MVMarijn Verbeke

Marijn Verbeke

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Marijn Verbeke has a total of 50 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • A Linear Modulator Driver With Over 70-GHz Bandwidth 21.8-dB Gain and 3.4-Vppd Output Swing for Beyond 120-GBd Optical Links

    by Shengpu Niu, Riaz Ahmed, Joris Lambrecht, Jakob Declercq, Marijn Verbeke, Michiel Verplaetse, Bart Moeneclaey, Cheng Wang, Joris Van Kerrebrouck, Nishant Singh, et al.

    This article investigates a linear differential modulator driver for optical communication links beyond 120 GBd. The driver was implemented in a 55-nm SiGe BiCMOS technology with an $f_T$ / $f_{\text{MAX}}$ of 330/370 GHz, featuring a traveling wave output stage driven by a lumped variable gain amplifier (VGA) and a predriver. The fabricated IC achieved a maximum low-frequency gain of 21.8 dB with more than 10-dB gain control capability and a 3-dB bandwidth of over 70 GHz. Power measurement results show the driver is able to deliver a constant output swing of 3.56 Vppd while keeping the THD below 2%, with the differential input swing varying from 0.3 to 1.03 Vppd at 1 GHz. In the large-signal measurement, the chip delivered 72, 106, 112, and 128 GBd pulse amplitude modulation-4 (PAM-4) signals with differential output swings of 4, 3.59, 3.434, and 3.425 Vppd, respectively. The chip’s total power consumption is 725 mW, resulting in a power efficiency of 2.83 pJ/bit. Overall, this is the first time an optical modulator driver reported the ability to accommodate such a wide input swing and demonstrated the operation at a data rate of 256 Gb/s. These features make it a desired candidate for beyond 120-GBd optical links.


  • Open Access

    Co-designed electro-optical integrated frontend circuits for high speed transceivers

    by Johan Bauwelinck, M. Verbeke, J. Lambrecht, M. Vanhoecke, L. Breyne, G. Coudyzer, J. Declercq, B. Moeneclaey, R. Ahmed, N. Singh, et al.

    Co-design of photonic and electronic chips has become a necessity to realize high-performance transceivers. This paper will illustrate a number of recent developments of high-speed transceiver circuits including broadband driver amplifiers and transimpedance amplifiers.


  • Open Access

    End-to-End 1024-Port Optical Packet Switching With 25 Gb/s Burst-Mode Reception for Data Centers

    by Nikolaos Terzenidis, Apostolos Tsakyridis, George Giamougiannis, Joris Van Kerrebrouck, Marijn Verbeke, Guy Torfs, Miltiadis Moralis-Pegios, Nikos Pleros

    Despite the fact that Optical Packet Switching (OPS) emerges as a promising solution for future Data Center (DC) networks, towards increasing capacity and radix, while retaining sub-μs latency performance, the requirement for ultra-fast burst-mode reception has been a serious restraining factor. We attempt to overcome this limitation and demonstrate, for the first time to our knowledge, an end-to-end optical packet switch link through the 1024-port 25.6 Tb/s Hipoλaos OPS, featuring burst-mode reception with <; 50 ns locking time. The switch performance for unicast traffic is evaluated via Bit-Error-Rate measurements and error-free performance at 10 9 is reported for all validated port-combinations, with a mean power penalty of 2.88 dB. Moreover, multicast flows from two different ports of the switch were successfully received validating the architecture's credentials for efficient multicast packet delivery. Taking one step further towards a realistic evaluation of an OPS-enabled DC, a simulation analysis was conducted, proving that low-latency performance, including the burst-mode reception time-overhead, can be successfully realized in a Hipoλaos-switched DC with up-to 100% throughput for a variety of traffic profiles.


  • The Truth About 2-Level Transition Elimination in Bang-Bang PAM-4 CDRs

    by Marijn Verbeke, Guy Torfs, Pieter Rombouts

    Reception of 4-level pulse amplitude modulation (PAM-4) requires a clock and data recovery (CDR) circuit, typically implemented by a PLL-like structure. An essential block in such a CDR is the phase detector which should detect whether the recovered clock leads or lags the incoming data edges. In typical implementations an incoming data edge is detected by sensing whether the incoming waveform crosses a data threshold level. However, there is some ambiguity in detecting the incoming data edge because PAM-4 modulation has 3 thresholds. If the waveform crosses multiple threshold levels, the level crossings will occur at different time instants due to the finite rise/fall time of the incoming waveform. In this work, we first analyze qualitatively and quantitatively CDR systems that use one threshold for phase adjustment. Here, eliminating the 2-level transitions decreases the amount of jitter injected by the phase detector. However, the available transitions for phase adjustment are also reduced, which lowers the CDR's robustness. Secondly, for CDR systems using three thresholds, a combination of two techniques: majority voting and elimination of 2-level transitions is investigated. We prove that in this case, the elimination of 2-level transitions is not needed and even gives a worse performance when implemented.


  • A 25 Gb/s All-Digital Clock and Data Recovery Circuit for Burst-Mode Applications in PONs

    by Marijn Verbeke, Pieter Rombouts, Hannes Ramon, Jochem Verbist, Johan Bauwelinck, Xin Yin, Guy Torfs

    The upcoming 100 Gb/s links in the next-generation ethernet passive optical networks will be based on four channels of 25 Gb/s. The corresponding transceivers in these optical links require a high-speed clock and data recovery circuit to extract a synchronous clock and recover the received data. To achieve a sufficiently fast settling time for 25 Gb/s burst mode upstream applications in passive optical networks (PONs), we introduce an architecture of the first 25 Gb/s all-digital clock and data recovery circuit (AD-CDR). Thanks to the implementation of a digital loop filter, our AD-CDR avoids the need of a system clock or a start-of-burst signal. This circuit is implemented in a 40-nm CMOS process and has a very compact active chip area of only 0.050 mm 2 . Furthermore, the performance of the burst-mode operation of our AD-CDR in an optical setup is measured and reported, resulting in a burst-mode lock time of 37.5 ns and consuming only 46 mW.


  • A 1.8-pJ/b, 12.5–25-Gb/s Wide Range All-Digital Clock and Data Recovery Circuit

    by Marijn Verbeke, Pieter Rombouts, Hannes Ramon, Bart Moeneclaey, Xin Yin, Johan Bauwelinck, Guy Torfs

    Recently, there has been a strong drive to replace established analog circuits for multi-gigabit clock and data recovery (CDR) by more digital solutions. We focused on phase locked loop-based all-digital CDR (AD-CDR) techniques which contain a digital loop filter (DLF) and a digital controlled oscillator (DCO) and pushed the digital integration up to a level where our DLF is entirely synthesized. To enable this, we found that extensive subsampling can be used to decrease the speed of the DLF while maintaining a good operation. Additionally, an Inverse Alexander phase detector and a 5.5-bit resolution DCO complete the AD-CDR architecture. As a result of the low complexity and digital architecture, the AD-CDR occupies a compact active chip area of 0.050 mm 2 and consumes only 46 mW at 25 Gb/s. This is the smallest area and the lowest power consumption compared with the state-of-the-art. In addition, our implementation is highly tunable due to the synthesized logic, and supports a wide operating range (12.5-25 Gb/s), which is a significantly larger range compared with the previous work. Finally, thanks to our digital architecture, the power dissipation decreases linearly while moving to the lower speeds of our operating range. This is in contrast with the most prior work, making our design truly adaptive.


  • Open Access

    Inverse Alexander phase detector

    by M. Verbeke, P. Rombouts, X. Yin, G. Torfs

    An improved bang‐bang phase detector (PD) for multi Gb/s clock and data recovery (CDR) circuits is presented. The proposed PD is based on inverting the Alexander PD. In a typical subsampled CDR circuit, this Inverse Alexander PD results in a ten times better bit error rate (BER) compared with the conventional Alexander PD. Additionally, in the case of duty‐cycle distorted input data, this Inverse Alexander PD can even reach 20 times better BER compared with the conventional Alexander PD.


  • The 40Gbps cascaded bit-interleaving PON

    by A. Vyncke, G. Torfs, C. Van Praet, M. Verbeke, A. Duque, D. Suvakovic, H.K. Chow, X. Yin


  • Voltage Controlled Oscillators for 40Gbit/s Cascaded Bit-Interleaving PON

    by Arno Vyncke, Guy Torfs, Marijn Verbeke, Christophe Van Praet, Xin Yin, Hungkei Chow, Dusan Suvakovic, Alex Duque

    Technologies such as the Internet-of-Things and cloud services demand dynamic bandwidth allocation flexibility, which is not offered by the currently deployed solutions. The Bit-Interleaving PON (BiPON) and its cascaded extension the Cascaded Bit-Interleaving PON (CBI-PON) offer a solution that allows to increase bandwidths, reduce power consumption and have a much more flexible dynamic bandwidth allocation scheme. CBI-PON consists of multiple levels of BiPON with different line rates. For each of these line rates, clock-and-data recovery must be performed, which requires a set of different Voltage Controlled Oscillators (VCOs). This paper presents the VCOs designed for the CABINET chip, an implementation of a CBI-PON network device, allowing clock-and-data recovery for 40Gbit/s, 10 Gbit/s and 2.5 Gbit/s line rates.

  • An 8-phase 10GHz voltage controlled ring oscillator for 40 Gbit/s BiPON clock-and-data recovery

    by Arno Vyncke, Guy Torfs, Marijn Verbeke, Xin Yin

    New technologies such as cloud services and the Internet of Things have led to a demand for higher bandwidths, as well as the need for dynamic bandwidth allocation. This flexible bandwidth is offered by the BiPON protocol. By using BiPON, a significant power consumption reduction can be realized at the receiver side by using a 1:4 sub-sampling clock-and-data recovery circuit (CDR). To realize this CDR, the phase detector needs 8 phases of a 10GHz sampling clock. In this paper, we present the design and measurements of such a VCO, which has a tuning range of 6.57 GHz to 10.61 GHz and a gain of 250 MHz/V, while consuming about 30mW.