
response of volume and flow transducers we must first
(Lower panel), of commercially available pressure transducers; Statham
determine whether compressions in each chamber are models PM15TC and PM5, Validyne model MP-45 with 2- (Va12) and
adiabatic or isothermal. To answer this question two 50-cmH20 (Val 50) sensitivities, Hewlett-Packard models 268B and
tests were performed. In the first test we measured the 270, and Ailtech catheter-tip transducer.
FREQUENCY RESPONSE OF TRANSDUCERS
size behave isothermally were also predicted from theo-
retical considerations (1) to be at frequencies well below
those used in this study. We thus concluded that
compressions in both chambers are adiabatic and that
Eqs. 5 and 6 with p = 1.4 were appropriate for the
frequencies used.
Pressure transducers. Figure 3 shows the amplitude
and phase response of a variety of pressure transducers
where a minimum of connectors were used to attach
them to the chamber. Over the frequency range tested
their response characteristics fell into three categories.
1) One group had amplitude responses that were essen-
tially flat and phase response either flat or linear with
frequency (Validyne 50 cmHa0, Hewlett-Packard 268B,
and Ailtech transducers). 2) One transducer had such an
amplitude response that it significantly underestimated
pressure at frequencies much above 10 Hz (Hewlett-
Packard 270). 3) A third group overestimated pressure
at higher frequencies (Validyne 2 cmHn0, Statham
PMETc, and Statham PM5).
To demonstrate how the response of a transducer is
affected by the addition of connectors and catheters Fig.
4 is included. Here the response of a Validyne Z-cmHa0
transducer alone is seen to be adequate up to 40 Hz.
However, its response is severely degraded with the
addition of additional connectors and air-filled catheters.
Vohme transducers. As an example the frequency
response of a volume transducer (constant-volume small-
mammal plethysmograph, described by Vinegar et al.
(13) (except with a Setra 239 as the pressure transducer)
is given (Fig. 5). Between 2 and about 60 Hz this device
behaves like a pure time delay: the amplitude response
Validyne 2
PE 200
-180
60
Frequency,
FIG. 4. Amplitude (upper panel) and phase (Lower
of Validyne MP-45 transducer with 2 cmH20 sensitivity,
attachments (unlabeled curve), with male-male
addition of appropriate needle stub adapter, then with
(b). and 45 cm (c) PE-200 air-filled catheters.
465
Plethysmograph
-180
0
G 0
mc
e
180
0 60 120
Frequency, Hz
FIG. 5. Amplitude and phase response of volume transducer (con-
stant volume, small mammal plethysmograph).
was flat and phase response was linear with frequency.
Above these frequencies the device begins to behave
more like a second-order underdamped system.
Flow transducers. We tested two commercially avail-
able pneumotachometers: Rudolph (clinical model) and
Fleisch no. 1. The pressure drop across each pneumo-
tachograph was measured using both a Validyne MP45
pressure transducer with a sensitivity of 2 cmH20 and a
/ 7 Validyne MP45 transducer with a sensitivity of 60
cmHzO. The frequency response of the pneumotachom-
eter-transducer systems depended on the pressure trans-
ducer used (Fig. 6). We tested the pressure transducers
alone with the same fittings used to connect them to the
pneumotachometer and found that the degradation in
the response of the pneumotachometer-transducer sys-
tem is primarily due to the pressure transducer and its
fittings and not the pneumotachometer itself (Fig. 7).
Like Finucane et al. (2) we found that the pneumo-
tachometer with the screen flow element (Rudolph) had
better frequency-response characteristics than the pneu-
motachometer with capillary tube elements (Fleisch).
Our response curve for the Fleisch no. 1 is qualitatively
similar to those of the Fleisch nos. 2 and 4 reported by
Peslin et al. (lo), but different than those of the Fleisch
nos. 2 and 3 reported by Finucane et al. (2). We have
shown here that the response of the pneumotachometer-
transducer system is largely dependent on the character-
istics of the pressure transducer and the tubing used to
connect it to the pneumotachometer. Peslin et al. (10)
used a Statham PM283 pressure transducer; Finucane et
Hz al. used either a Hewlett-Packard model 268B or 270
(they did not specify which model was actually used).
panel) response The differences between the reported responses could
with minimum
Luer slip fitting, with
therefore be due to the differences in the characteristics
15 cm (a), 30 cm of the pressure transducers and the tubing used to con-
nect the pneumotachometer and transducer.
466 A. C. JACKSON AND A. VINEGAR

Fleisch #I

FIG. 6. Amplitude and phase re-

sponse of commercially available pneu-
motachographs: A, Rudolph, clinical
model; B, Fleisch no. 1 using Validyne
MP-45 pressure transducers with 2- and
480 60-cmHz0 sensitivities.
-180

180 180
60 60
Frequency, Hz Frequency, Hz

on the dynamic response characteristics of transducers

are not severe. In normal adult humans 95% of respira-
tory signals are below 3.5 Hz during quiet breathing,
below 26 Hz during panting, and below 4 Hz during the
forced expiratory maneuver (8). All the pressure trans-
ducers that we tested (Fig. 3) have adequate amplitude
response to accurately record these maneuvers. The
pneumotachometer-transducer systems tested should as
well accurately reproduce flow signals during quiet
breathing and the forced expiratory maneuver. However,
during panting these pneumotachometers with the Vali-
dyne 2-cmH,O transducer will tend to overestimate the
-180 high-frequency components of the signal. It is important
to note that the response of pressure and flow transducers
might be severely degraded when attached through
adapters or tubes other than those used here.
Several recent reports are concerned with the re-
sponsesof lungs to high-frequency oscillations. Peslin et
al. (11) reported transfer impedances at frequencies up
to 70 Hz, Michaelson et al. (9) measured input impedance
up to 35 Hz, Lanser et al. (7) measured input impedances
180
60 120
up to 20 Hz, and Jackson et al. (5) reported input imped-
0
Frequency, Hz
ance values at frequencies from 150 to 10,000 Hz. In all
these instances the selection of appropriate transducers
FIG. 7. Amplitude and phase response of Rudolph pneumotacho- becomes most critical.
graph (solid curues) using Validyne 2- and 60-cmH20 pressure trans-
ducer with necessary fittings, and pressure transducers themselves with Measurements of respiratory variables in infants and
fittings (dashed curue) used to generate solid curve. small mammals may also require transducers with im-
proved response characteristics. A systematic analysis of
the frequency content of breathing in either infants or
DISCUSSION
small mammals has not appeared in the literature. But,
Much effort in the field of respiratory mechanics is since infants and small mammals have breathing rates
oriented toward estimating parameters such as compli- higher than human adults their respiratory signals might
ance, resistance, and inertance from measurements of be expected to have higher frequency contents. Thus, the
pressure, flow, and volume. Measurements are usually data of McCall et al. (8) might not be applicable in these
made on human subjects during spontaneous maneuvers cases and some of the transducers tested here might be
such as quiet breathing and panting, or in some cases inadequate for making these measurements.
during respiratory gymnastics such as the forced expira- In summary, we have described a device that can be
tory vital capacity maneuver. In such casesthe demands used to measure the frequency response of pressure,
FREQUENCY RESPONSE OF TRANSDUCERS 467

volume, and flow transducers. This device was tested by under the same conditions as those that are to be used in
measuring the response characteristics of several trans- the experiment. The device, here described, allows one
ducers and these are included as examples. Investigators to make these determinations.
should be warned not to assume that another transducer
of a make and model similar to one studied here will
necessarily have the same response characteristics as we
This work was supported in part by National Institutes of Health
found. We did not investigate the variability among Grants HL-17382 and HL-22178. A. Vinegar was the recipient of Na-
several different transducers of the same type. But we tional Institutes of Health Fellowship NRSA-07118.
did find that the response of a pressure or flow transducer Present address of A. Vinegar: Dept. of Environmental Health,
is markedly affected by the geometry of connectors. We University of Cincinnati Medical Center, 3223 Eden Ave., Cincinnati,
OH 45267.
thus suggest that if response characteristics are crucial
then the responses of transducers should be determined Received 8 December 1978; accepted in final form 7 April 1979.

REFERENCES
1. DANIELS, F. B. Acoustical impedance of enclosures. J. Acoust. Sot. 8. MCCALL, C. .B., R. E. HYATT, F. W. NOBLE, AND D. L. FRY.
Am. 19: 569-571, 1947. Harmonic content of certain respiratory flow phenomena of normal
2. FINUCANE, K. E., B. A. EGAN, AND S. V. DAWSON. Linearity and individuals. J. AppZ. Physiol. 10: 214-218, 1957.
frequency response of pneumotachographs. J. Appl. Physiol. 32: 9. MICHAELSON, E. D., E. D. GRASSMAN, AND W. R. PETERS. Pul-
121-126, 1972. monary mechanics by spectral analysis of forced random noise. J.
3. FRY, D. L., R. E. HYATT, C. B. MCCALL, AND A. J. MALLOS. CZin. Inuest. 56: 1210-1230, 1975.
Evaluation of three types of respiratory flowmeters. J. AppZ. Phys- 10. PESLIN, R., J. MORINET-LAMBERT, AND C. DUVIVIER. Etude de la
iol. 10: 210-214, 1957. reponse en frequence de pneumotachographes. BUZZ. Physio-
4. FRY, D. L., F. W. NOBLE, AND A. J. MALLOS. An evaluation of PathoZ. Respir. 8: 1363-1376, 1972.
modern pressure recording systems. Circ. Res. 5: 40-46, 1957. 11. PESLIN, R., J. PAPON, C. DUVIVIER, AND J. RICHALET. Frequency
5. JACKSON, A. C., J. P. BUTLER, E. J. MILLET, F. G. HOPPIN, JR., response of the chest: modeling and parameter estimation. J. AppZ.
AND S. V. DAWSON. Airway geometry by analysis of acoustic pulse Physiol. 39: 523-534, 1975.
response measurements. J. AppZ. Physiol.: Respirat. Environ. Ex- 12. SHAPIRO, G. S. AND L. J. KROVETZ. Damped and undamped fre-
ercise Physiol. 43: 523-536, 1977. quency responses of underdamped catheter manometer systems.
6. JACKSON, A. C., P. GULESIAN, AND J. MEAD. Glottis aperture Am. Heart J. 80: 226-236, 1970.
during panting with voluntary limitation of tidal volume. J. AppZ. 13. VINEGAR, A., E. E. SINNETT, AND D. E. LEITH. Dynamic mecha-
Physiol. 39: 834-836, 1975. nisms determining functional residual capacity in mice, lMus mus-
7. LANSER, F. J., J. NAGELS, M. DEMEDTS, L. BILLIET, AND K. P. VAN culus. J. AppZ. Physiol.: Respirat. Environ. Exercise Physiol. 46:
DE WOESTIJNE. A new method to determine frequency character- 867-871, 1979.
istics of the respiratory system. J. AppZ. PhysioZ. 41: 101-106, 1976.