Considerations for Design of ASIC of Thermal Flow Sensors

Xiang Zheng Tu

Thermal mass flow sensors are manufactured using MEMS (Micro-Electro-Mechanical Systems) technology. The sensor chip comprises of a thermally isolated pad supporting a heater and one or two thermopile(s), all integrated in a silicon substrate. Heater temperature is typically controlled to be several degrees above ambient temperature. Thermal flow sensors operate on the principles of heat transfer across the surface of the sensing element. The upstream sense thermopile is cooled, the downstream sense thermopile is heated, and the combined differential electrical signal is proportional to flow.

The thermal flow sensors enable the ASIC to fulfill the basic market requirement for the thermal flow sensors: low-power, low-cost, able to run on battery, and with automated meter reading. The main attraction of this design is that the flow sensing module of the ASIC keeps running even when the ASIC goes into low-power mode. Since the ASIC is in low-power mode for most of the time, it reduces the power consumption.

The thermal flow sensors allow the ASIC to support a battery driven power supply and be capable of time keeping. It senses the signals from the flow sensor, calculates the flow and then accumulates it. The total flow accumulated and the month wise profile of the flow are stored and updated in the memory. The user key available on the board can be used to display the flow accumulated in a month and the date on the LCD. The ASIC also supports wireless communication with another handheld device. Thus, the ASIC supports a user deriving the flow readings using a handheld device from a distance.

For operating the thermal flow sensor a voltage reference is needed. The voltage reference is a DAC output of the ASIC, which is generated by a modulated bandgap voltage reference. The heater of the sensor is heated by an additional DAC output of the ASIC which is generated by modulating a regulated voltage. So calibration and correction of the sensor can be achieved by varying the offset and gain of a programmable-gain-amplifier and by varying the sensor heater excitation current or voltage.

The offset of a CMOS amplifier is usually in the order of 1mV and can be reduced only by increasing the area of the CMOS devices. Almost the same is true for the 1/f noise of the amplifier. It is preferred to utilize dynamic offset cancellation techniques, such as Auto-zeroing and chopping. This technique can reduce the offset to the microvolt level, while also removing 1/f noise. The offset cancellation is done in two phases a sampling phase and an amplification phase. During phase 1 the input signal is disconnected and the input of amplifier is connected to ground. So during the amplification phase the offset is subtracted, resulting in an output voltage free from offset.