13 January 2023 in IEEE Transactions on Circuits and Systems I: Regular Papers
This paper presents a high data rate wireline Discrete Multitone (DMT) transceiver capable of operating at 252 Gbps and higher for short range chip-to-chip and chip-to-module applications. The use of DMT allows for higher spectral efficiency than PAM SERDES systems, using similar bandwidth as proposed 224 Gb/s PAM-4 systems but achieves higher data rates. Data rates have improved over other published wireline DMT designs with the use of narrow sub-channels with both adaptive bit-and power-loading along with a combined transmit and receive side equalization. Simulation and measured results show that the transmit side pre-compensation has improved the BER by a factor of $10^{4}$ at 250 Gb/s. The system is measured over a channel with a loss of 18 dB at 50 GHz, showing a data rate of 252 Gb/s with a BER of $10^{-6}$ , and a data rate of 400 Gb/s with a BER of $10^{-2}$ .
31 August 2021 in IEEE Transactions on Very Large Scale Integration (VLSI) Systems
Sigma–delta ( $\Sigma \Delta$ ) digital-to-analog converters (DACs) are commonly used to realize programmable dc voltage generators in built-in self-test (BIST) schemes, largely on account of their high linearity. Unfortunately, such $\Sigma \Delta $ -DACs require large amounts of digital memory and a reconstruction filter with a large silicon area footprint. In this article, a segmented DAC architecture is proposed. The proposed architecture is realized using two sub-DACs, where both are $\Sigma \Delta $ -based. This DAC provides two advantages: filter footprint size reduction and memory savings, thereby allowing for a simpler BIST solution. The benefits of using the segmented $\Sigma \Delta $ -DAC are shown using two experimental prototypes. The first is an IC prototype design using the TSMC 65-nm CMOS technology. It will be shown that the IC achieves 12 bits of resolution from 1020 memory elements, whereas an unsegmented $\Sigma \Delta $ -DAC uses 4095 elements to obtain the same resolution. This is about a 75% savings in memory elements. The segmented prototype occupies 0.5 mm 2 of silicon area, whereas the unsegmented design occupies 0.77 mm 2 for the same resolution. A 35% silicon area savings compared with its unsegmented counterpart. A second prototype implements the segmented $\Sigma \Delta $ -DAC architecture using discrete components. The discrete prototype achieves 16 bits of resolution from 1020 memory elements, whereas the unsegmented counterpart uses 65 535 bits for the same resolution. This is a 98% saving in memory elements.
11 March 2024 in IEEE Transactions on Instrumentation and Measurement
This paper proposes an adaptive all-digital background compensation of timing skew mismatches in time-interleaved analog-to-digital converters (TI-ADCs). The proposed approach uses the orthogonality between the desired component representing ideal samples of the TI-ADC input signal and the timing skew mismatch errors. Our technique is the first attempt to use a source separation mechanism to recover the desired component in the digital output of TI-ADC. A finite impulse response (FIR) filter implementing the Hilbert transform and two separate adaptive FIR (A-FIR) filters are exploited to implement this recovery process. The proposed technique works at frequencies close to the full admissible Nyquist frequency range, and its operation does not depend on the number of sub-ADCs (SADCs). This method operates without any extra SADC. The main cost introduced by the proposed compensation technique is the requirement of two adjustable FIR filters. This cost is proportional to the number of taps in these FIR filters that can be optimized according to the specific application. This technique is validated through various behavioral simulations and a field-programmable-gate-array (FPGA) implementation. With the proposed method, the signal-to-noise and distortion ratio (SNDR) is enhanced from 23.06 dB to more than 66.82 dB in a 12-bit TI-ADC when the standard deviation of timing skew mismatches is 10% of the TI-ADC’s sampling interval.
4 November 2021 in IEEE Transactions on Circuits and Systems II: Express Briefs
This brief presents a fully integrated energy-efficient memory-based direct digital frequency synthesizer (DDFS). To overcome the limitation of using an accumulator and a Read-Only Memory (ROM) Lookup Table (LUT) within a DDFS, the proposed design utilizes a 6-bit memory with a configured mode of operation to synchronize the digital output clock with the input sampling clock. The proposed DDFS is able to tune the output phase and frequency once a phase rotator is used in the input clock path. To realize the proposed DDFS, a 6-bit memory in front-end and some analog blocks in back-end are employed. The prototype DDFS achieves jitter below 12.7ps $_{rms}$ integrated over a 100MHz bandwidth with 28.25Hz resolution while consuming less than 3mW at sampling clock 3.12GHz. The proposed DDFS is fabricated in a 0.9V TSMC 7nm CMOS process and occupies a core area of only 0.014mm 2 .
14 February 2020 in IEEE Open Journal of Circuits and Systems
Settling time is an important performance metric in digital-to-analog converters (DACs) that are used for dynamic calibration applications. To obtain an area efficient DAC design, periodic sequences are generated from sigma-delta modulators (ΣΔ Ms) using software, stores the sequence in memory and then passes it to a low-pass filter (LPF). In this brief, the sigma-delta (ΣΔ) bitstream generated in software will be optimized such that it yields a system that settles faster and is more area efficient. Two optimization routines are used, namely, serial method and random method algorithms. The random method is chosen to optimize the ΣΔ bitstreams for its computational simplicity. It will be shown, that using the random method routine provides at least 46 improvement in settling time performance of a 12-bit DAC, used in the paper as a design example. Moreover, it offers silicon area savings by at least 10% for resistor components and 35% for capacitor components. Similar results were obtained for 8 to 11 bits of resolution.
8 July 2020 in IEEE Transactions on Circuits and Systems II: Express Briefs
New charge-steering DFEs are proposed and simulated in 55-nm CMOS for use in wireline transceivers. Four new methods of applying DFE taps demonstrate significant power savings of over 43% in the clock drivers and 28% in the total DFE compared to charge-steering DFEs using traditional taps. The power savings are achieved by reducing the number of switches in the taps and using pMOS transistors to apply the taps, which complements the reset state of charge-steering latches. Ultimately, a new method of applying charge-steering DFE taps using a variable resistor as the tap coefficient is presented. As a result, clock generation and routing for the DFE has been made significantly easier. The new topologies also make the charge-steering DFE FinFET friendly by eliminating stacked switches, which allows operation from a low supply voltage. All results are demonstrated analytically and verified by post-layout simulations. The new equalizers are shown to recover data at 56-Gb/s, with a BER of <10-12 across a channel with 13-dB of loss at Nyquist.