15.2 A 4.5mW CT self-coupled ΔΣ modulator with 2.2MHz BW and 90.4dB SNDR using residual ELD compensation

Abstract

A high-dynamic-range (DR) CT ΔΣ modulator is required to relax the analog front-end filter design for wireless communication applications. To achieve high resolution (DR>90dB) and low power dissipation (FoMs>170dB), architecture selection and circuit techniques are the main design issues. In [1], a CT ΔΣ modulator embedded with a 2 nd -order active filter and VGA is reported to extend the DR. However, the additional active filter results in a complicated architecture as well as extra area that is not preferred in advanced processes. An alternative method that improves the DR is to adopt the self-coupled noise-injection technique introduced in [2] to increase by one the order of the noise transfer function (NTF). Unfortunately, it requires an accurate clock cycle delay, which is only available in the DT ΔΣ modulator. Apart from DR considerations, power efficiency is still limited by the building block design. Conventional excess loop delay (ELD) compensation [1,3,4] is implemented by an inner DAC, which increases the power consumption and loads the last integrator with a large parasitic capacitance, especially for a multi-bit modulator. Therefore, a high-bandwidth opamp is required in the last integrator to alleviate the phase delay of the loop filter. Furthermore, to address the nonlinearity of a multi-bit DAC, 2 nd -order dynamic element matching (DEM) is used in [5] to reduce data-dependent switching. The SFDR is still limited to 90dB. In this paper, CT self-coupling (CTSC), residual ELD compensation, and DAC linearity enhancement techniques are introduced to overcome these challenges. Our CT ΔΣ modulator achieves an SNDR of 90.4dB with an FoMs (SNDR) of 177.3dB in a 2.2MHz bandwidth.In Fig. 15.2.1 (a), the modulator, clocked at 140MHz, employs a 4 th -order feed-forward (FF) architecture with a 4b asynchronous successiveapproximation register (ASAR) quantizer (QTZ) [3] and advancing data-weighted averaging (ADWA) [6]. The high-gain loop filter provides strong noise-shaping capability to suppress the noise and nonlinearity at the back-end stages. The high-speed FF path for the S -1 term [4], realized by R f4 and R 4 , is used for robust stability. The CTSC technique is used to enhance the SQNR since the selfcoupling path forms an additional 1 st -order integrator. In the last integrator, the positive feedback implemented by C 4 and R 5 provides a 2 nd -order transfer function. The equivalent linear model is shown in Fig. 15.2.1 (b). To realize the self-coupling in the S-domain, Z -1 is transformed to 1/(1+KS), where K is determined by the coefficient synthesis. To reuse the opamp in the last-stage integrator, S/(1+KS) is applied to keep the same transfer function as the self-coupling path. By choosing C 3 =C 4 , the SC term is cancelled. Therefore, the expression, 1/(1+KS) can be derived by the product of the transconductance part (C 4 R 5 ) and the integration path (C 3 ). With the CTSC technique, in-band SQNR is improved by an amount that depends on the loop filter order, OSR, and QTZ-bit number. In this wor...

AI-Managed Cognitive Radio Digitizers

“…This idea -originally proposed in [156] to implement hybrid Pipeline-Σ∆ ADCs -can also increase the reconfigurability and programmability of the resulted ADC. Indeed, there has been a number of hybrid Σ∆M-Nyquist ADCs featuring a competitive performance in very diverse application scenarios [82], [157]- [166]. In the majority of cases, the basic strategy followed by hybrid Σ∆M/Nyquist ADCs consists of replacing (power and area)-hungry Flash quantizers, by another type of Nyquist-rate ADCs, such as pipeline [156], [157], [161], [164], two-step flash [161], SAR [82], [159], [163], [165], [166], cyclic [158], [160] and integrating ADC [162].…”

Estimating Non-Ideal Effects within a Top-Down Methodology for the Design of Continuous-Time Delta-Sigma Modulators

“…To solve these problems, many papers suggest different approaches of DSM architecture: The architecture with a feedforward ELD compensation path by residual signal can allow the specification of the feedback DAC to be relaxed [10]. A mixed CT/DT 2-1 cascaded DSM architecture is proposed to achieve both high-speed operation and stability [11].…”

5.2 A 118dB-PSRR 0.00067%(−103.5dB) THD+N and 3.1W fully differential class-D audio amplifier with PWM common-mode control

“…As a result, the internal nodes of the loop filter are input-signal independent and the non-linearity of the loop filter has less effect on the linearity of the CDA. [6]. The summation of signals is conventionally performed by an active summing amplifier [1].…”