Smart Asthma Inhaler

Product Idea

Smart Asthma Inhaler is one of the most innovative projects that I have worked on. While working at Tritek Micro Controls, I handled the complete development of the electronics of this inhaler. The smart inhaler kept a track of the time and the accuracy of the last 1000 doses. Its main function was to detected if the patient is taking the medication correctly. The correct way to take medicine via an inhaler involves 3 steps - shake, inhale, and press for dose release. This smart inhaler would capture all the steps and note whether the dose taken was taken correctly. It then sent this information to a phone app via Bluetooth (BLE). 

Constraints

When working on this device, there were three major constraints. The first one was the price of the device. The cost of the electronics component of the device could not be more than INR 900.00 (approximately, $11.00). The other major constraint was the size of the device. The entire electronics portion of the device had to be accommodated inside the casing of a traditional asthma inhaler. A small hand held device meant a limited room for the PCB, battery, charging circuit, etc. The third constraint was the  battery life. To make the product viable, a longer battery life was needed. Many innovative solutions were needed to make sure we fit into these constraints.

Innovation

Detecting the inhalation of the medicine was one of the key challenges that had to be addressed. We designed a unique mechatronics solution to capture the inhalation correctly. We added a small flap inside the plastic casing of the inhaler. When the patient inhaled, the flap moved away from the PCB towards the patient’s mouth. When the patient stopped inhaling, the flap moved back to its initial position. We added a small aluminium foil to this plastic flap, and a proximity sensor to the PCB. The proximity sensor could detect the movement of this foil away from the PCB and towards the patient. We had to make this proximity sensor fit the size and the cost constraints.

An inductive proximity sensor works with two connected inductors, one is the sensing inductor and other is the reference inductor. The proximity sensor detects the inductance of the connected inductors by passing a current through them. When another metal object is close to the sensing inductor, its inductance value changes. The difference between the sensing and the reference inductance is detected by the proximity sensor.

Usually an inductive proximity sensor is used with a physical coil inductor, or an inductor circuit is added to the PCB. Because we had severe size constraints, these options were not available to us.  So I decided to use SMD inductors. SMD stands for surface mount device. Such devices are small, flat, and can be soldered directly on the PCB. I gathered as much information as possible from online forums and the manufacturer’s website. I also carried out a small experiment by soldering all needed components of the sensing circuit to a small general-purpose PCB. However, during prototype testing I discovered that only half the devices worked as expected.

Further investigation revealed that this was due to calibration issues. Fixed calibration had been used for every unit. After sampling the inductance values of SMD inductors, I discovered that inductors had different inductance values even if they were from the same batch. This was due to the tolerance that inductors have. Each inductor has a value L ± tolerance. Due to this small difference in values, fixed calibration was not working. I posted this difficulty on the manufacturer’s forum and received a recommendation of using a slightly higher value for reference inductor. This change partially solved the problem. Now 75% devices were working as expected. To make sure that all devices work perfectly, I decided to perform dynamic calibration. I calibrated devices at runtime. 

When a device is waking up, the inhaler flap would be in its steady state, that is, it will be closer to the sensing inductor. The proximity sensor would return 0×0 if the calibration is correct. In the dynamic calibration protocol, the device looped through 10 different calibration values during the wake up process. The first value where the proximity sensor transitioned from 0x1 to 0x0, was chosen for that device. This solution gave us 100% correct results. For all the devices we were able to detect inhalation correctly. There were no false positives / negatives. For this purpose, I used the analog output pin of the microprocessor.

Process Changes

Due to the small size of the PCB and price constraints, we had to focus on reducing the overall footprint of the PCB. This required the use of SMD components. Most of these components were of the BGA type (Ball Grid Arrays - where pins are at the base of the chip instead of the sides).

Manufacturing a PCB with BGA components was new to our business at the time. To make sure we could manufacture and test the prototypes properly, I did research on how these types of products are manufactured and tested. We purchased an SMD soldering machine and I setup the flow in the factory for proper soldering of the PCBs. This involved educating the staff on how to use the machine, how to place stencils properly to apply solder paste, how to place components, and then run them through the machine.

To make sure testing was glitch free and accurate, I identified the test points and designed a small test jig. I also developed a software program for testing. We could load this program on the device and with the test jig we could test the functioning of all the components of the device. The tests covered the functioning of the individual components like the proximity sensor, accelerometer, LCD, battery, the charging circuit etc.

Other small innovations and simplifications include, how sync between the inhaler and phone would take place, how a shake will be detected and how code and circuit was designed for optimization of power consumption.