Magnetically Levitated Blood Pump Impeller for Life Support

Minkyun Noh, MIT

"Prototyping mechatronic systems typically requires multiple iterations between ideation and implementation across hardware and software. The high flexibility of the LabVIEW development environment and NI hardware helped us expedite the prototyping phase."

- Minkyun Noh, MIT

The Challenge:

Prototyping a bearingless (magnetic-levitation) blood pump that requires integrating various types of analog and digital devices, such as power amplifiers and digital sensors. Magnetic-levitation control, due to its open-loop unstable nature, especially requires fast and deterministic loop timing to realize robust feedback stabilization.

The Solution:

Using CompactRIO and myRIO solutions to integrate all analog and digital devices and achieve the deterministic loop timing required to prototype the bearingless blood pump control system.

MIT-Ension Collaboration for Innovative Bearingless Blood Pumps

Conventional blood pumps, with impellers driven on mechanical shafts and bearings, are prone to stress and heat concentration on the shaft-bearing contact areas. This increases the level of hemolysis and thrombosis. Recently, engineers have applied magnetic levitation technology to the new generation of blood pumps, mainly due to advantages from non-contact operation. However, the high cost of the technology limits wide clinical adoption.



The Precision Motion Control Lab at MIT and Ension, Inc. are collaborating to develop a new type of bearingless blood pump for extracorporeal life support (ECLS). An ECLS system uses a disposable blood pump module for each patient each time to prevent infection. Therefore, the unit price of the blood pump module is directly reflected to the overall medical service fee. We are developing a new bearingless blood pump technology that can reduce the unit price of the disposable module, as well as vibration and power consumption.


Bearing-Free and Magnet-Free Pumps

The impellers of bearingless pumps are typically mounted with permanent magnets to generate levitation forces and a driving torque. This is a repeated cost, since the permanent magnets are wasted with the disposable pump module each time. In our new bearingless pump (Figure 1), we replaced the permanent magnet with a D2 steel ring to reduce cost. As the stator applies controlled magnetic flux on the rotor, levitation forces are generated through reluctance coupling and driving torque is generated through hysteresis coupling. We have presented the operating principle and the proof-of-concept motor in Figure 1, in which we used optical sensors to measure the position of the rotor. Currently, we are improving the prototype system toward a pumping test with water, and then an in vitro blood pumping test. A key challenge to this end is to design and implement a new sensing system that can measure the rotor position through a blood channel.



NI myRIO Interfacing With Digital Eddy Current Sensors

We replaced the previous optical sensors with eddy current sensors because they can measure the rotor position through a blood channel. Figure 2 shows the prototype system, which we equipped with digital eddy current sensors called inductance-to-digital converter (LDC) from Texas Instruments. The LDCs drive sense coils to measure the distance to a conductive target. We placed four sense coils, which four LDCs drive individually, in the inner air gap to measure the radial positions of the steel ring. We used a myRIO device to interface with the LDC digital sensors. The myRIO includes a total of 40 digital I/O channels with 3.3 V output and 3.3 V/5 V-compatible input, which provides voltage compatibility to a wide range of third-party digital chips. In addition, we can reconfigure all the digital I/O channels through FPGA programming. These offer a flexible interface with peripheral chips to save time and effort, especially during a prototyping phase.


We programmed the myRIO FPGA with LabVIEW to implement four Serial Peripheral Interface (SPI) ports for communicating with the LDC digital sensors. Figure 3 shows the schematic diagram of the sensing system. We used two pairs of LDC digital sensors to differentially measure the x- and y-position of the rotor. We connected each LDC to the myRIO with five digital I/Os—four for SPI communication and one for a reference clock.



We downloaded the SPI IP by NI (v1.3.0.1) from VI Package Manager and modified it to implement additional functionalities such as taking the sensor data at a hardware-timed fixed sampling rate, computing the x- and y-position estimates, and sending out the position estimates through two analog output channels for the CompactRIO to read. As a result, the myRIO together with the LDC digital sensors turns into a programmable two-channel analog sensor.





CompactRIO System Runs the Time-Critical Levitation Control Algorithm

Our prototype bearingless motor includes twelve-phase windings, driven by lab-made linear transconductance power amplifiers. We used a cRIO-9076 controller mounted with three NI 9263 modules and one NI 9205 module. This configuration delivers a total of 12 analog output channels, which can individually drive the twelve-phase stator windings. Such redundant degrees of freedom allow testing diverse control algorithms during a prototyping stage. Figure 4 shows the control system overview.


We implemented the feedback control algorithm on the FPGA part of the controller. We did this because, due to its open-loop unstable nature, the magnetic levitation control especially requires fast and deterministic loop timing to realize robust feedback stabilization. Missing a single time step can deteriorate the stability of the closed-loop system significantly. Based on a benchmark test result, the CompactRIO FPGA can execute our control algorithm at up to a 46 kHz loop rate, which is more than enough for most mechanical system controls. We set the actual loop rate to 10 kHz and tune the feedback controller to achieve a cross-over frequency of 130 Hz. We used the real-time processor side of CompactRIO for data logging and user interface.


LabVIEW and NI Hardware for Prototyping Mechatronic Systems

Research conducted in the Precision Motion Control Lab at MIT focuses on new types of mechatronic systems, including the bearingless blood pump described above. Prototyping mechatronic systems typically requires multiple iterations between ideation and implementation across hardware and software. The high flexibility of the LabVIEW development environment and NI hardware helped us expedite the prototyping phase. In particular, the reconfigurable I/O (RIO) architecture and FPGA-programming capability allows a larger design space for the mechatronic system designer. Also, we used LabVIEW trainings online (such as LabVIEW Real-Time 1, LabVIEW Real-Time 2, and LabVIEW FPGA) as trustworthy sources for students and researchers in our lab.


What’s Next

Our short-term goal is to perform a water pumping test. Then we plan to improve system performance, such as maximum torque and speed, toward an in vitro blood pumping test. Our long-term goal is to incorporate this new blood pump into Ension’s existing portable cardiopulmonary assist system (pCAS); thereby, developing an artificial heart-lung combination as a single disposable module. We believe this new technology can significantly increase the clinical adoption rate of magnetic-levitation blood pumps, thereby enabling more patients to enjoy high-quality medical services at low cost.


Author Information:

Minkyun Noh
77 Massachusetts Avenue
Cambridge, MA 02139
Tel: (617) 258-6098

Figure 1. CAD Model of Our Bearingless Blood Pump
Figure 2. Prototype of Bearingless Motor
Figure 3. Block Diagram for Sensing System
Figure 4. Prototype of Bearingless Motor