Digital Radio Receiver

1. Introduction

The evolution of the digital wireless communication market is prompting the development of new receivers capable to support new services and applications. In such systems, early Analog-to-Digital conversion at Radio-Frequencies (RF) or Intermediate-Frequencies (IF) permits digital control and programmability of both the gain and the filter coefficients of the IF stage. This enables the receiver to support multiple communication standards with different modulation schemes, carrier frequencies, channel bandwidths, etc…

The models and techniques presented in the paper are demonstrated through the practical realization of two integrated BP-S?M prototypes in CMOS 0.8 µm technology. Both prototypes have been fabricated and experimentally characterized. One of them uses a single-loop fourth-order architecture and is capable of digitizing signals up to 1.63 MHz with 10.5-bit resolution and 60 mW power consumption from a 5 V supply voltage (1.63 MHz at 10.5-bit at 60 mW). The other modulator is a single-loop second-order architecture featuring, 1.63 MHz at 8-bit at 42 mW, also from a 5 V supply voltage. In both cases, fully-differential regulated-folded cascode memory cells, and fully-differential current-mode buffers are employed to keep SI errors under control, on the one hand, and to attenuate the influence of bonding pad parasitics, on the other. Measurements show that correct noise-shaping filtering is achieved with a sampling frequency of up to 16 MHz, thus demonstrating the possibility of using SI BP-S?Ms in narrowband high frequency communication systems.

2. Band-pass S? architectures for AM digital radio receivers

Fig. 1 shows the conceptual block diagram of a single-loop BP-S?M with a 1-bit quantizer. Assuming that the quantization error is modeled as white, additive noise , the z-domain output can be expressed as:(1)Y(z)=STF(z)X(z)+NTF(z)E(z),where STF(z) and NTF(z) represent the signal transfer function and the noise transfer function, respectively. In a BP-S?M, NTF is of the band-stop type and has ideally the following expression:(2)NTF(z)=(1+z-2)Lwith 2L being the loop order. On the other hand, STF(z) is of the all-pass type. By making z=exp(j2pf/fs), where fs is the sampling frequency, it can be seen that NTF(z) has L transmission zeroes at fs/41, and that the filtering around this frequency is actually of the band-stop type. Fig. 2(a) shows the filtering performed by a BP-S?M. The input signal is allowed to pass while, at the same time, most of the quantization noise power is 'shaped' so that is pushed out of the signal band. This is illustrated through an ideal modulator output spectrum in Fig. 2(b).

Fig. 1: Conceptual block diagram of a 2L-order, N-bit BP-S?M.

Fig. 2: Noise shaping in BP-S?Ms. (a) Filtering functions. (b) Ideal modulator output spectrum for L=2.

The in-band quantization noise power can be calculated by integrating the output power spectral density within the band,(3)where SQ=?2/(12fs) is the power spectral density of the quantization noise error, ? the quantization step, Bw the signal bandwidth and M=fs/(2Bw) is the oversampling ratio. From Eq. (3), and assuming that the modulator input is a sinewave of amplitude A, the Signal-to-Noise ratio (SNR) at the output results in:(4)

The modulator Dynamic Range (DR) is obtained by making A=?/2 in the above expression, yielding:(5)

Expressions (4) and (5)(4) and (5) ...