Gradient Amplifier Design with Advanced Power Semiconductor Devices and Simple Topology
Ruxi Wang1, Juan Sabate1, Eladio Delgado1, Xiaohu Liu1, and Fengfeng Tao1

1GE global research center, Niskayuna, NY, United States

Synopsis

Gradient Amplifier is one of the key components in a MRI system to supply the gradient coil with large current (>1000A) and high voltage (>2000V) to achieve strong gradient field and fast slew rate. In this paper, a high efficiency two H-bridge in cascaded gradient driver design with 1700V SiC MOSFET is designed and fabricated. The power module switching frequency is 31.25 kHz and amplifier total output ripple frequency is 125 kHz. The amplifier loss and efficiency (around 99.5%) is verified in experimental results.

Introduction

Gradient Amplifier is one of the key components in a MRI system to supply the gradient coil with large current (>1000A) and high voltage (>2000V) to achieve strong gradient field and fast slew rate. In addition, extremely high fidelity for reproducing the current command from the central system is very critical for imaging quality. Therefore, high power and high band width gradient amplifier is desired for high performance MRI system [1-4]. This paper presented a high performance modular solution gradient amplifier system with advanced wide-band gap (Silicon Carbide) devices.

Power Stage Evolution and Challenges

Fig.1 (a) shows the existing gradient amplifier topology with silicon devices in a stacked structure. In order to meet high power and high bandwidth capability, two middle bridges powered with lower voltage are connected in parallel with phase shift to output higher ripple frequency. As indicated in Fig.2, the new wideband gap devices made of materials such as SiC provide a much higher bandgap, breakdown field, thermal conductivity which offer them the potential capability to overcome the frequency, temperature and power management limitations of traditional silicon devices. Fig.1 (b) shows the improved two stages structure with high voltage (1700V) GE SiC MOSFETs. The modular structure brings the benefit of higher components volume, simpler implementation and lower lost.

Silicon Carbide (SiC) Gradient Driver

The two bridges SiC based gradient amplifier as shown in Fig.1 (b) is fabricated and tested. Each bridge is powered with isolated 1050V. The load is typical MRI gradient coil. Fig.3 shows the power loss test schematic. The power stage loss is measured through the liquid coolant temperature change with the assumption that most of the loss is dissipated through the liquid cooled cold plate. The coolant pipe is connected in series and four Resistance Temperature Detectors (RTDs) are embedded inside the coolant to measure the inlet and outlet temperature for cold plate 1 and cold plate 4. Then, the temperature data is documented as shown in Fig.4 through data acquisition. The total calculated loss is around 1846.15W at the tested operating point which is almost 4~5 times lower compared with baseline structure using silicon IGBT module as shown in Fig.1 (a).

Conclusion

In this paper, the gradient amplifier power stage topologies evolution and challenges are discussed. Modular cascaded structure with SiC devices enables simple architecture with lower engineer and implementation cost without sacrificing the high bandwidth and high power capability. A high efficiency two H-bridge in cascaded gradient driver with GE 1700V SiC MOSFET is designed and fabricated. The amplifier power stage loss is around 1846W which is verified in experimental results. The SiC gradient amplifier power loss is 4~5 times lower compared with silicon version.

Acknowledgements

No acknowledgement found.

References

[1] R. Wang, et al., Pittsburg, ECCE, 2014. [4] J. Sabate, et al., Proc. ISMRM. 2007. [3] R. Wang, et al., Pittsburg, ECCE, 2015. [4] R. Wang, et al., Pittsburg, EPE, 2015.

Figures

Fig.1 Gradient amplifier topology evolution and challenges

Fig.2 Comparison between Silicon and Silicon Carbide

Fig.3 Power loss test schematic

Fig. 4 Temperature measurements through the data acquisition



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
3623