Our Design Process: A Quick Overview
Starting from the requirements, we held a series of meetings with our Advisors who suggested various options to explore for hardware and software needs. Given the trail blazed by MIT & MPS prototypes, some decisions were relatively easy to make:
- Using an Arduino as the microcontroller for the AutoBVM.
- Using a system of gears to exert rotational force, resulting in a smaller device (for portability) and better duty cycle (reliability).
- Utilize encoder on the motor to granularly control the speed and range of motion of the respiration phases. This was one of key learnings (MIT project).
- Implement the same user interface recommendation made by the MIT project (user controls for BPM, Tidal Volume, I/E ratio and lung pressure).
By leveraging the existing state-of-the-art, we were able to make rapid progress on many aspects of the design and implementation. There were some requirements unique to the Indus Hospital use case that necessitated innovation and/or deviation from the original design. As a result, we:
- Reduced the physical size by making efficient use of the space within the device. The Auto-AmbuBag device is significantly smaller than both MIT and MPS devices.
- Incorporated a power management and battery charging module to enable switching over the battery in case of power outages.
- Switched to a 12V ecosystem to enable charging and operating the device from the power outlet of an ambulance during patient transportation.
- Conserved power by using a 9 N-m motor (lower power than the 15 N-m minimum for COVID patients), a non-touch screen display with aggressive brightness controls, and removing the need for any active cooling.
- Added an SpO2 sensor that comes with the Auto-AmbuBag to monitor Oxygen saturation, displayed on the screen for the clinician.
The modifications above made it possible to take advantage of existing open-source knowledge where possible, while adjusting the design for the unique needs we had to meet.
Entry 1: Power System
Battery used in the design
To design a power system that could run continuously for 4+ hours when plugged in and 2+ hours on battery.
- Considered a 24V motor for efficiency, but rejected it as it would require a boost converter from 12V input, introducing complexity. We used 12V motors which efficiently manage the required load.
- Selected a low power, 2.4 inch LCD display with configurable backlight. With aggressive backlight control, we reduced the consumption to approximately 0.05Ah, which is negligible.
- For typical Indus configuration (15 bpm, Tidal vol of 750 & I:E of 2), the device drew less than 1 A of power hourly. Local testing used latex balloons to simulate lung resistance.
We picked an off-the-shelf 4-cell and 6Ah Lithium-Ion battery that is easily procured and replaceable in Pakistan. We are confident we will comfortably exceed the 2-hour battery operation time. For future iterations, we are considering a different battery (8Ah, 4000+ cycle life, $18) for improved capacity and cost-efficiency.
Entry 2: The Motor - The Heart of the system
goBILDA Yellow Jacket motor with encoder
To generate controlled torque to squeeze the AmbuBag while operating in limited power conditions or during transport. This requires a high torque, low revolutions per minute (rpm) motor.
- Attempted a high-rpm brushless DC motor, but it required a complex 100:1 planetary gear and additional gearing, making the device too large.
- Tested multiple low-rpm stepper motors and various drivers. The stepper motors consistently skipped under load and proved difficult to control, and were deemed unsuitable.
- The TorqueNADO MAX motor (5 N-m torque) lacked the necessary torque at higher respiration settings.
- Selected the goBILDA Yellow Jacket DC motor with 9 N-m torque at 84 rpm speed, which performed perfectly across all configuration ranges and includes a built-in encoder.
We picked the goBILDA Yellow Jacket DC brushed motor. This motor met all our needs of low rpm, high torque, encoder support for closed-loop operation, and low cost ($55). The goBILDA ecosystem provides easily available mounts and drivers, resulting in a reliable and well-supported core component.
Entry 3: User Interface (UI) & Controls
From a simple potentiometer to the final LCD and encoder interface.
To create a simple, intuitive, and power-efficient interface for clinicians to view and precisely set ventilation parameters (BPM, Tidal Volume, I/E Ratio, Pressure, User messages).
- Large Touch Screen: Rejected due to high cost and power-intensive nature, which conflicted with the battery-operation requirement.
- Multiple Potentiometers: Rejected because they were imprecise and lacked the required numerical feedback for clinical parameter setting.
- LCD Screen with Rotary Encoder: This option offered the best balance of cost-effectiveness, low power consumption, and precise digital control through a menu system.
We selected a 16x2 LCD screen with a single rotary encoder and push-button. This interface was the most robust and cost-effective, allowing me to program a simple menu where clinicians could precisely set and confirm each ventilation parameter, meeting the hospital's exact usability requirements.