A Thermal Flow Sensor Based on Printed Circuit Technology in Constant Temperature Mode for Various Fluids

Abstract

We present a thermal flow sensor designed for measuring air as well as water flow velocities in heating, ventilation, and air conditioning (HVAC) systems. The sensor is designed to integrate the flow along the entire diameter of the pipe also quantifying the volume flow rate of the streaming fluid where the calorimetric principle in constant temperature operation is utilized as a readout method. In the constant temperature mode, a controller keeps a specific excess temperature between sensing elements at a constant level resulting in a flow dependent heater voltage. To achieve cost-effective sensors, the fabrication of the transducer is fully based on printed circuit board technology allowing low-cost mass production with different form factors. In addition, 2D-FEM simulations were carried out in order to predict the sensor characteristic of envisaged setups. The simulation enables a fast and easy way to evaluate the sensor's behaviour in different fluids. The results of the FEM simulations are compared to measurements in real environments, proving the credibility of the model.

Keywords: CT mode; HVAC systems; calorimetric; printed circuit board; thermal flow sensor.

Figure 1

Image of the sensor layout with a meander length $l_{\mathrm{m}}$ of 250 mm. The left inset shows the copper leads (dark stripes) in detail: In the middle there is the heater and above and below there are the four sensing leads; $g_{\mathrm{I}}$, $g_{\mathrm{O}}$, and w denote the geometrical parameters (see Table 1). The right zoom section shows the interconnection area of the sensor with holes for alignment pins and screws. The substrate is coloured beige and the solder resist is coloured green. On the left end, there is a hole for a bracket to fasten the sensor to the pipe.

Figure 2

FEM model cross-section. Heating and sensing elements (copper) are placed on top of a substrate (FR4 glass epoxy), passivated by a solder resist (polymide).

Figure 3

Results of the FEM simulation (for air as fluid) around the sensor. The colour indicates the temperature distribution for a heating power of 1 W while the blue arrows indicate the flow velocity field at a velocity of 1 m/s. Due to convection, the temperature profile is shifted downstream.

Figure 4

Simulated heating power versus the the flow velocity for three preselected temperature differences ($\Delta T$) with air as test fluid.

Figure 5

Simulated heating power versus the the flow velocity for three preselected temperature differences ($\Delta T$) with water as test fluid.

Figure 6

(a) setup in the office building with the position of both PCB and reference sensor. The flow is regulated via two flaps, differential pressure sensors ($\Delta p$), and an exhaust fan. The test flow channel is a merging pipe from two offices with a diameter (d) of 250 mm; (b) schematic view of the electronics. Four amplifiers evaluate the sensing lead voltages. Then, the four signals are summed according to Equation (1). The resulting signal is the command value for the PI-controller, which regulates the voltage of the heating lead to maintain the desired excess temperature (configured via the offset voltage).

Figure 7

Measurement setup where the PCB sensor is mounted across a pipe. A reference flow sensor is located 750 mm downstream of the PCB sensor. The signals from both sensors are recorded with an ADC module which sends the data via WLAN to a central server.

Figure 8

Mounting of the sensor in the flow channel via two slots. (a) 3D printed part (green) for tip of the sensor with a screw for tightening. (b) Sensor locked on the linkage PCB. The sensor is placed on a female connector and tightened via a 3D printed part (green).

Figure 9

Measurement of the PCB sensor in constant temperature mode compared with the simulation. The heater voltage is plotted as a function of the mean flow velocity where the error bars indicate the error ($\pm 0.5$ m/s + 7% from reading).

Figure 10

Long-term measurement where the sensor was tested in a normal operation mode of the HVAC system that was controlled by an automatic system. PCB and reference sensor were measured every 10 s during three days. The scaling factor for the PCB sensor is given in Equation (2).

Figure 11

Start-up detail (timely zoomed in) of a measurement (similar to Figure 10). The PCB sensor recognises small flow velocities (at $2.5$ min) before the reference does. Both sensors need the same settling time before an accurate value is achieved. Detailed information about the response property can be found here [12].

Figure 12

Picture of the water experimental setup. The sensor is stretched across a PVC water pipe (inner diameter of $21.2$ mm), located at 135 cm at a strait pipe which is 20 cm before a bend, and connected to the controller by an adapter board. The volumetric flow is controlled via two separate water pumps and measured with two ultrasonic flow sensors. From a control cabinet, all sensors and actuators are controlled.

Figure 13

Simulated and measured heater voltage depending on the flow velocity in constant temperature mode where the temperature difference was set to $0.5$ $\mathrm{K}$.