(A) study on multichannel receivers with enhanced lane expandability and loop linearity

Abstract

Two types of serial data communication receivers that adopt a multichannel architecture for a high aggregate I/O bandwidth are presented. Two techniques for collaboration and sharing among channels are proposed to enhance the loop-linearity and channel-expandability of multichannel receivers, respectively. The first proposed receiver employs a collaborative timing scheme recovery which relies on the sharing of all outputs of phase detectors (PDs) among channels to extract common information about the timing and multilevel signaling architecture of PAM-4. The shared timing information is processed by a common global loop filter and is used to update the phase of the voltage-controlled oscillator with better rejection of per-channel noise. In addition to collaborative timing recovery, a simple linearization technique for binary PDs is proposed. The technique realizes a high-rate oversampling PD while the hardware cost is equivalent to that of a conventional 2x-oversampling clock and data recovery. The first receiver exploiting the collaborative timing recovery architecture is designed using 45-nm CMOS technology. A single data lane occupies a 0.195-mm2 area and consumes a relatively low 17.9 mW at 6 Gb/s at 1.0V. Therefore, the power efficiency is 2.98 mW/Gb/s. The simulated jitter is about 0.034 UI RMS given an input jitter value of 0.03 UI RMS, while the relatively constant loop bandwidth with the PD linearization technique is about 7.3-MHz regardless of the data-stream noise. Unlike the first receiver, the second proposed multichannel receiver was designed to reduce the hardware complexity of each lane. The receiver employs shared calibration logic among channels and yet achieves superior channel expandability with slim data lanes. A shared global calibration control, which is used in a forwarded clock receiver based on a multiphase delay-locked loop, accomplishes skew calibration, equalizer adaptation, and the phase lock of all channels during a calibration period, resulting in reduced hardware overhead and less area required by each data lane. The second forwarded clock receiver is designed in 90-nm CMOS technology. It achieves error-free eye openings of more than 0.5 UI across 9− 28 inch Nelco 4000-6 microstrips at 4− 7 Gb/s and more than 0.42 UI at data rates of up to 9 Gb/s. The data lane occupies only 0.152 mm2 and consumes 69.8 mW, while the rest of the receiver occupies 0.297 mm2 and consumes 56 mW at a data rate of 7 Gb/s and a supply voltage of 1.35 V.