Science Series Data Report Vol 4, No. 5;May 2012 DAC Mismatching Compensation in Multibit Sigma-Delta Modulators with Two-Step Quantization Sakineh Jahangirzadeh Department of Electrical Engineering, Shahid Chamran University of Ahvaz, Ahvaz, Iran E-mail: s_jahangirzadeh@yahoo.com Ebrahim Farshidi Department of Electrical Engineering, Shahid Chamran University of Ahvaz, Ahvaz, Iran Tel: 61357-8313 E-mail: farshidi@scu.ac.ir Abstract The use of multibit quantizers in sigma-delta modulators can increase SNR, improve stability and reduce integrator power consumption. However, using multibit quantizers causes nonlinearity in DAC converter. In recent decades the uses of data weighted averaging (DWA) methode has been proposed for reducing mismatch errors. However, each added bit to quantizer resolution causes an exponential increase in the power dissipation, required area and complexity of the DWA circuit required to attenuate DAC mismatch errors. This paper proposes the prospect of using a segmented feedback path with coarse and fine signals to reduce DWA complexity for modulators with large internal quantizers. This reduces the DWA circuit complexity, power dissipation, and size. But this method adds an additional noise to system. To overcome this problem, two solutions are suggested: one that uses calibration method to cancel mismatch between coarse and fine DACs, and another that frequency-shapes this mismatch error using requantization method. Keywords: Sigma-Delta modulator, Data Wighted Averaging (DWA), Segmentation 1. Introduction The sigma-delta analog to digital modulator (Σ∆ ADC) has been widely used in recent decades for low frequency, high resolution applications such as digital audio and high-precision instrumentation [1]. Recently, however, is extending the signal bandwidths of Σ∆ ADCs into the MHz range while maintaining high resolution [3]. Three of the key design parameters that affect the resolution of a Σ∆ ADC are: the sampling frequency relative to the bandwith of interest (OSR), the order of the noise transfer function, and the number of internal quantization levels. As each is increased, the theoretical resolution of the ADC is increased. However, the complexity, power dissipation, and required chip area also increase. 32 Early Σ∆ converter used a single bit quantizer in the loop because of their suitability for VLSI implementation and their superior linearity [4],[5]. The use of multibit quantization has been limited because non linearity in the DAC of a sigma-delta modulator translates directly into non linearity of the entire modulator, producing a distorted output. Non linearity in the DAC also modulates the quantization noise into the signal band, thus degrading the SNR. However, multibit modulators have several advantages such as increased resolution for the same oversampling ratio, improved stability, relaxed amplifier requirements and better tone behaviour [1]. Attempts to eliminate the non linearity problem associated with multibit Σ∆ modulators have resulted in the use of DWA techniques which SSDR@SCIENCERECORD.COM Science Series Data Report shape the noise generated by DAC unit element mismatch, shifting it to higher frequencies which are out of the band of interest. Increasing internal quantization levels beyond five bits improves SNR but presents significant challenges. Both the internal quantizer and the DWA logic grow exponentially in complexity, size, and power dissipation as the internal quantizer resolution increases. Using two-step ADC is a logical alternative for reducing quantizer power. A folding ADC could provide quantization above eight bits while still maintaining the low latency required of the internal quantizer. Recent work has also shown that it is possible to incorporate two-step ADCs with in asingle loop modulator, permitting lower power quantizers while maintaining loop stability [6], [7]. This paper will present two architectures that permit the uses of DWA with two step quantizer. The paper is organized as follows. In section 2, basic principle of the data weighted averaging algorithm is presented. In section 3, the problems associated with applying traditional DWA algorithm to a segmented coarse/fine DAC structure are discussed. In section 4, calibration method and requantization method in section 5 to overcome these problems are proposed. Simulation results are presented and discussed in section 6,and concluding remarks are provided in section 7. 2. Vol 4, No. 5;May 2012 beginning with the next available unused element. The operation principle is illustrated in Figure 2. V(n) denotes the DAC input at clock cycle n. In the 1st clock four unit elements are selected. Then in the next clock the elements are selected from the first unused, that is the 5th element. If the last element is selected, DWA will start to select the 1st one again. DWA shapes the nonlinear errors with the first-order transfer function (1-z-1) [ 8],[11]. 3. DWA With Segmented Quantizer A two-step architecture for the internal quantizercan solve some of the problems arising Figure 1. Block diagram of the data weighted averaging (DWA) method The Data Weighted Averaging (DWA) Algorithm Multi bit quantization improves the stability and the signal to quantization noise performance of sigma-delta converters, but it also necessitates the use of dynamic element matching (DEM) to filter the nonlinearity error in the signal band . Data weighted averaging (DWA) is the most widely used DEM algorithms, due to its simplicity and low hardware overhead. Figure 1 shows block diagram of the data weighted averaging (DWA) method. The basic concept of DWA is to guarantee that each of the elements is used with equal probability for each digital input code. This is realized by sequentially selecting elements, 33 Figure 2. The DWA operation principle SSDR@SCIENCERECORD.COM Science Series Data Report Vol 4, No. 5;May 2012 from increasing the internal quantization levels beyond five bits. Since these architectures provide the digital data in two sections, coarse bits and fine bits, a logical way to interface with the DWA is to simply perform DWA independently on the coarse and fine DAC banks, as illustrated in Figure 3. The quantizer produces NC bits as the coarse signal and NF bits as the fine signal, for a total of N bits (N = NC+NF). Figure 4 shows a mathematical representation of the segmented architecture from Figure 3. The two-step quantizer resolves the NC coarse bits, and then subtracts this value from the input Figure 3. Block diagram of segmentated Σ∆ ADC and generates the NF fine bits from this signal. The gain of 2 ( N NC ) inside the quantizer represents a binary right shift to insure the correct place value of the bits, since the coarse bits are the NC most significant bits of an N bit signal. Since DWA shapes the error due mismatch unit element DAC with first order transfer function , the DWA blocks can be represented as (1-z-1) , as seen in Figure 4. The coarse and fine outputs are each applied to separate DACs using smaller, independent DWA circuits, reducing DWA complexity significantly. The coarse DAC transfer function is weighted by 2 ( N N C ) times that of the fine DAC to insure that the original place values are preserved. However, since this weighting depends on the size of the unit elements involved, a gain mismatch, 1- , will be present. The quantization noise, QC, present in both signals YC and YF , ideally will cancel when the coarse and fine signals are summed together at the modulator input. This result be the same as if a single DWA circuit with a single DAC had been in the feedback path. However, because of the gain Figure 4. Mathematical Block Diagram of Segmented ∆Σ ADC and will be transmitted to the output. When the coarse and fine signals are summed together, the quantization noise will not completely cancel while in the single-path method would completely cancel because of gain mismatch between the coarse and fine DAC banks in the segmented method. The non-canceled portion of the quantization noise will be added directly to the input signal, and thus be transmitted to the output of the ΣΔ ADC. The output, Y, of the ΣΔ ADC in Figure 4 can be written as : mismatch between the DACs, the coarse quantization 34 error will not completely cancel Y ( z ) Yc ( z ).2 N N C YF ( z ) (1) SSDR@SCIENCERECORD.COM Science Series Data Report Vol 4, No. 5;May 2012 where YC (z) 2(NNC)[QC (z) (X (z) (1 Z 1)YF )H(z)] 1 (1 )(1 Z 1)H(z) (2) Y F ( z ) QC ( z ) Q F ( z ) (3) by substituting (2) and (3) into (1), it is obtained as follows : Y (z) X ( z)H ( z ) 1 (1 )(1 z 1 ) H ( z ) QF ( z ) 1 (1 )(1 z 1 ) H ( z ) (4) (1 z 1 )(QC ( z ) QF ( z )) H ( z ) 1 (1 )(1 z 1 ) H ( z ) For comparison, a single-path approach would lead to: Y ( z) X ( z) H ( z) 1 (1 z 1 ) H ( z ) Q( z) (5) 1 (1 z 1 ) H ( z ) A comparsion of equations (4) and (5) show that the segmented system has error term ε · (QC − QF )(1-z-1) present in addition to normal quantization noise. The value of the mismatch term, ε, changes with each clock cycle due to the operation of the DWA. For realistic unit-element mismatch values, ε is small, but still large enough to significantly affect the SNR of the system. 4. value. So the mismatch, ε from equation (4) can be reduced by matching the average element values between DACs with the ratio of 2N-Nc : 1. The individual DAC element values are not important, as long as the average element values meet this ratio. In an attempt to measure and match the average element values between coarse and fine DACs, a start-up calibration procedure can be used as shown in Figure 5. During calibration, the connections from the coarse/fine ADC to the feedback path are broken and a single-bit path provides the modulator feedback. A one-bit quantizer is used because it is immune to the mismatch that plagues multi-bit quantizers. The modulator input is grounded and a fixed, DC test signal is presented to the coarse and fine DACs through the DWA circuitry. The output is sent to an averaging synck filter, and then measured for each DAC individually, with the other DACs disconnected from the circuit. The relative measurement between coarse and fine DAC banks can then be used to make the necessary adjustment to match the ratio of the average DAC element values. These individual measurements are then compared and used to calculate how much a single fine element must be adjusted to insure an average element size ratio of 2N-Nc : 1. A distinct advantage of the calibration method is the low complexity of the feedback path. For an 8-bit quantizer, two independent, 4-bit DWA implementations are required. Thus both the complexity and the timing delay of the digital feedback path are minimal. Calibration Method The mismatch term, ε, from equation (4) represents the deviation from the desired gain ratio of the coarse and fine DACs. Each DAC’s gain changes with every cycle due to the DWA operation. However, since ΣΔ ADCs use oversampling, the average gain of the DAC over time is more important than any instantaneous Figure 5. Block diagram of calibration method 35 SSDR@SCIENCERECORD.COM Science Series Data Report 5. Vol 4, No. 5;May 2012 The Noise-Shaped Requantization (ReQ) Method The mismatch error between coarse and fine banks can be noise shaped if the coarse quantization is performed within a digital Σ∆ modulator. This method was initially proposed in [9] for a Σ∆ DAC, then this method in [10] was proposed for Σ∆ ADC along DEM with gain of unity. This paper extends this concept to Σ∆ ADCs with DWA algorithm. The basic idea is to generate a new coarse signal with a digital Σ∆ modulator and use this coarse signal to generate a new fine signal. This insures that both the coarse and fine signals are individually noise shaped, which is performed in a way that causes the quantization error leakage to be noise shaped as well. Even though it does not completely cancel errors due to DAC mismatch, the quantization error noise power will be outside the signal band. The process is modeled in Figure 6. Figure 7 shows a mathematical representation of the ReQ architecture from Figure 6. The digital coarse and fine signals from the quantizer are first concatenated to form an N-bit signal. This signal is then requantized to NC bits using a digital first-order Σ∆ modulator. Then coarse signal is subtracted from the original N-bit signal to form the new fine signal, comprised of NF+1 bits. After requantization, the new coarse and fine signals become: YC ( z ) 2 ( N NC ) [Y ( z ) Q C ( z )(1 z Y F ( z ) QC ( z )(1 z 1 ) 1 )] Figure 7. Mathamatical block diagram of REQ method (6) noise-shaped away from the signal band. The output of the system with ReQ as in Figure 7 is derived as follows: (7) Y ' ( z ) Yc( z ).2 N N C Y F ( z ) Signals Y'C (z) and Y'F (z) then pass through independent DWA blocks and DACs and are summed at the input of the modulator. With first-order requantization (ReQ), the quantization error that is not completely cancelled due to coarse/fine DAC mismatch as in Equation (4) is 36 Figure 6. Block diagram for REQ method (8) where YC ( z ) 2 ( N NC ) [QC ( z ) ( X ( z ) (1 )(1 z 1 ) 2 N N C .YC ( z ) (9) Y F ( z )(1 z 1 )) H ( z )] SSDR@SCIENCERECORD.COM Science Series Data Report Vol 4, No. 5;May 2012 and YF ( z ) QC ( z ) Q F ( z ) (10) by substituting (9) and (10) into (8), it is obtained as follow: Y ' ( z) X ( z) H ( z) 1 (1 )(1 z 1 ) H ( z ) QF ( z) 1 (1 )(1 z 1 ) H ( z ) For the calibrated simulation, one of the fine DAC elements was adjusted to correct the gain ratio between the coarse and fine DACs. The simulation results for the calibrated system and the power spectrum density of its output signal are shown in figures 9 and 10 respectively. For the mismatch of 1% the results show that the calibrated system has a drop of only 0.5 dB in the SNR which is much lower than that of the segmented system. (11) ( .QC ( z ))(1 z 1 ) 2 H ( z ) 1 (1 )(1 z 1 ) H ( z ) which shows that that the coarse quantization noise leakage is first-order shaped. Simulations show that higher order REQ is not necessary, first order REQ sufficiently suppresses coarse/ fine mismatch errors below the noise shaped by the DWA within each DAC. 6. Simulation Results In order to verify the validity of the proposed calibration and REQ methods in this paper a second order sigma delta modulator with 8-bit quantizer (4 bit coarse, 4 bit fine) and an OSR of 30 has been simulated using MATLAB. Figure 8 shows the simulation results for both segmented and single-path systems versus various unit-element mismatch percentages. The element sizes were selected assuming that the coarse element percent mismatch be (24)1/2 times lower than the fine element mismatch due to the 24 sizing ratio. As shown in figure 8, with the addition of unit element mismatch, the overall SNR of the segmented system drops much faster than the single-path system. The results of simulation show the SNR of segmented system with 1% mismatch, is 20 dB lower than that of single-path system. Figure 8. Simulated results for segmented system Figure 9. Simulated results for calibration system 37 SSDR@SCIENCERECORD.COM Science Series Data Report Vol 4, No. 5;May 2012 Figure 10. Power spectrum density of output the Figure 11. Simulated Results for REQ system ∆Σ modulator using calibration method Figure 11 compares the performance of the segmentation and ReQ methods against the single-path method. Again, a (4+4)–bit quantizer, second-order modulator with an OSR of 30 was used, and also the coarse and fine percent mismatches were scaled assuming a 16:1 size ratio between coarse and fine elements. The figure 12 shows power spectrum density of the output signal for REQ system for the case of 1% fine element mismatch. The ReQ method achieves an average SNR of 106.337 dB, which is only 2 dB less than that of the full 8-bit DWA method however number of unit elements of DWA circuit much less than that of single-path method. The simulated results are summarized in Table 1. Figure 12. Power spectrum density of output the ∆Σ modulator using requantization method Table1: Simulated Results 38 SSDR@SCIENCERECORD.COM Science Series Data Report 7. Conclusion The DWA algorithm modulates the nonlinarity error of the DAC due to mismatch unit elements, moving the harmonic distortion out of the signal bandwidth. However each added bit of quantizer casuses an exponential increase in complexity of DWA and DAC circuitry. The segmented architecture with coarse/fine DAC and DWA combined with either the calibration or ReQ methods proposed in this paper allow for larger internal quantizers without the exponential increase in DWA logic, while still maintaining performance close to the single path system. References [1] S. R. Norsworthy, R. Schreier, and G.C. Temes, Delta-Sigma Data Converters, IEEE Press, 1997. [2] R. Schreier and G. C. Temes, Understanding Delta-Sigma Data Converters,Hoboken, NJ: Wiley, 2005. [3] R. Jiang and T. S. Fiez, “A 1.8V 14b Delta-Sigma A/D Converter with 4Msamples/s Conversion,” Digest of Technical Papers, IEEE 2002 International Solid-State Circuits Conference (ISSCC), vol. 1, pp 220-461, Feb. 2002. doi:10.1109/ISSCC.2002.993015 [4] R. Koch, B. Heise, F. Eckbauer, E. Engelhardt, J. A. Fisher, and F.Parzefall, “A 12-bit sigma-delta analog-to-digital converter with a 15-MHz clock rate, ” IEEE Journal of Solid-State Circuits, vol. 21, pp. 1003-1010, Dec. 1986. doi:10.1109/JSSC.1986.1052642 [5] M. Rebeschini, N. R. van Bavel, P. Rakers, R. Greene, J. Caldwell, and J. R. Haug, “A 16-b 160-kHz CMOS A/D converter using sigma-delta modulation, ” IEEE Trans. Circuits Syst, vol. 25, pp. 431-440, Apr. 1990. doi:10.1109/4.52167 [6] S. Lindfors and K. A. I. Halonen, “ Two-step Quantization in Multibit Delta-Sigma Modulators,” IEEE Transactions on. Circuits and systems II, vol. 48, no. 2, pp. 171-176, Feb. 2001. doi:10.1109/82.917785 [7] Y. Cheng, C. Petrie and B. Nordick, “A 4th-Order Single- Loop Delta-Sigma ADC with 8-Bit Two-Step Flash Quantization,” submitted to Proc. ISCAS 2004,pp. 1156-1159, Oct. 2004. doi:10.1109/ISCAS.2004.1328405 39 Vol 4, No. 5;May 2012 [8] R. T. Baud and T. S. Fiez, “Linearity Enhancement of Multibit AID and D/A Converters Using Data Weighted Averaging, ” IEEE Transactions on Circuits and Systems II. 119 vol. 42, pp. 753- 762, Dec. 1995. doi:10.1109/82.476173 [9] R. Adams, K. Nguyen, and K. Sweetland, “A 113-dB SNR Oversampling DAC with Segmented Noise-Shaped Scrambling,”IEEE Journal of Solid-State Circuits, pp. 1871-1878, Dec. 1998. doi:10.1109/4.735526 [10] Brent Nordic, Craig Petrie, and Yongjie Cheng, “ Dynamic Element Matching Techniques For Delta-Sigma ADCS With Large Internal Quantizers” Accepted to Proc. 2004 IEEE International Symposium on Circuits and Systems (ISCAS), vol. 1,pp. 653-656, May 2004. doi:10.1109/ISCAS.2004.1328279 [11] I. Galton, “Why Dynamic-Element-Matching DACs Work, ” IEEE Transactions on Circuits and Systems II: Express Briefs, vol. 57, no. 2, pp. 69 –74,February2010. doi:10.1109/TCSII.2010.2042131 [12] A. A. Hamoui and K. W. Martin, “High-order multibit modulators and pseudo data weighted-averaging in low-oversampling AS ADCs for broad-band applications,” IEEE Transactions on Circuits and Systems I, vol. 51, pp. 72-85, Jan. 2004. doi:10.1109/TCSI.2003.821291 [13] E. Najafi Aghdam , P. Benabes,J. Abbasszadeh “Completely first order and tone free partitioned data weighted averaging technique used in a multibit delta sigma modulator,” IEEE 19th European Conference on Circuit Theory and Design (ECCDT'09), Antalya, pp. 53-56, 2009. doi:10.1109/ECCTD.2009.5275129 [14] A. Lavzin, M. Kozak,G. Friedman “A Higher-Order Mismatch-Shaping Method for Multi-Bit Sigma-Delta Modulators,” In Proceedings of SoCC, pp. 267-270, 2008. doi:10.1109/SOCC.2008.4641525 [15] S. Zouari, H. Daoud, M. Loulou, P. Loumeau, N. Masmoudi ,“ High Order Cascade Multibit ΣΔ Modulator for Wide Bandwidth Applications,” International Journal of Electrical and Computer Engineering, 2007. SSDR@SCIENCERECORD.COM Science Series Data Report [16] E. Najafi Aghdam , P. Benabes, “Reducing multibit DAC circuits errors by a simplified dynamic element matching algorithm used in delta-sigma converters,” 15th ICEE - Iranian Conference on Electrical Engineering, 2007. [17] I. Fujimori, A. Nogi, and T. Sugimoto, “A multibit delta-sigma audio DAC With 120dB dynamic range,” IEEE Journal of Solid-State Circuits, vol. 35, no.8, pp. 1066-1073, August 2000. doi:10.1109/4.859495 [18] D. H. Lee and T. H. Kuo, “Advancing data weighted averaging technique for multi-bit sigma-delta,” IEEE Trans. Circuits Syst. II: Expr. Briefs, vol. 54, no. 10, pp. 838–842, Oct. 2007. doi:10.1109/TCSII.2007.901575 Vol 4, No. 5;May 2012 [24] K. L. Chan, J. Zhu, and I. Galton, “Dynamic Element Matching to Prevent Nonlinear Distortion from Pulse-Shape Mismatches in High-Resolution DACs,” IEEE J.Solid-State Circuits, vol. 43, no. 9, pp. 2067–2078 ,Sep. 2008. doi:10.1109/JSSC.2008.2001931 [25] Nevena Rakuljic, Member, and Ian Galton, Senior Member, “Tree-Structured DEM DACs with Arbitrary Numbers of Levels,”IEEE Transactions on Circuits and Systems II: Regular Papers, vol. 57, no. 2, 313-322, Feb 2010. doi:10.1109/TCSI.2009.2023931 [19] Alex Jianzhong Chen, Member, and Yong Ping Xu, Senior Member, “Multibit Delta-Sigma Modulator with Noise-Shaping Dynamic Element Matching,” IEEE Transactions on Circuits and Systems I: Regular Papers, vol. 56, and no. 6, pp.1125-1133,June2009.doi:10.1109/TCSI.2008. 2008485 [20] A. Fishov, E. Siragusa, J. Welz, E. Fogleman, and I. Galton, “Segmented Mismatch-Shaping D/A Conversion,” Proc. 2002 IEEE International Symposium on Circuits and Systems (ISCAS), vol. 4, pp. IV-679-IV-682, May 2002. doi:10.1109/ISCAS.2002.1010547 [21] M. R. Miller and C. S. Petrie, “ A Multibit Sigma-Delta ADC for Multimode Receivers,” IEEE J. Solid-State Circuits, vol. 38, no. 3, pp. 475-482,Mar.2003. doi:10.1109/JSSC.2002.808321 [22] C. Petrie and M. Miller, “A Background Calibration Technique forMultibit Delta-Sigma Modulators,” Proc. 2000 IEEE International Symposium on Circuits and Systems (ISCAS), vol. 2, pp. 29-32, May 2000. doi:10.1109/ISCAS.2000.856250 [23] I. Fujimori, and T. Sugimoto, “A 1.5 V, 41 mW dual-channel audio deltasigma D/A converter”, IEEE Journal of Solid-State Circuits, vol. 33, no.12, pp. 1863-1998, December 1998 40 SSDR@SCIENCERECORD.COM