flowing, for example (from left to right), the hot air
is blown from left to right, and thus the sensor sees
a higher temperature than the temperature sensor
at the left side. Since the temperature difference is
proportional to the flow rate (linear at low flow rate and
non-linear at high flow rate), we can determine the flow
rate by measuring the temperature difference of the two
In one implementation of Thermal MEMS flow
measurement, the ACEINNA MDP200 sensor, MEMS
sensor, and signal conditioning circuit is built together
monolithically on a single chip by utilizing standard
materials in the CMOS process to build the MEMS
flow sensor. In the CMOS process, polysilicon and
aluminum are readily available and commonly used for
interconnects. In a Thermal MEMS sensor, polysilicon
acts as the heater resistor, while the polysilicon and
aluminum contact each other to create a thermocouple-based temperature sensor.
With this modern approach, a MEMS flow sensor
can be built without needing any special materials
and processes. The signal conditioning circuits can be
naturally integrated with the sensor on a single chip.
Such monolithic integration enables smaller size,
lower cost, higher precision, and control of the sensor.
With integrated CMOS circuit, one can monitor the
temperature difference down to micro-Kelvin allowing
very high resolution and low flow rate sensing. The
resulting MDP200 device and its internal sensor is shown
below in Figures 2 and 3.
Figure 2: Packaged flow sensor. Figure 3: Sensor and ASIC.
Low-cost semiconductor flow sensors can be simply
connected to any standard microcontroller with I2C
interface. Software integration can be completed in a
couple of hours. Low-power consumption of less than
10 mA at 3. 3 V is also generally acceptable. In battery-powered applications, the device can be easily duty-cycled
because power-up and measurement time is less than 0.1
When selecting and specifying a Thermal MEMS Flow
sensor there are a couple more subtle considerations:
1. Gas Type
3. Connection Mechanism and Altitude Correction
A flow sensor returns a value proportional to flow;
hence accuracy is modeled in classic terms with zero-
point (offset) and span (linearity) errors.
There are two ways to determine if the accuracy of
a flow or differential measurement device is suitable
for your requirement. The first and simplest way is to
consider the manufacturer’s total accuracy specification.
This combines the initial offset and span errors, along
with the change in these errors over temperature and
other subtle effects.
If this error turns out too large for the intended
application, then it is often useful to consider the
individual errors separately. In many applications,
initial offset error can be removed once the device
is installed, reducing the total error budget. In other
applications, the required range may be less than the
device specification, hence the contribution of span
error may also be less than included in the total error
Finally, how to connect the flow sensor into the
flow stream is often an interesting challenge. Instead
of inserting the device directly in the flow stream, a
bypass flow circuit can be used. The Figure 4 diagram
below shows how to properly set up a pressure
drop circuit, which will force a portion of the flow
stream through the device. Such flow drop circuits
can be custom made or provided by the flow sensor
Figure 4: A bypass flow circuit can measure flow inside a tube.
Another issue is that the DP sensor can require
altitude correction for precision applications. Altitude
correction is typically accomplished by adding a
barometer to the end device. Manufacturers can usually
provide an appropriate altitude correction formula. In
lower precision applications, this compensation is not