See

discussions, stats, and author profiles for this publication at: http://www.researchgate.net/publication/50248629

A simulator for oscillometric blood-pressure

signals to test automated noninvasive
sphygmomanometers
ARTICLE in THE REVIEW OF SCIENTIFIC INSTRUMENTS FEBRUARY 2011
Impact Factor: 1.58 DOI: 10.1063/1.3549803 Source: PubMed

CITATIONS

DOWNLOADS

VIEWS

123

4 AUTHORS, INCLUDING:
Stephan Mieke
Physikalisch-Technische Bundesanstalt
33 PUBLICATIONS 195 CITATIONS
SEE PROFILE

Available from: Stephan Mieke

Retrieved on: 07 September 2015

REVIEW OF SCIENTIFIC INSTRUMENTS 82, 024303 (2011)

A simulator for oscillometric blood-pressure signals to test automated

noninvasive sphygmomanometers
W. Riedel, S. Mieke, R. Seemann, and B. Ittermanna)
Physikalisch-Technische Bundesanstalt, Abbestr. 2-12, D-10587 Berlin, Germany

(Received 5 November 2010; accepted 27 December 2010; published online 28 February 2011)
A device was developed allowing to generate simulated human blood pressure signals for the purpose of testing the performance of automated noninvasive sphygmomanometers. The apparatus reproducibly generates blood-pressure oscillations synthesized from prerecorded measurements on human subjects. These real-life data allow for a much better evaluation of the accuracy of blood-pressure
measurements than the existing simulators using artificial and thus less realistic waveforms. To assess
the performance of a given sphygmomanometer under both stable and varying conditions, generated
signals can be repeated in their original shape or distorted by well-defined artifacts. In comparison to
clinical tests, the procedural influences on the performance testing of sphygmomanometers are largely
reduced when the simulator is used. 2011 American Institute of Physics. [doi:10.1063/1.3549803]
I. INTRODUCTION

Hypertension is estimated to contribute about 4.5% to the

global disease burden, causing at least 7.1 106 premature
deaths per year.1 As a variety of effective treatments for hypertensions are known, the reliable detection of this disease,
i.e., reliable blood pressure measurements in primary healthcare, is key to address this global problem.
Sphygmomanometers are devices to measure human
blood pressure. The prevalent automated method for this purpose is the oscillometric principle.2 In most cases an inflatable cuff is wrapped around the upper arm. It is inflated to
about 30 mmHg above the systolic blood pressure3 to stop the
blood flow and is then gradually deflated. The arterial blood
pressure in the arm superimposes small pressure pulses (referred to as oscillations) onto the (slowly varying) baseline
pressure generated by the cuff. As long as the baseline pressure exceeds the systolic blood pressure, the brachial artery
is fully closed; small oscillations, however, are still present.
With decreasing baseline pressure they are getting larger and
less finely structured. Also, the baseline becomes less defined.
After passing through a maximum, the oscillation amplitudes
decrease again.
In an automated noninvasive sphygmomanometer, the
cuff pressure, i.e., the sum of baseline pressure and oscillations, is measured by a suitable sensor, digitized, and fed
to the internal microcomputer. At least the envelope function
is evaluated to compute systolic and diastolic blood-pressure
values. The algorithms applied thereby are derived empirically from the clinical studies and are usually not disclosed
by the manufacturers. To check the measurements of such a
device, it is thus necessary to test it in a black-box manner
by comparing it to a reference method in clinical trials, as described in several standards and protocols.47
Reference values for the diastolic and systolic blood pressures in a clinical trial are typically obtained by the auscultatory method which is the most commonly used technique for
manual blood-pressure measurements. This method is based
a) Electronic mail: Bernd.Ittermann@PTB.de.

0034-6748/2011/82(2)/024303/7/$30.00

on the detection of the so-called Korotkoff sounds which are

audible through a stethoscope applied to the artery in the inner
elbow below the cuff.
To perform an accurate reference measurement, the precautions described in the aforementioned standards and protocols have to be taken. Clinical trials are usually performed
on subjects with different blood-pressure ranges, arm circumferences, gender, age, etc.
At present, almost all automated oscillometric sphygmomanometers require a fairly constant heartbeat to measure
correctly. They do not permit artifacts otherwise their accuracy is lost. Some devices try to handle such situations by
special software routines, error messages, or corrective procedural measures. Under these conditions, it is practically impossible to evaluate the full performance in clinical trials, as
arrhythmic signals cannot be delivered in a reproducible way.
For functional rechecks of a device in the fielde.g., by
legal authorities or calibration servicesclinical trials are too
time-consuming and too expensive. Therefore, only the static
indication of the baseline pressure is checked. A deviation
of more than 3 mmHg is regarded as a malfunction.
This approach assumes that the clinically tested software
is unchanged and that the hardware other than the pressure
transducer, the signal amplifier, the analog-to-digital converter (ADC), and the cuff design does not affect the accuracy
of the measurement. A simple pressure pulse generator8 can
then be applied to detect the changes of a limited set of other
parameters.

II. SIMULATOR CONCEPT

Suitability and limitations of existing blood-pressure

simulators for the testing of sphygmomanometers were recently reviewed by Balestrieri and Rapuano.9 In the present
paper a new simulator is described. It is similar to the previous
devices in so far as it superimposes oscillations to a baseline
pressure by connecting the simulator directly to the pneumatic
system of a sphygmomanometer under test (SUT). The distinguishing feature is its ability to play back real-life oscillations

82, 024303-1

2011 American Institute of Physics

Downloaded 17 Mar 2011 to 192.53.103.200. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions

024303-2

Riedel et al.

Rev. Sci. Instrum. 82, 024303 (2011)

FIG. 1. Schematic view showing the interplay of recording unit, bloodpressure simulator, and sphygmomanometer.

recorded from human subjects under well-controlled clinical

conditions. The concept is illustrated in Fig. 1; it comprises
four procedural steps.
(i)

Acquisition of oscillometric blood-pressure data from

human subjects with a specifically developed recording
unit.10
(ii) Preparation of a simulator database containing all oscillometric data together with clinical reference values
obtained by auscultation.
(iii) Processing the recorded raw data to obtain the model
signal to be output by the simulator.
(iv) Simulation, i.e., application of the model signal to the
SUT.
For step (iv), the simulator is connected to a SUTs pneumatic system, as shown in Fig. 1. The SUT itself is set up
as normally, except that its cuff is wrapped around a rigid
metal cylinder instead of a human limb. Once a measurement
is started, the SUT creates its generic baseline-pressure curve;
frequently a rapid inflation followed by a slow deflation of the
cuff. The simulator detects the momentary baseline pressure,
looks up the best-fitting pressure oscillation in its database,
and reproduces it with a digitally controlled membrane pump.
This artificial oscillation on top of the baseline pressure is detected by the SUT and evaluated according to its internal algorithm. The resulting blood-pressure indication of the SUT
can be then compared to the reference value in the simulator
database.
Signal output is exclusively based on the prerecorded
data; within the concept of the simulator, it is not intended
to change any of the signal parameters in the field. This requires a considerable amount of datasets to be recorded but
eliminates handling errors during operation.
A detailed description of the simulators construction and
components will be given in Sec. IV. Also, the considerations
leading to a particular technical solution will be discussed.
III. OSCILLOMETRIC SIGNALS

In the following, a brief description of the aforementioned recording unit is given. The device in itself is of minor
importance for this paper, but some details on the oscillometric measurements may facilitate the understanding of the data
processing. Figure 2 shows a real-life blood-pressure measurement from the simulator database. Controlled by a personal computer, the cuff is inflated rapidly and then deflated
at a moderate rate of 23 mmHg/s. The cuff pressure is measured by a piezoresistive pressure transducer.11 Its analog sig-

FIG. 2. (a) Measured cuff pressure p, i.e., baseline pressure overlaid by oscillations from a human subject. The baseline pressure is deflated at a rate of
about 3 mmHg/s. The average heart rate is 78 beats per minute. (b) Pressure
oscillations p separated by subtracting a least-squares fit to the pressure
baseline. Insets: Regions around the systolic and diastolic pressures of this
signal.

nal is converted by a 16-bit ADC (Ref. 12) at a sampling rate

of 2 kHz and stored on the hard disk. The 16-bit dynamic
range is required to separate the small oscillations from the
large baseline pressure. Previous data acquisition systems13
only had 12-bit resolution and employed electrical high-pass
filters to separate the slow decrease from the oscillation signal. Due to this unrecoverable loss of information, 12-bit signals cannot be used to fully regenerate the measured data at a
later time.
The spectral content of the signal in Fig. 2(b) and its integral are shown in Fig. 3. Integration starts at 0.5 Hz to exclude residual near-dc signals which remained after baseline
subtraction. There is no noticeable content beyond 12 Hz, and
95% of the spectral content is below 8 Hz.
The systolic and diastolic reference blood-pressure values were determined simultaneously but independently by
two trained observers utilizing a double stethoscope for auscultation. If both observers agreed within 4 mmHg, the mean
value was used as reference value. Together with the acquired
cuff-pressure dataset, the observed blood pressure values and
the settings of the recording unit are stored in the database.

Downloaded 17 Mar 2011 to 192.53.103.200. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions

024303-3

Rev. Sci. Instrum. 82, 024303 (2011)

FIG. 3. Spectral content (solid line) and integrated spectrum (dashed line) of the oscillometric signal shown in Fig. 2(b). The fundamental heartbeat frequency
at 1.31 Hz is normalized to 100%. The integrated spectrum is normalized to 100% at 12 Hz.

IV. ARTIFICIAL OSCILLATION SIGNALS

A. The generation of pressure oscillations

To regenerate the recorded pressure data in the pneumatic system of a SUT, a pneumatic pulse generator with
high dynamical capabilities is needed. In principle, artificial
oscillations in the cuff could be generated by a model of the
human arm with a mechanical actuator inside. The actuator
could change the diameter or volume of the arm model. The
undefined volume and different elasticities of the various inflatable cuffs, however, would result in an undefined pressure
response at a given mechanical deformation. Therefore, we
opted to determine the momentary pressure in the cuff and
apply a feedback to the actuator in order to generate the desired waveform. The measurement of the baseline pressure is
performed by simply inserting a T-fitting into the tube connecting the cuff and the measurement unit (console) of the
SUT and attaching a pressure sensor.
At the low frequencies occurring here (below 12 Hz),
the pressure propagates in the whole pneumatic system of
the SUT without noticeable delay. It is sufficient, therefore, to generate the oscillations externally by a pneumatic
pump and feed them into the tubing of the SUT.13 The cuff
is wrapped around a rigid cylinder substituting the human
arm.
The diaphragm pump used to feed pressure oscillations
into the pneumatic system of a SUT is schematically depicted

in Fig. 4. The pressure chamber above the diaphragm is connected to the SUT tubing by means of a T-fitting which has
been inserted for the test.
The baseline pressure generated by the SUT exerts a considerable force onto the diaphragm. In order to withstand this
force, an electronically commutated, brushless dc-motor14
is used as an actuator. The torque of the motor is transferred to the diaphragm via a pulley-belt mechanism and a

FIG. 4. Schematic sketch of the blood pressure simulator. Three-way cocks

A and T are depicted in their simulation position, i.e., after all initial adjustments have been performed.

Downloaded 17 Mar 2011 to 192.53.103.200. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions

024303-4

Riedel et al.

lever. The stroke of the diaphragm is limited to a few millimeters in order to avoid any undue deformation. The pulley has to exceed a certain minimum diameter in order to
avoid an excessive bending of the belt and to keep the shaft
rotation well below 360 . These constraints determined the
choice of the appropriate lever ratio and suitable diaphragm
diameter.
The waveform of the oscillations is taken from the simulator database mentioned above and output by means of a
combined ADC/DAC-adapter.12 The DAC signal controls the
motor current via the servocontroller,15 operated in current
output mode. The torque generated by a dc motor with a permanent magnetic or constantly excited stator is proportional
to the armature current and independent of the speed (torque
motor).
Due to the small displacements, the transmission ratio
and the active area of the diaphragm are nearly constant. The
force actuated on the diaphragm and the resulting pressure
is, therefore, proportional to the motor torque, and ideally,
this should allow for a simple feedforward control. As the
force does not depend on the deflection or speed, the generated pressure oscillation does not depend on the cuffs volume
and the resulting baseline pressure: larger cuffs need larger
pumping strokes but the same motor current for a given pressure oscillation. At higher baseline pressures from the cuff,
the required stroke for a given oscillation amplitude becomes
smaller, but the force to be added to the inflation force remains
the same again.
In practice, however, this appealing approach to control
the oscillation pressure directly via the motor current and
without feedback was not feasible due to the disturbances
caused by mechanical friction, the stiffness of the diaphragm
under pressure load, and by stickslip effects of the motor
bearings. In addition, the transfer characteristics from the setpoint voltage to the output current were somewhat coarse due
to the pulse-width modulation scheme of the power amplifier.
Attempts to reduce these disturbances by applying mechanical dither brought no satisfying success. Therefore, a feedback scheme utilizing stroke-amplitude control to overcome
the perturbations was required.
The control scheme of the pneumatic pulse generator has
to superimpose the oscillations onto the baseline pressure in
the cuff. The inflation and subsequent deflation of the cuff
are not controlled by the simulator, but are generated by the
pump and the exhaust valve of the SUT. This baseline pressure exceeds the oscillation amplitudes by 2 orders of magnitude, not to mention the finer structures of the oscillation
signal. In the simulator, the motor current has to compensate the considerable force exerted by the baseline pressure
and in addition has to generate the small oscillations. A feedback control for the two tasks based solely on the signal of
the pressure sensor was not feasible due to the aforementioned friction and noise artifacts in the pressure signal. Furthermore, a feedback scheme requires a considerably higher
sensor bandwidth than the moderate frequency content of the
oscillations in Fig. 3. The well-separated time scales of the
slowly varying baseline pressure and the much faster oscillations allow the motor to be controlled by a common feedback
loop.

Rev. Sci. Instrum. 82, 024303 (2011)

The motor current compensating the baseline pressure

is controlled by sensing the position of the diaphragm.
The position signal is supplied by a conductive plastic
potentiometer16 attached to the motor shaft. The potentiometer is fed with a constant voltage; the position-dependent
wiper voltage is input into the ADC. The feedback loop is
closed by real-time software running on the personal computer. It controls the motor torque via the DAC output and the
servoamplifier.
The diaphragm tries to respond to the baseline pressure
by moving downward. The feedback counteracts this position
change. Roughly speaking, the position control holds the diaphragm near the center of the stroke range, regardless of the
baseline pressure. This feedback is slow enough not to interfere with the oscillation strokes.
The oscillations are added to the set-point signal of the
position control scheme. Since the stroke depends on the cuff
volume, its inner pressure, and the atmospheric pressure, a
self-calibrating feedforward principle has to be applied. When
the cuff is fully inflated, the diaphragm performs some initial
probing pump-strokes under control of the angular transducer.
The pressure generated by the probing strokes is measured
and a slight digital averaging is applied for noise reduction.
The ratio between pressure signal and angular displacement
is the calibration factor for the oscillation strokes at the
current baseline pressure. Multiple short strokes are applied
in order to mimic the timing of anatomic oscillations. This
is important due to thermal transients between adiabatic and
isothermal compressions in the cuff, which occur on the same
time scale as the oscillations. The fast pressure variations
are henceforth controlled solely by means of the position
sensor.
The output of the pressure sensor is digitized and
smoothed by software low-pass filter to give the momentary
pressure in the cuff. This quantity is used to synchronize the
oscillations of the model signal as described in Sec. IV B. In
addition, the momentary cuff pressure is needed to calculate
the stroke required to generate a given oscillation amplitude,
which is inversely proportional to the absolute pressure (internal plus atmospheric pressure). This dependency is accounted
for by the real-time software.
The initial probing strokes could trigger an irregular reaction of the SUT or could be disturbed by the beginning deflation of the SUT. To avoid this, the T-fitting is equipped with a
three-way cock, the T-cock in Fig. 4, blocking the connection
to the SUTs console during the probing strokes. The volume
of the cuffs was found to be rather constant once they had
been fully inflated. Therefore, the volume needs no further
software correction after initial probing.
Some recently developed sphygmomanometers measure
the oscillations during slow inflation and do not exploit the deflation phase at all. In these cases, an auxiliary pump inflates
the cuff prior to testing. This facilitates the initial stroke calibration as well as the settling of the cuff-fastener to achieve a
constant cuff volume. The auxiliary pump is connected to the
pneumatic system via the A-cock (see Fig. 4). The probing
strokes are performed after full inflation of the cuff, while the
T-cock blocks off the SUT. Afterward, the auxiliary pressure
is released again through the A-cock.

Downloaded 17 Mar 2011 to 192.53.103.200. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions

024303-5

B. Synchronization of the oscillometric signal with

the deflation of the SUT

The oscillation amplitude depends on the baseline

pressure from the cuff in a manner that is specific for each
subjectthis is the rationale of sphygmomanometry. For
any model signal in the database the baseline pressure was
released at a rate controlled by the recording unit during
the clinical data collection. A SUT will, in general, deflate
the cuff at a different rate. At a given baseline pressure, the
SUT expects a defined oscillation amplitude and waveform
to indicate the same blood pressure values as recorded. Blood
pressure measurements dowithin wide marginsnot depend on the deflation rate which is why this parameter varies
considerably across sphygmomanometers on the market.
Even stepwise deflation is practical. For the simulator it is
necessary, therefore, to output the recorded model oscillations
with a timing corresponding to the actual deflation rate of
the SUT. This cannot be achieved by simply shrinking or
stretching the time scale of the model signal, as this would
change the heartbeat rate and the oscillation waveforms. The
individual oscillation segments in the model signal have to be
treated separately, instead, such that neither their waveform
nor their frequency is compromised.
This requires a processing of the original blood pressure
recordings before they can be used for a simulation. The main
steps are as follows:
(i) separation of oscillation and baseline pressure;
(ii) segmentation of the oscillations.
The first step is achieved by subtracting a least-square fit
to the slowly varying baseline pressure. The remaining oscillations [see Fig. 2(b)] are then dissected into single oscillation
periods (model segments) at their minima. A residual skew in
the model segments is removed by subtracting the tangent at
both minima to facilitate the concatenation of segments. An
impression of deskewed and concatenated segments is given
in Fig. 6(b).
Each segment is tagged with its accompanying baseline
pressure, i.e., the pressure which has been exerted while the
respective model segment was recorded. The individual segments are stored in a dataset for the real-time control of the
test process; the original dataset itself is no longer required.
During the testing of a SUT, the momentary baseline
pressure exerted by the SUT is measured. The software
selectsby means of the pressure tagthe correct model segment for that baseline pressure of the SUT. The selected segment is fed to the diaphragmpump controller. After the current model segment has been completed a new measurement
of the baseline pressure is performed and the next matching
model segment is selected and output by the drive-controller.
This will be repeated with the simulated heartbeat rate until
the baseline pressure has reached a lower limit. The subsequent model segments are easily stitched to the previous
ones without discontinuities due to the mentioned subtraction
of the tangents.
The model signals are recorded with a low deflation rate.
This way an excess of model segments is available, facilitating the selection of the model segment closest to the momen-

Rev. Sci. Instrum. 82, 024303 (2011)

tary baseline pressure of the SUT. Before the segment amplitude is fed to the control software as a position set-point, it
has to be scaled to the varying transfer ratio between stroke
and oscillation pressure, as described in Sec. IV A.
V. PERFORMANCE TESTS

A prototype of the blood-pressure simulator was built

(Fig. 5) and several tests were conducted to evaluate its performance:

r comparing the generated pneumatic signal with electrical sinusoidal signals;

r comparing the generated pneumatic signal with

recorded human signals;

r checking the repeatability and reproducibility of the

generated signals;

r investigating the effect of different cuff volumes on the

generation of the pneumatic signal.
Tests performed with clinical sphygmomanometers gave
standard deviation for repeated measurements with a fixed
setting staying within 1.5 mmHg for 70% (within 2.0 mmHg
for 90%) of the displayed results. The standard deviation
of the individual generated pressure pulse was below 0.02
mmHg for all pulse amplitudes up to 4 mmHg. Arbitrary errors, intentionally imposed on the self-calibration of the simulator for different cuff sizes, produced accordingly shifted
results, thus proving that different cuff volumes are compensated correctly. The frequency response of the pneumatic
system and its dependence on the diaphragm material (see
Sec. IV A) were investigated at Dresden Technical
University.17
Figure 6(a) shows the recorded test cycle of a rather
quickly (7 mmHg/s) deflating sphygmomanometer with
oscillations overlaid by the simulator. The model segments
are taken from the signal in Fig. 2(b). Due to the faster
deflation, only a subset of oscillations is used and the overlaid
oscillations are less visible than those in Fig. 2(a). In the fast
inflation phase they are actually indiscernible. The model segments, selected and concatenated in real-time, are shown in
Fig. 6(b). In contrast to Fig. 2(b), the segments of Fig. 6(b)

FIG. 5. (Color online) Mechanical interior of the prototype, (A) motor, (B)
belt, (C) lever, and (D) bellow-coupled angular transducer. The right front
pole is removed for clarity. The part partly obscured by the bellows is the
pressure chamber.

Downloaded 17 Mar 2011 to 192.53.103.200. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions

024303-6

Riedel et al.

Rev. Sci. Instrum. 82, 024303 (2011)

ment, i.e., a clinical trial, more closely than any previously

existing instrument. Especially important in this context is its
capability to correctly reproduce pathological conditions and
artifacts, e.g., arrhythmia artifacts. The adaptive time scale to
adopt for different heart rates and the combination of real-life
signals with artifacts are both, to the best of our knowledge,
unique features of the device. In this particular respect the
new simulator even outperforms the clinical trial in assessing a given sphygmomanometer, as it is capable to play back
irregular or otherwise rare waveforms repeatedly and reproducibly, which is inherently impossible in a clinical study on
a limited cohort of human subjects.
After successful testing of the prototype, another five
units of the simulator were produced, one by PTB and four by
a commercial manufacturer19 which are now in use for clinical research in the Asian and European Hospitals.
Given that some benefits of this simulator cannot be met
by clinical trials, it is intended to introduce its utilization into
future editions of regulatory standards governing the testing
of sphygmomanometers. This will be a lengthy process, however, requiring considerable international effort.
The simulator reaches its technical limits when it comes
to the testing of sphygmomanometers with non-standard cuff
geometry, especially, with a so-called wide-range cuff. For a
given blood pressure a wide-range and a normal cuff generate different oscillation signals. This is taken into account by
the manufacturer. The testing of such wide-range cuff devices
would require the recording of a new set of model-signals for
the respective cuff.
FIG. 6. (a) Test cycle of an automated SUT connected to the simulator. (b)
Oscillation segments to be superimposed to the fast pressure inflation and to
the slower measuring deflation of the SUT.

show a nonzero average. This is compensated by the slow

position control described in Sec. IV A. Evidently, the
synchronization of the oscillation segments with the baseline
pressure works even during the very fast initial inflation phase
(+30 mmHg/s). This is advantageous as certain sphygmomanometers measure the oscillations coarsely during the inflation phase in order to adapt the maximum inflation pressure
to the actual blood pressure of the subject to be examined.
The indication of the SUT in Fig. 6 was 116 mmHg systolic pressure, 69 mmHg diastolic pressure, and 76 heartbeats
per minute. A second SUT with significantly slower deflation rate (3 mmHg/s) indicated 121 mmHg, 74 mmHg, and
78 beats per minute for the same model signal. Such discrepancies, especially, if human artifacts come into play, are well
known from detailed surveys18 of the behavior of commercial
SUTs.

VI. SUMMARY AND OUTLOOK

The basic concept of the blood-pressure simulator presented in this paper differs substantially from commercial devices for the testing of sphygmomanometers. With its use of
prerecorded human signals which can be flexibly combined to
a wide range of realistic oscillometric waveforms, this simulator emulates the only currently accepted reference measure-

ACKNOWLEDGMENTS

The blood pressure simulator and the recording unit

were developed and manufactured under the EU contract
G6RD-CTC-2002-0076 with participating institutions from
Germany, the Netherlands, Slovenia, and the United
Kingdom.
1 J.

A. Whitworth; World Health Organization, International Society of Hypertension Writing Group, J. Hypertens. 21, 1983 (2003).
2 L. A. Geddes, M. Voelz, C. Combs, D. Reiner, and C. F. Babbs, Ann.
Biomed. Eng. 10, 271 (1982).
3 The prevailing medical pressure unit is mmHg; permitted by the international system of units (SI). 1 mmHg = 133.3 Pa.
4 International Standard ISO 81060, Non-invasive sphygmomanometers,
Part 1 (2007); Part 2 (2009).
5 European Standard EN 1060-4, Non-invasive sphygmomanometers, Part
4 Test procedures (2004).
6 American National Standard ANSI/AAMI, SP10; Manual, electronic or
automated sphygmomanometers (2002).
7 E. OBrien, J. Petrie, W. Littler, M. de Swiet, P. L. Padfield, D. G. Altman,
M. Bland, A. Coats, and N. Atkins, J. Hypertens. 11, 677 (1993).
8 J. N. Amoore, J. Med. Eng. Technol. 17, 25 (1993).
9 E. Balestrieri and S. Rapuano, IEEE Trans. Instrum. Meas. IM-59, 2391
(2010).
10 Informations-Technologie
Babelsberg GmbH, Potsdam, Germany
(recordingunit.itbb.de).
11 PMP 4015; Piezoresistive pressure sensor, 10 PSI, 690 mbar, 517
mmHg. GE Measurements and Control Solutions, Billerica, USA
(www.gesensing.com).
12 APCI-3120; 16-bit-ADC, 14 bit DAC, PCI-Adapter. Addi-Data GmbH,
Rheinmnster, Germany (www.addi-data.com).
13 H. Gro, S. Mieke, and M. Ulbrich, J. Med. Eng. Technol. 20, 75 (1996).
14 EC 45; brushless dc-motor, 250 W. Maxon Motor AG, Sachseln, Switzerland (www.maxonmotor.com). The rated power is never applied due to the

Downloaded 17 Mar 2011 to 192.53.103.200. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions

024303-7
small angular speeds occurring here. The motors torque is needed to withstand the inflation pressure.
15 DES 50/5; four-quadrant, electronic commutating controller. Maxon
Motor AG, Sachseln, Switzerland (www.maxonmotor.com). A single quadrant controller would be sufficient here because the inflation pressure of the SUT loads the drive always in one direction and because the inertial braking forces never exceed this
preload.

Rev. Sci. Instrum. 82, 024303 (2011)

16 P2201

A502; plastic potentiometer. Novotechnik OHG, Ostfildern,

Germany (www.novotechnik.com). Lifetime is 107 cycles or 100 full days
of uninterrupted operation.
17 M. Mehner, B. Fischer, S. Mieke, R. Seemann, and F. Wonka, Biomed.
Tech. 52(Suppl.), 1569047551 (2007).
18 J. N. Amoore, Y. Lemesre, I. Murray, E. Vacher, S. Mieke, S. King,
F. Smith, and A. Murray, J. Hypertens. 26, 35 (2008).
19 B. V. Lode, Groningen, The Netherlands.

Downloaded 17 Mar 2011 to 192.53.103.200. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions