3–188
Motorola Sensor Device Data
For More Information On This Product,
Go to: www.freescale.com
! !# !
" "
Prepared by: Carl Demington
Design Engineering
INTRODUCTION
This application note describes a technique to improve the
linearity of Motorola’s MPX10 series (i.e., MPX10, MPXV10,
and MPX12 pressure sensors) pressure transducers when
they are interfaced to a microprocessor system. The
linearization technique allows the user to obtain both high
sensitivity and good linearity in a cost effective system.
The MPX10, MPXV10 and MPX12 pressure transducers
are semiconductor devices which give an electrical output
signal proportional to the applied pressure over the pressure
range of 0–10 kPa (0–75 mm Hg). These devices use a unique
transverse voltage–diffused silicon strain–gauge which is
sensitive to stress produced by pressure applied to a thin
silicon diaphragm.
One of the primary considerations when using a pressure
transducer is the linearity of the transfer function, since this
parameter has a direct effect on the total accuracy of the
system, and compensating for nonlinearities with peripheral
circuits is extremely complicated and expensive. The purpose
of this document is to outline the causes of nonlinearity, the
trade–offs that can be made for increased system accuracy,
and a relatively simple technique that can be utilized to
maintain system performance, as well as system accuracy.
ORIGINS OF NONLINEARITY
Nonlinearity in semiconductor strain–gauges is a topic that
has been the target of many experiments and much
discussion. Parameters such as resistor size and orientation,
surface impurity levels, oxide passivation thickness and
growth temperatures, diaphragm size and thickness are all
contributors to nonlinear behavior in silicon pressure
transducers. The Motorola X–ducer was designed to minimize
these effects. This goal was certainly accomplished in the
MPX2000 series which have a maximum nonlinearity of 0.1%
FS. However, to obtain the higher sensitivity of the MPX10
series, a maximum nonlinearity of
±
1% FS has to be allowed.
The primary cause of the additional nonlinearity in the MPX10
series is due to the stress induced in the diaphragm by applied
pressure being no longer linear.
One of the basic assumptions in using semiconductor
strain–gauges as pressure sensors is that the deflection of the
diaphragm when pressure is applied is small compared to the
thickness of the diaphragm. With devices that are very
sensitive in the low pressure ranges, this assumption is no
longer valid. The deflection of the diaphragm is a considerable
percentage of the diaphragm thickness, especially in devices
with higher sensitivities (thinner diaphragms). The resulting
stresses do not vary linearly with applied pressure. This
behavior can be reduced somewhat by increasing the area of
the diaphragm and consequently thickening the diaphragm.
Due to the constraint, the device is required to have high
sensitivity over a fairly small pressure range, and the
nonlinearity cannot be eliminated. Much care was given in the
design of the MPX10 series to minimize the nonlinear
behavior. However, for systems which require greater
accuracy, external techniques must be used to account for this
behavior.
PERFORMANCE OF AN MPX DEVICE
The output versus pressure of a typical MPX12 along with
an end–point straight line is shown in Figure 1. All nonlinearity
errors are referenced to the end–point straight line (see data
sheet). Notice there is an appreciable deviation from the
end–point straight line at midscale pressure. This shape of
curve is consistent with MPX10 and MPXV10, as well as
MPX12 devices, with the differences between the parts being
the magnitude of the deviation from the end–point line. The
major tradeoff that can be made in the total device
performance is sensitivity versus linearity.
Figure 2 shows the relationship between full scale span and
nonlinearity error for the MPX10 series of devices. The data
shows
the
primary
contribution
nonproportional stress with pressure, while assembly and
packaging stress (scatter of the data about the line) is fairly
small and well controlled. It can be seen that relatively good
accuracies (<0.5% FS) can be achieved at the expense of
reduced sensitivity, and for high sensitivity the nonlinearity
errors increase rapidly. The data shown in Figure 2 was taken
at room temperature with a constant voltage excitation of
3.0 volts.
to
nonlinearity
is
SEMICONDUCTOR APPLICATION NOTE
REV 3
F
Freescale Semiconductor, Inc.
n
.