Chapter 5 Integrated Electronics toward Microsystem
5.3 Multichannel Analog Front End
The studies of biomedical signals, or say biopotential, play an important role in the researches of biomedical science. Generally, biopotentials used in diagnoses are monitored with electrodes to acquire these electrical signals. Generally speaking, EEG and ECG are most widespread used in clinical diagnoses, and they span over a relatively low frequency range, say dc to about 100Hz [107]. The collected weak physiological signals have to be amplified, filtered and conditioned for the actual clinical applications because the raw signals are so small that they may be strongly interfered by the environmental noise. The demands of small portable, wearable and mobile sensing system are increased rapidly for the long term monitoring requirements; a low-power, low-noise amplifier system for the neural sensing applications is therefore needed.
Results in this section were co-worked with Tan-Jiun Ho and Yen-Chang Chen in Microsystem Control Laboratory, National Chiao-Tung University.
5.3.1 Differential-Difference Amplifier
The conventional instrumentation amplifier (IA) designs include traditional three operational amplifiers structure (3OIA) [108], current-balance structure (CBIA) [109-110]
and differential difference amplifier (DDA) [111]. 3OIA structure requires completely resistor match to obtain very high common-mode rejection ratio (CMRR); however, mismatched resistor can cause serious CMRR debasement [112]. The advance of CBIA is its simple structure but also suffer from the requirement of current-mirror match, or the mismatch can also affect the CMRR efficiency [113].
In this work, a 16-channel analog front end (AFE) neural amplifier using DDA as first stage is presented. Comparing to previous works, resistor mismatch in the DDA design will solely influence the amplification gain. Block diagram of the complete 16-channel amplifier is shown in Fig. 5.13 (A). The differential difference amplifiers (DDA) are used as first stage of the AFE amplifier for high-CMRR and low-noise requirements. A 16:1 MUX controlled
by the clock signal allows all 16 DDAs share the same second and third amplifier/filter stages. The MUX frequency can be up to 200 kHz to scan the 16-channel inputs. The second stage and third stage use operational transconductance amplifier (OTA) to implement low-pass filter, high-pass filter and gain amplifier with selectable bandwidth and gain. The gain and low-pass filter frequency are controlled by digital signals LPF-Select and Gain-Select.
The DDA used in this work is modified from [112], and the circuit is depicted in Fig. 5.13 (B). To reduce the original flicker noise, the input stage (M1-M2 and M3-M4) are PMOS with wide width and operated in weak inversion region. The input stage transfers the input voltage into current. M9 and M10 construct a common source amplifier. The differential currents flowing through M7 and M8 are transferred back to voltage signal by the load device M11 and M12. The output gain stage of DDA is constructed by the two-stage amplifier (M13-M19), which acts as a differential-to-single- ended converter and buffer. C1 and C2 are Miller capacitances used to increase the phase margin. Finally, the DDA-based non-inverting amplifier is implemented from the topological placement of R1 and R2. The input/output relation is defined as
inout
R R V
V
2 1
1 ( 5 - 1 5 )Fig. 5.13 (A) Electrical structure of the 16-channel amplifier (B) Schematic of the DDA Fully differential OTA is used in the second and third stage of our 16-channel neural amplifier. The transconductance Gm of the amplifier is controlled by the bias current, while the input transistors are operated in the weak inversion, and the current mirror transistors are operated in the strong inversion. This design minimizes input-referred noise for a given bias current [114]. Fig. 5.14 shows the schematic of the second and third stages, which serve as a programmable gain and filter. In Fig. 5.14 (A), the filter is implemented by two series PMOS transistors across C2 operated in sub-threshold region. The mid-band gain AM
is set by C1/C2=40. The cutoff frequency is approximately defined as Gm/(AMCL), where Gm is the transconductance of OTA. The selectable on-chip capacitive loads provide 50 Hz
to 10 kHz low-pass cut-off frequency. Fig. 5.14 (B) shows the amplifier stage with tunable gain and fixed high cutoff frequency. The ratio of capacitors determines the gain factor of 15 to 1. The two series PMOS resister across a capacitance is employed to provide the high-pass cut-off frequency of 0.1Hz. The final output of the AFE chip is serial data, of which the packaging volume and wire bonding pads can be reduced due to less I/O numbers.
Fig. 5.14 Schematic of the 2nd and 3rd stages which serve as programmable gain and filter
5.3.3 Characterization and Performance
The 16-channel neural amplifier is fabricated using the TSMC 0.35um two-poly four-metal CMOS process and mounted on a dedicated printed-circuit board for testing. To measure the chip performance, a DAQ device (National Instruments, USA) is used to collect and transmit the 16-channel signal to a laptop. The DAQ device also provides a 200k Hz clock to operate the MUX. A microphotograph of the complete chip is shown in Fig. 5.15(A). The whole chip achieves a size of 4.18mm2 including pads. It drains a power of 108 μW when operating for full 16-channel neural recording. The chip operates from a single 1.8V supply without off-chip components. The measured input referred noise is 2μVrms in the band of dc to 200Hz.
Fig. 5.15 (B)-(C) shows the measured frequency response with tunable gain/band, and the noise spectrum. The measured input referred noise is 2μVrms in the band of dc to 200Hz.
To verify the neural recording capability of the present AFE chip, simultaneous recording performance comparison with commercial ICs is tested by using a neural signal simulator (Cyberkinetics, US). A similar 3-stage amplifier structure which consisted of an instrumentation amplifier LT1789 (Linear Technology, US) and two operational amplifiers AD8607 (ADI, US) was used for comparison. Both present AFE chip and commercial ICs provides a band of 0.2-6k Hz with 66dB in gain. A 100-ms segment of comparison result is show in Fig. 5.15(D). In Fig. 5.15(D), the upper solid line and the lower dotted line display the results from present neural amplifier and commercial ICs, respectively. These two recordings result in a high correlation (>0.99) in average, which demonstrates the correct behavior of the present chip in neural recording.
Fig. 5.15 (A) Microphotograph of the 16-ch neural amplifier chip (B) Measured tunable gain/band frequency response (C) Noise spectrum (D) Simultaneous comparison by neural
simulator, upper solid line and the lower dotted line display the results from present amplifier and commercial chip. Result shows high correlation (>0.99) in average Human EEG recording is obtained using Ag/AgCl electrodes on occipital region with amplifier ground connected to the A1, as shown in Fig. 5.16. Alpha waves (8-12 Hz) are observed when the subject’s eyes were closed initially. Fig. 5.17 display the 16 channel recording consists of 9 leads of ECG from an ECG generator and 7 sinusoid waves from 10Hz to 40Hz, which are used for functional demonstration of the analog multiplexer. Table 5.4 summarizes the comparison of the measured parameters of this neural chip with those of reported multichannel works. Present neural amplifier offers technical merits of reduced supply voltage, sufficient low power per channel and reasonable low noise performance, yet offers comparable measured results such as input offset, CMRR and PSRR.
Fig. 5.16 Human EEG recording is obtained from on occipital region with amplifier ground connected to the A1, Alpha waves (8-12 Hz) are observed when eyes were closed
Fig. 5.17 16-channel recording consists of 9 leads of ECG from an ECG generator and 7 sinusoid waves from 10Hz to 40Hz
Table 5.4 Measured performance of the proposed neural amplifier
This work [115] [116] [117] [118] [119]
JSSC2007 JSSC2009 BioCAS2009 BioCAS2009 BioCAS2010
CMOS Technology[μm] 0.35 1.5 0.35 0.5 0.18 0.35
Supply voltage[V] 1.8 ±2.5 3 3.3 1.8 3
Number of channels 16 6 256 16 16 128
Mid-band gain[dB] 48-65 39.5 48-68 39.6 70 54-73
Highpass frequency[Hz] 0.1 0.025 0.01-70 0.2-94 100 0.5-50
Lowpass frequency[Hz] 50-6k 7.2k 500-5k 140-8.2 9.2k 500-10k
Current per channel[μA] 4.5 -- -- 8 2.6 4.25
Input-referred[μVrms] 2 2.2 7 1.94 5.4 6.08
PSRR[dB] >78 >85 -- >70 -- --
CMRR[dB] >90 >83 -- >76 -- --
Amplifier Power[μW] 108(6.75/ch) 80(13.33/ch) 3840(15/ch) 422(26.4/ch) 8.6/ch 1632(12.75/ch)
5.3.4 Chip-on-Broad Integrated Microsystem
Wireless microsystem constructed with fabricated 16-channel neural amplifier, 8051-based micro controller unit (C8051F582, Silicon Laboratories, Inc., USA) and Bluetooth transceiver (BTM-182, Rayson Technology Inc., Taiwan) is implemented. Detailed system structure is shown in Fig. 5.18. The 16-ch AFE amplifier provides signal conditioning of the weak biopotentials, produces time-division serial data output. The series data are then digitized and packaged with marker, wirelessly transmitted by the Bluetooth transceiver.
The C8051F582 is 8051 core based micro controller with embedded 200ksps 12-bit SAR
ADC, 50M Hz internal oscillator, operation voltage from 1.8 to 5.25V. The UART port is used to communicate with the Bluetooth module. The BTM-182 consists of CSR Bluecore-4 single chip, complete 2.4GHz radio transceiver and print circuit antenna, is designed for transparent wireless serial connection. Benefit to the easy usage of Bluetooth, single lap top is utilized as transceiver host.
Fig. 5.18 System block diagram of the chip-on-Broad level integrated microsystem Fig. 5.19 (A) shows the control flow of the MCU, and detailed timing chart of the data is in Fig. 5.19 (B). The microsystem provides two operation modes, say 16-channel mode and single channel mode. In the case of 16-channel mode, the clock generated by MCU is sent to the AFE amplifier to scan the 16 channels and outputs series data. A counter controlled Sample-control-Bit (SCB) is used to sample the analog data to ADC at the falling edge of the clock. Red dot in Fig. 5.19 (B) denotes the sample point for ADC. Digitalized 12-bit data is then divided into 2 bytes to UART. After 16 channels are sent, two FF which acts marker are sent into UART before next 16-channel data. The marker is used by the software to divide and reconstruct the serial data back to simultaneous 16 channels. Windows-based software is designed for receiving data from the presented microsystem and sends commands to tune on, reset or parameter setting to the microsystem. Additionally, the full system is proposed to be packed onto the back of a rat as a backpack with wires link to the implanted sensors in brain.
Fig. 5.19 (A) Control flow of the MCU (B) Data timing and packaging method
A 4-layer PCB design with Chip-on-Broad (COB) technique is used to divide the analog (neural amplifier) and digital circuit into two side of the PCB with two powering layer in middle layers. Fig. 5.20 shows the fabricated microsystem. The over all PCB is 39mm x 39mm in size, weight 20g with a 380mAh battery. A PCB antenna is also designed on the Bluetooth module. Commercial regulators are employed for providing 3.3V, 1.8V and 0.7V for ESD circuit, system supply and reference bias, respectively. Additionally, programming ports, power on LED, reset button and debugging circuitry are embedded on the broad for function verification although this can seriously increase system dimension.
The Flexible cable wire is designed by a 2-layer polymer-based ribbon, connected to the microsystem via a ZIP connector. As shown in Fig. 5.20, a presented 3D neural probe array is attached to the microsystem. In fact, the sensors can easily be replaced for versatile applications, such as EEG or ECoG by using presented MEMS dry electrode (MDE) or polymer based brain surface grid electrode array. Table 5.5 summarizes the microsystem specification.
Fig. 5.20 Fabricated microsystem (A) Digital side (B) Analog side Table 5.5 Specification list of the microsystem
Parameter Value
Size 39 x 39 mm2
Weight with/without battery 20g/9.96g Resolution /Channel number 12bit/ 16 channels
AFE Gain 400-4000V/V
Sample Rate 1.25k Hz per channel
Transmission rate Max 961200bps
Current Consumption/ Battery capacity 20 mA / 380mAh
5.4 Summary
In this chapter, key components including inductive coupled coil for wireless powering, low dropout voltage regulator and multi-channel analog-front-end amplifier for biopotential conditioning are developed for the biomedical microsystem design. Section 5.2 describes the RF-powering coil and regulator design, following with a 16-channel AFE chip development reported in Section 5.3. Spiral coils as a wireless power module with LDO linear regulator circuit convert RF signal into DC voltage provide a batteryless implantation for truly free-behavior monitoring. The 16-channel AFE provides pre-amplification and signal conditioning. Also, a chip-on-broad (COB) integrated microsystem consisting of fabricated AFE chip, commercial micro control unit (MCU) and wireless transceiver module are developed for the conceptual representation of the proposed microsystem.
Chapter 6 Conclusion
6.1 Summary of Thesis
In this dissertation, a neural sensors and microsystems toward neuroprosthesis applications is presented. Detailed designs and results for the sensors and systems are described in previous chapters. The MEMS surface-mounted dry electrode provides superior low interface impedance. Alternative MDE designs including DS-MDE and transparent MDE for better self-stability and PDT applications are also reported. Flexible gird electrode array using parylene-C as structure for ECoG recording is also proposed. In-vivo recording of auditory response experiment in rat demonstrates its practical effectiveness. New stacking method for assembling a 3D microprobe array is developed for implantable 3D neural recording. Proposed method provides simple process, solid structure with possibility for system integration, design flexibility and volume usage efficiency. The neural signal data acquired by 3D multichip array achieves the recording and mapping of the neural signal network and interconnections among the target brain structure, which allows further studies for event-related observation.
Wireless RF-powering electronics approach is described for implantable biomedical microsystem applications. Miniaturized spiral coils as a wireless power module and low-dropout regulator circuit convert RF signal into DC voltage for batteryless implantation with low quiescent-current and line/load regulation, high antenna/current efficiency with thermal protection to avoid damage to the implanted tissue. A 16-channel analog front end neural amplifier is also introduced, which offers technical merits of reduced supply voltage, sufficient low power per channel and reasonable low noise performance, yet offers comparable measured results such as input offset, CMRR and PSRR. Finally, an integrated microsystem using commercial micro control unit and wireless transceiver module are shown as a conceptual representation of proposed system.
To summarize, proposed sensors and electronics provide versatile neural recording and key component realization in microsystem design, as well as achieving the development of biomedical prosthesis applications by integrating with commercial modules. However, there are still plenty of rooms needed to be discovered beyond this work.
6.2 Future Work
In this dissertation, versatile neural sensors and electronics in biomedical system had been studied preliminarily. There are some prospects for the further researches including detailed neural-electrode interface studies, all-polymer based implantable sensor array for true 3D analysis, long wireless/batteryless transmission distance, system-on-chip design and reliable packaging integration with neural sensors.
Furthermore, specifically sensing electrodes for localized neuron recording require special microfabrication process with particular impedance performance. Localized neural recording can reduce the difficulty of neuron sorting or classification because of the specified signal source. Additionally, standard electrode model should be characterized by a small-signal source to compare and identify the electrode performance with prior arts. The auditory evoked potential (AEP) studied in the dissertation should also be comparing with standard samples in prior arts, as well as the neural potential recorded form different layer of the tissue by the presented neural probes. The practical survival time of the implant device is another important issue, which is considered as the next step beyond current achievements. Noise measurements and analysis is another weakness part in the work, including MDE development and the wireless neural microsystem. Comparison of the noise performance between MDE and wet electrode should be detailed investigated. The improvement of the power transmission efficiency (PTE) in photodynamic therapy by transparent MDE is clearly observed, however, current result lack of studies in the influence of the interface variance by the surface mounted MDE structure. Quantitative design procedures should be involved in the development the next generation MDE with logical and empirical verification. Finally, fully integrated system on chip design will be the ultimate goal toward neural prosthetic microsystem implementation.
Reference
[1] N. V. Thakor, ―Biopotentials and Electrophysiology Measurement‖ in The Measurement, Instrumentation, and Sensors Handbook. Webster, John G., CRC Press, 1999.
[2] R. Plonsey, ―Bioelectric Phenomena,‖ New York: McGraw-Hill, 1969.
[3] R. Plonsey and R. C. Barr, ―Bioelectricity,‖ New York: Plenum, 1988.
[4] R. C. Barr, ―Basic electrophysiology,‖ in The Biomedical Engineering Handbook, Bronzino J., Ed., Boca Raton, FL: CRC Press, 1995.
[5] N. V. Thakor, "Neuroprosthesis: An Inspiration for Brain-Machin Interface," Lecture Note Presented at the International Workshop on Bio-Inspired System and Prosthetic Devices, Hsinchu, Taiwan, 2010 [6] Legatt, J. Arezzo and H.G. Vaughan, ―Averaged multiple unit activity as an estimate of phasic changes in
local neuronal activity: effects of volume-conducted potentials,‖ Journal of Neuroscience Methods, vol.
2(2), pp.203-217, 1980.
[7] E. Niedermeyer and D. Silva F.L., ―Electroencephalography: Basic Principles, Clinical Applications, and Related Fields,‖ Lippincot Williams & Wilkins, 2004.
[8] B. Abou-Khalil and Musilus, K.E., ―Atlas of EEG & Seizure Semiology,‖ Elsevier, 2006.
[9] E.N. Brown, R.E. Kass and P.P. Mitra, ―Multiple Neural Spike Train Data Analysis: State-of-The-Art and Future Challenges,‖ Nature Neuroscience, vol.7, pp.456-461, 2004.
[10] P.K. Campbell, K.E. Jones, R.J. Huber, K.W. Horch and R.A. Normann, ―A silicon-based, 3-dimensional neural interface—manufacturing process for an intracortical electrode array,‖ IEEE Transaction on Biomedical Engineering, vol. 38, pp758-768, 1991.
[11] S. Kim, R. Bhandari, M. Klein, S. Negi, L. Rieth, P. Tathireddy, M. Toepper, H. Oppermann and F.
Solzbacher, "Integrated wireless neural interface based on the Utah electrode array," Biomedical Microdevices, vol. 11, pp. 453-466, Apr 2009.
[12] B. Rubehn, C. Bosman, R. Oostenveld, P. Fries, T. Stieglitz, ―A MEMS-based flexible multichannel ECoG-electrode array,‖ Journal of Neural Engineering, vol. 6, 2009.
[13] P. Cong, N. Chaimanonart, W. H. Ko, and D. J. Young, ―A wireless and batteryless 130mg 300µW 10b implantable blood-pressure sensing microsystem for real-time genetically engineered mice monitoring,‖
IEEE International Solid-State Circuits Conference, 2009.
[14] A.M. Sodagar, G.E. Perlin, Y. Yao, K. Najafi and K.D. Wise, ―An Implantable 64-Channel Wireless Microsystem for Single-Unit Neural Recording,‖ IEEE Journal of Solid-State Circuits, Vol. 44, pp.2591-2604, 2009.
[15] M. Teplan, ―Fundamentals of EEG measurement,‖ Measurement Science Review, vol.2, 2002 [16] H. A. Miller, D.C. Harrison, ―Biomedical Electrode Technology,‖ Academic Press, NY, 1974.
[17] A. P. Spence, ―Basic Human Anatomy,‖ Benjamin Cummings, Redwood City, CA, 1990.
[18] S.I. Fox, ―Human Physiology,‖ 9th Ed., McGraw-Hill, 2004.
[19] P. Griss, E. Peter, T. Heli, M. Pekka, O. Stig and S. Göran, ―Micromachined Electrodes for Biopotential Measurements,‖ Journal of Microelectromechanical Systems, vol. 10, no. 1, pp. 10-16, 2001.
[20] C.W. Chang, L.W. Ko, F.C. Lin, T.P. Su, T.P. Jung, C.T. Lin and J.C. Chiou, ―Drowsiness Monitoring with EEG-Based MEMS Biosensing Technologies,‖ Journal of Gerontopsychology and Geriatric Psychiatry, Vol. 23, pp.107-113, Jun 1, 2010.
[21] P. Griss, E. Peter and S. Göran, ―Micromachined barbed spikes for mechanical chip attachment,‖
Sensors and Actuators Journal-A: Physical, vol. 95, pp.94-99, 2002.
[22] J. G. Webster, ―Medical Instrumentation Application and Design,‖ 3rd ed., New York: Wiley, 1998.
[23] R. Matthews, N.J. McDonald, H. Anumula, J. Woodward, P.J. Turner, M.A. Steindorf, K. Chang, and J.M. Pendleton, ―Novel hybrid bioelectrodes for ambulatory zero-prep EEG measurements using multi-channel wireless EEG system,‖ Lecture Notes in Artificial Intelligence, vol. 4565, pp.137–146, 2007
[24] A. Amditis, A. Polychronopoulos, ―System architecture of a driver's monitoring and hypovigilance warning system,‖ IEEE Intelligent Vehicle Symposium, vol.2, pp.527–532, 2002
[25] C. T. Lin, L.W. Ko, I.F. Chung, T.Y. Huang, Y.C. Chen, T.P. Jung and S.F. Liang ―Adaptive EEG-based Alertness Estimation System by Using ICA-based Fuzzy Neural Networks,‖ IEEE Transactions on Circuits and Systems I: Regular Papers, vol.53, pp.2469–2476, 2006
[26] W. Klimesch, ―EEG alpha and theta oscillations reflect cognitive and memory performance: a review
and analysis,‖ Brain Research Reviews, vol.29, pp.169–195, 1999
[27] M. Steriade, ―Central core modulation of spontaneous oscillations and sensory transmission in thalamocortical systems,‖ Current Opinion in Neurobiology, vol.3, pp.619–625,1993
[28] T. P. Jung, S. Makeig, M. Stensmo and T.J. Sejnowski, ―Estimating alertness from the EEG power spectrum,‖ IEEE Transactions on Biomedical Engineering, vol.44, pp.60–69, 1997
[29] R. Grace, V.E. Byrne, D.M. Bierman, J.M. Legrand, D. Gricourt, B.K. Davis, J.J Staszewski, and B.
Carnahan, ―A drowsy driver detection system for heavy vehicles,‖ Proceedings of the 1998 17th AIAA/IEEE/SAE Conference on Digital Avionics Systems, 2, I36/1–I36/8. 1998
[30] S. Chatterjee, and A.S. Hadi, ―Influential observations, high leverage points, and outliers in linear regression,‖ Statistical Science, vol.1, pp.379–416, 1986
[31] C. T. Lin,I.F. Chung, L.W. Ko, Y.C. Chen, S.F. Liang and J.R. Duann, ―EEG-based Assessment of Driver Cognitive Responses in a Dynamic Virtual-Reality Driving Environment,‖ IEEE Transactions on Biomedical Engineering, vol. 54, pp.1349–1352, 2007
[32] X. Sun and W.N. Leung, ―Photodynamic therapy with pyropheophorbide-a methyl ester in human lung carcinoma cancer cell: efficacy, localization and apoptosis,‖ Photochem Photobiol, vol. 75, no. 6, pp.
644-651, 2002.
[33] Goldberg DJ, ―Photodynamic therapy in skin rejuvenation,‖ Clinics in Dermatology, vol. 26, pp.608-613, November, 2008.
[34] M. R. Hamblin, D.A. O'Donnell, N. Murthy, C.H. Contagand and T. Hasan, ―Rapid Control of wound
[34] M. R. Hamblin, D.A. O'Donnell, N. Murthy, C.H. Contagand and T. Hasan, ―Rapid Control of wound