PIERS Proceedings, Suzhou, China, September 12–16, 2011 1004 Filter for Processing of NMR Signal Martin Friedl, Jiřı́ Sedláček, Lubomı́r Fröhlich, and Radek Kubásek Brno, FEEC BUT, UTEE, Kolejnı́ 2906/4, Brno 612 00, Czech Republic Abstract— One of the fields of science, where is going forward continual development, is the area of nuclear magnetic resonance (NMR). Bulk of the NMR issues concerns on signal processing in analog and digital form in NMR signal. In this paper is focused our attention on design of active analog filter with frequency dependent negative resistor (FDNR). There is designed active ARC filter based on ladder structure of the LRC filter, then the filter DCR is made through the use of Brutton transformation and finally the DCR filter is converted to the filter with FDNR elements. The designed and realized filter is used as a low pass filter, which aims to filter out the noise of NMR signal. The filter is placed behind the mixer, which converts the useful NMR signal to baseband. The designed and realized filter performs role of the anti-aliasing filter before the A/D converter. 1. INTRODUCTION The paper deals with the design of low-pass filter with lossy FNDR element. The purpose of the using the filter is as anti-aliasing filter in NMR spectrometer signal path, see Fig. 1. Signal of resonant nucleus detected by probe can be expected at µV level. Low noise preamplifier will increase the level of NMR signal before mixer. The mixer bring useful signal to base-band. Interfrequency signal required to be filtered by low-pass filter with appropriate order. There is possible to use Bessel or Butterworth approximation. Filter has to have low noise parameters. In general, noise parameters are the most significant in all signal path design. 2. FILTER DESIGN There is by the design of frequency RLC filters the biggest problem with the quality, size and price of coils. Consequently, for low and medium frequencies, RLC filters are preferably replaced by active filters RC (ARC). Their basic principle consists in replacement the coil through the use of an active element with resistors and capacitors [1]. One of the suitable method is solution consists in the creation the ARC circuit with transfer function of the 2th order, which is equal to transfer functions of RLC filter through the use of FDNR blocks [2]. The filters with FDNR elements replace coils indirectly using Bruton’s transformation [3], which transforms the initial RLC structure into equivalent behaving the CRD structure (Fig. 2). This new structure does not contain an element of inductive character, but uses the properties of the synthetic element FDNR, which are preferably used for low pass filter. The active circuits realized of the FDNR elements can be separated into different types according to their basic circuit characteristics [4]. The lossy FDNR grounded circuit (Fig. 3(b)) can be realized by the use one active element (OA) and in this circuit the lossy are represented by a parallel MAGNET PROBE OSCILATOR FDNR + GAIN A/D COUPLER GAIN 40 dB TRANSCEIVER MIXER LOW PASS FILTER GAIN 20 – 60 dB Figure 1: The anti-aliasing filter for mirror band after mixer and noise suppression. Progress In Electromagnetics Research Symposium Proceedings, Suzhou, China, Sept. 12–16, 2011 L1 L3 L2 C1 R1 C2 Rp2 Rp1 R R2 1005 R3 Bruton transformation C Cp1 FDNR 2 FDNR 1 Cp2 Figure 2: Principle of the active LP filter design. CB R1 IN 1kΩ L1 L2 9,8 mH 31,8 mH C1 R L3 CA 9,8 mH C2 25,7 nF Dp _ R2 1kΩ 25,7 nF Cp + (a) (b) Figure 3: (a) The LRC low pass 5th order, (b) the grounded FDNR with the parallel lossy. 1 kΩ C3 4,7 nF IN C1 R1 C2 33 nF 820 Ω 4,7 nF R2 R3 820 Ω C4 R4 27 kΩ OPA355 4,7 nF C5 4,7 nF R2 163Ω R5 27 kΩ C6 10 nF OPA355 Figure 4: The optimized ARC filter with the lossy FDNR blocks. capacitor. These lossy correspond to resistors Rp1 and Rp2 parallel connected to capacitors in the initial circuit of the RLC filter (Fig. 2). Thus it is very easy these lossy to simulate and to observe filter response to different sizes of the resistors Rp1 and Rp2. The final design of the low-pass RLC filter of the 5th for cut-off frequency 10 kHz is on Fig. 3(a). Through the use Bruton’s transformation and grounded lossy FDNR element (Fig. 3(b)) we obtain circuit of the ARC filter with lossy FDNR elements. The results of the simulation (Fig. 6) show that parallel lossy of FDNR elements have highly influence to the resulting characteristic (the black curve). Therefore the optimization [5–7] was done and the resulting transfer characteristic of ARC filter with the real OA (OPA355) [8] corresponds to the red curve. It is clear that by the using of the optimization was achieved almost the same characteristic as it is at the RLC filter (the blue curve). There is possible of the using the optimization to reach almost identical transfer as with the lossless FDNR elements with the lossy FDNR elements. The optimization consists in spreading lossy from the lateral branch of the filter to the whole circuit. The resulting realization of the filter is on the Fig. 5. The measuring was done with the help of vector analyzer Bode 100, by company OMICRON Lab [9]. During parameters verifying of designed circuit was also searched the characteristic sentence of real active elements. Very crucial is mainly sufficient width of the OA band, which is one of the key presumptions of the distortion-less function of the filter in demanded band-pass and it is necessary to solve it with the regard to cut-off frequency of the filter with the sufficient reserve, minimally one or two orders higher [10]. Very good results were achieved by OA by company Texas Instrument [8] OPA355, which are determined also for active filters and they have sufficient width of the band (GBW 200 MHz). The designed low-pass filter was completed with the three amplifiers for processing of NMR signal. At first, the transfer characteristic of the filter was measured (Fig. 6 — the green curve) and after the transfer characteristic of the filter with amplifiers (Fig. 7) consequently there is obvious gain 22 dB. PIERS Proceedings, Suzhou, China, September 12–16, 2011 1006 Figure 5: The realized filter with the amplifiers. Figure 6: The magnitude response of the filters: LRC, ARC, ARC optimization and the realized filter. Figure 7: The measured magnitude response of the filter with amplifiers. 3. CONCLUSIONS This contribution deals with synthesis and optimization of ARC low pass filter based on modified simple and economic lossy building blocks (FDNR). The designed frequency filter will be used for pre-processing of analogue signals before digitalization in the NMR signal processing area whereas the resulting characteristics of the designed and measured filter show, that it is possible using it with utilize of the optimization. The quick progress of modern technologies enables realization of modern structures of analogue frequency filters. In this context, there is necessary to evolve the synthesis and optimization meth- Progress In Electromagnetics Research Symposium Proceedings, Suzhou, China, Sept. 12–16, 2011 1007 ods of these structures with regard to the possibilities of the modern active components (voltage OA with GBW approximately 1 GHz, current conveyors, transimpedance OA). In the future research, it will focused on optimization of the basic building block of 2nd order and their characteristics with the modern active elements above they will be analyzed from the point of view of their usage in circuits of higher orders and area of higher frequencies. ACKNOWLEDGMENT This work has been supported by the project of the BUT Grant Agency FEKT-S-11-5. REFERENCES 1. Sedláck, J. and K. Hájek, Kmitocové Filtry, 1. vydánı́, Praha, 535s, BEN — Technická Literatura, 2002, ISBN80-7300-023-7. 2. Pactitis, S. Active Filters: Theory and Design, CRC Press, 274 str., USA, 2008, ISBN 978-14200-5476-7. 3. Bruton L. T., RC-active Circuits Theory and Design, Prentice-Hall, Inc., Englewood Cliffs, New Jersey, 1980, ISBN 0-13-753467-1. 4. Friedl, M., Synthesis of Modern Structures Frequency Filters, 21s, Dizertacı́ Práce, Department of Theoretical and Experimental Electrical Engineering, Faculty of Electrical Engineering and Communication, Brno University of Technology, Brno, 2010. 5. Hájek, K. and J. Sedláck, “Lossy LC ladder prototypes and their use for ARC filter optimization,” Wseas Transactions on Electronics, Vol. 2, No. 3, 94–99, July 2005, ISSN1109-9445. 6. Friedl, M., L. Fröhlich, and J. Sedláck, “Modified approximation types for lossy building blocks,” PIERS Proceedings, 521–525, Xi’an, China, March 22–26, 2010. 7. Hájek, K., V. Michal, J. Sedláck, and M. Steinbauer, “A simple method of goal-directed lossy synthesis and network optimization,” Advances in Electrical and Electronic Engineering, 249– 253, Zilina, 2006, ISSN1336-1376. 8. Focus.ti, OPA355, Dostupné z, [online], 2010 [cit. 2010-04-28], www:http://focus.ti.com.cn/cn /lit/ds/sbos195d/sbos195d.pdf. 9. Omicron-lab, Vector Network Analyzer, Bode 100 Extended frequency range 1 Hz–40 MHz. Dostupné z, [online], 2010 [cit. 2010-04-28], www:http://www.omicron-lab.com/. 10. Baker, C., Chyba Zisku U Operaccı́ch Zesilovacu, www.Pandatron.cz: Elektrotechnický Magazı́n, Dostupný z, [online], 2010 [cit. 2011-01-31], www:http://pandatron.cz/?1382&chyba zisku u operacnich zesilovacu.