生醫訊號感測系統之低雜訊低功耗類比前端電路設計

Low-noise and Low-power Readout Circuits for Biomedical Applications

李峯文(Feng-Wen Lee)

指導教授 ： 林宗賢

國立臺灣大學/電機資訊學院/電子工程學研究所/碩士(2014年)

https://doi.org/10.6342/NTU.2014.02956

摘要

生理訊號感測在醫學臨床診斷上具有重要意義。以往生理訊號的量測皆須仰賴大型機台，拜科技進步之賜，現今的半導體工業已允許我們將整個系統微小化並實作於一個電路板，甚至是單一晶片上。將整個感測系統包含前端電路，類比數位轉換器，數位訊號處理器以及無線傳輸/接收模組全部整合在一個晶片內，是近幾年國際間熱門的研究課題。本論文著重於探討應用於此系統之類比前端電路設計。此電路必須將極其微弱且低頻的生理信號放大，去除來自外界與電路本身可能的雜訊干擾，同時亦要求低功率以滿足可攜式需求。
本論文實作並量測三個不同架構的晶片。第一個架構為一個四通道儀表放大器，利用時間分工技術達到僅使用一個電容回授式儀表放大器達到可同時偵測四組不同訊號的功能，然而時間分工技術雖然可以大幅降低晶片面積及功率消耗，但其非連續時間的特性會將儀表放大器的輸出雜訊疊回低頻，導致訊號品質下降。此架構採用台積電0.18微米製程，每個通道面積消耗為0.015平方毫米，在1.8伏特的電源供應下，每個通道平均消耗4.25微安培的電流，並達到11.56的雜訊效率指標(NEF) 。
第二個架構實現一個雙通道儀表放大器利用正交頻率調變技巧。此電路會先將訊號做正交調變，以利於訊號在進行放大過程不會互相干擾，並且在解調後可成功將訊號分離。正交頻率調變技巧為連續時間操作，所以不會發生雜訊疊回低頻的現象，且此技巧可將閃爍雜訊載到高頻上，在利用低通濾波器可有效的去除，讓載低頻的訊號有良好的訊號品質。此架構採用台積電0.35微米製程，每個通道面積消耗為0.035平方毫米，在3伏特的電源下，每個通道平均消耗13微安培的電流，並達到3.74的雜訊效率指標(NEF) 。 I
第三個架構為胸腔阻抗檢測電路，此架構可成功量測出胸腔阻抗訊號，並且採用方波刺激電流有別於傳統的刺激方法，方波刺激電流可大大降低整體電路的功率消耗，此架構還提出新的解調方法讓解調出的胸腔阻抗訊號有良好的線性度及精準度。此架構採用台積電0.18微米製程，面積消耗為0.5平方毫米，在1.8伏特的電源下最大電流消耗為75微安培，刺激電流大小為10到50微安培之間。

關鍵字

低功率 ； 低雜訊 ； 感測器 ； 類比前端電路 ； 儀表放大器 ； 胸腔阻抗

並列摘要

The thesis presents the design of analog front-end circuits for biomedical applications. The analog-front-end circuit is the most critical building block in bio-potential monitoring SoC, since it should amplify very weak signals under noisy aggressors.
Three circuit architectures are implemented and in this thesis. The first chip proposes a 4-channel low noise amplifier that utilizes a time-multiplexing topology to share a single primary LNA with four independent input channels. So adopting this technique can save large area and large power consumption. However, the characteristic of time-multiplexing technique is not continuous of each channel and it will fold the output noise of LNA back to low frequency. So the noise folding problem will make signal quality degrade. This chip occupies 0.015 mm2 and consumes 4.25 μA per channel. The NEF is 11.56 due to the noise folding problem. It is implemented with TSMC 0.18 μm process.
The second chip demonstrates a LNA that adopts an orthogonal frequency chopping (OFC) technique to realize a continuous 2-channel LNA with only one active amplifier. The OFC technique modulates two input signals to be orthogonal to each other before feeding into LNA. After two signals are amplified by LNA, the two signals can be demodulated back easily. Due to the OFC technique, the flicker noise will be modulated to high frequency. The signal quality of this chip is better than chip 1. This chip occupies 0.035 mm2 and consumes 13 μA per channel. The NEF is 3.74. It is implemented with TSMC 0.35 μm process.
III
The third chip implements a thoracic electrical bio-impedance(TEB) readout system. It uses square-wave current as injection current. The square-wave current can save large power and have good linearity rather traditional way. The proposed demodulated way can achieve high linearity and good accuracy. This chip occupies 0.5 mm2 and consumes 75 μA. The injection current range is from 10 μA to 50 μA. It is implemented with TSMC 0.18 μm process.

並列關鍵字

Low Power ； Low Noise ； Sensor ； Analog Front-End Circuits ； Instrumentation Amplifier ； Biomedical Applications ； TEB

參考文獻

[1] R. F. Yazicioglu, C. Van Hoof, and R. Puers, Biopotential Readout Circuits for Portable Acquisition Systems, Springer, 2009.

[3] R.F. Yazicioglu, P. Merken, and C. Van Hoof, “A 200μW Eight-Channel EEG Acquisition ASIC for Ambulatory EEG Systems,” IEEE J. Solid-State Circuits, vol.43, no.12, pp. 3025 - 3038, Dec. 2008.

[5] V. Majidzadeh, A. Schmid, and Y. Leblebici, “Energy Efficient Low-Noise Neural Recording Amplifier With Enhanced Noise Efficiency Factor,” IEEE Trans. Bio. Circuits Syst., vol. 5, no. 3, pp. 262–271, Jun. 2011.

[7] D. P. Bernstein, “Impedance cardiography : Pulsatile blood flow and the biophysical and electrodynamic basis for the stroke volume equations,” in J. Electrical Bioimpedance, vol, 1, pp. 2–17, Dec. 2009.

[8] J. H. Meijer, A. Smorenberg, and A. B. J. Groeneveld, “Assessing cardiac preload by the Initial Systolic Time Interval Obtained from impedance cardiography,” in J. Electrical Bioimpedance, vol, 1, pp. 80–8., Dec. 2010.