Expand this Topic clickable element to expand a topic
Skip to content
Optica Publishing Group

Silica micro-rod resonator-based Kerr frequency comb for high-speed short-reach optical interconnects

Open Access Open Access

Abstract

Conventional data center interconnects rely on power-hungry arrays of discrete wavelength laser sources. However, growing bandwidth demand severely challenges ensuring the power and spectral efficiency toward which data center interconnects tend to strive. Kerr frequency combs based on silica microresonators can replace multiple laser arrays, easing the pressure on data center interconnect infrastructure. Therefore, we experimentally demonstrate a bit rate of up to 100 Gbps/λ employing 4-level pulse amplitude modulated signal transmission over a 2 km long short-reach optical interconnect that can be considered a record using any Kerr frequency comb light source, specifically based on a silica micro-rod. In addition, data transmission using the non-return to zero on-off keying modulation format is demonstrated to achieve 60 Gbps/λ. The silica micro-rod resonator-based Kerr frequency comb light source generates an optical frequency comb in the optical C-band with 90 GHz spacing between optical carriers. Data transmission is supported by frequency domain pre-equalization techniques to compensate amplitude–frequency distortions and limited bandwidths of electrical system components. Additionally, achievable results are enhanced with offline digital signal processing, implementing post-equalization using feed-forward and feedback taps.

© 2023 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement

1. Introduction

The growing popularity of online services like video streaming, large-scale data applications including cloud storage and computing, and video call services is driving an unprecedented increase in the bandwidth requirements of data centers (DC). To meet the growing demand for bandwidth, DCs must evolve towards higher performance and throughput while improving spectral efficiency (SE) and reducing power consumption [1]. Intensity modulation and direct detection (IM-DD) are still promising schemes in intra- and inter-data center interconnects (DCI) thanks to their low latency, high reliability, and reasonable cost performance [2]. The transmission of multiple data streams, using wavelength division multiplexing (WDM), allows scaling of the optical interconnect density in DCs and is based on arrays of discrete wavelength laser sources [14]. The spectral efficiency of laser-based transmission systems suffers from the uncertainty of individual radiation frequencies [5]. In this context, an optical frequency comb (OFC) that generates frequency-locked carriers with fixed frequency spacing of the order of 10–100 GHz can replace laser optical carriers from individual sources in WDM links, providing a reduction in transmitter energy consumption, as well as increasing spectral efficiency [3,6]. One promising and cost-effective technique is OFC generation in a silica whispering gallery mode resonator (WGMR) manufactured on a silica micro-rod platform (silica micro-rod WGMR-based Kerr OFC light source). An OFC is generated by pumping a high-quality (high Q-factor) optical resonator with Kerr nonlinearity using a single continuous-wave (CW) laser source [711]. While optical data transmission has been well studied and demonstrated using Kerr-OFC in integrated resonators that show even terabit communications [5,10,12], the experimental demonstration of high-speed data transmission based on OFCs generated specifically in micro-rod resonators has, to the best of our knowledge, not been demonstrated until now. Micro-rod resonators have several advantages compared to integrated resonators. First, the effective coupling of the pumping light into integrated resonators is complicated and fixed when integrated on a chip without the possibility of changing it [13]. In addition to the possibility of fine-tuning the coupling conditions (between the resonator and tapered fiber), micro-rod resonators also have the advantage of fast and simple fabrication using laser-machining techniques, which simultaneously shape and polish a quartz rod and allow the fabrication of micro-rod resonators with a user-defined diameter (changing the spacing between generated harmonics), thickness, and curvature to be shaped with ±10 µm precision and Q∼5 × 108 in a couple of minutes [14]. On the other hand, producing integrated resonators as microring is rather complicated. It is required to grow nanometer-scale waveguides as thick as 500 nm without disrupting the integrity of the waveguide [15,16]. Let's compare bulk resonators as microspheres and micro-rods. The latter can offer spacing between generated carriers below 100 GHz. This spacing is almost impossible to achieve with microspheres due to the integrity issues of a microsphere at diameters higher than 660 µm. Generated carrier spacing below 100 GHz is attractive for WDM applications following ITU-T spectral grid.

In this paper, we experimentally demonstrate a short-reach optical interconnect IM/DD link powered by a silica micro-rod WGMR-based Kerr OFC light source. Digital equalization techniques, such as a linear equalizer with feed-forward (FF) and feedback (FB) taps, are used to improve signal quality. The record bitrate achieved is 60 Gbps/λ using non-return to zero (NRZ) on-off keying (OOK) modulated signals and 100 Gbps/λ using 4-level pulse amplitude modulated (PAM-4) signals for transmission over a 2 km single-mode fiber (SMF) link. The rest of the paper is structured as follows. Section 2 describes the manufacturing and characteristics (such as size, curvature radius, and Q-factor) of the silica micro-rod WGMRs (fabricated on one fused quartz rod) used for the Kerr-OFC light source. Section 3 presents the experimental setup of the designed silica micro-rod WGMR-based Kerr-OFC light source and shows the output spectrum together with the used micro-rod WGMR. In addition, we show the stability of the generated Kerr-OFC over 20 hours. Also, the experimental setup for short-reach optical interconnects is discussed. Further, Section 4 demonstrates 50 and 60 Gbaud data transmission employing NRZ-OOK and 50 Gbaud data transmission using PAM-4 with and without post-equalization for both modulation formats. Finally, Section 5 concludes the paper.

2. Manufacturing and characteristics of silica micro-rod WGMR-based Kerr-OFC light source

Micro-rod fabrication begins with preparing a sample of a fused quartz rod having a cylindrical shape. The diameter of the cylindrical rod is selected depending on the desired microresonator free spectral range (FSR) between the generated comb carriers. Figure 1(a) shows a micro-rod fabrication setup using a carbon dioxide (CO2) laser. The fused quartz rod is attached to a spindle stage, which includes an air-cushioned lathe with less than 100 nm vibrations, in a fixed position to rotate the micro-rod. However, a galvanometer and an F-theta zinc-selenite (ZnSe) lens are used to position the laser beam on a micro-rod. The fused quartz rod's rotation occurs perpendicular to the transmitted laser beam. For the first step, a rotating fused quartz rod is illuminated for a few seconds with a CO2 laser beam focused through a ZnSe lens that selectively removes material by ablation to ensure the rotational axis of the mounted rod is parallel to the rotational axis of the motor. In the next step, a microresonator is produced by applying the CO2 laser at different positions in quick succession along the rod's axis. The cutting process actively generates dust from the evaporated material since the top layer of the fused quartz rod is removed. After the primary process, the cut region is exposed to the laser beam, which significantly improves the Q factor of the microresonator. Adhering to the fabrication protocol and following the cutting configurations – laser power, irradiation, duration, rod rotation speed, and beam positioning - allows the creation of almost identical samples repeatedly. The method allows for the repeated production of several rods with the same parameters in a relatively short period of time (∼ 5 min).

 figure: Fig. 1.

Fig. 1. (a) The fabrication of fused quartz micro-rod WGMRs using CO2 laser machining, where the spindle stage is used to rotate and a galvanometer is used to position the laser beam on the ZnSe lens. (b) The captured image of a fabricated micro-rod with a 700- µm diameter and 520- µm gap between WGMRs (Res1 to Res5). Images of micro-rod WGMRs with curvature radius of 250 µm, 250 µm, 200 µm, 175 µm, and 150 µm from the first resonator (Res1) to the fifth resonator (Res5). (c) The captured images of individual micro-rod resonators (Res1 to Res5) indicating different micro-rod curvature radii with white circles.

Download Full Size | PDF

Several WGMRs with the same or slightly varying parameters can be fabricated on one fused quartz rod. Figure 1(b) shows five microresonators produced on a single fused quartz rod with a 520 µm gap between the WGMRs, the same micro-rod diameter D = 700 µm but with a different curvature radius r. The curvature radius for each of the five WGMRs is calculated mathematically using image measurements taken with a visible light camera with a resolution of 320 × 240 pixels; please see Fig. 1(b,c). During manufacturing, five WGMRs on the fabricated fused quartz micro-rod are obtained with the curvature radius r of 250 µm, 250 µm, 200 µm, 175 µm, and 150 µm (depicted as Res1 to Res5, respectively), see Fig. 1(c). The curvature radius allows fine-tuning the Q-factor of the resonator. Although light loss due to scattering from the roughness of the surface of the micro-rod reduces the Q-factor, the most critical factor is the curvature of the resonator [17].

The curvature helps to confine and focus light within the resonator mode. Therefore, the curvature of the resonator sidewalls is controlled during CO2 laser machining to obtain an ultrahigh optical Q-factor and avoid crossings between different mode families [8]. The second micro-rod WGMR (Res2, D = 700 µm and r = 250 µm) is used in Section 3 to obtain Kerr-OFC with a mode spacing of about ∼90 GHz (89 GHz). This micro-rod WGMR is chosen as the combination of the resonator diameter D = 700 µm and the curvature radius r = 250 µm ensures the highest measured Q-factor of 2.6 × 107 compared among our 5 fabricated micro-rod WGMRs on the fabricated fused quartz micro-rod. Our achieved Q-factor is lower compared to [14,1820], comparable to microspheres (∼ 107-109), but higher than integrated resonators ∼105-106 [21]. The reason for choosing the second micro-rod is based not only on the best Q-factor (impacted by a combination of micro-rod diameter and curvature radius) among five fabricated micro-rods but also on experimental observations. They indicated that a comb achieved in the second micro-rod was not fluctuating and ensured lower noise, making it eligible for data modulation. The combs in other resonators greatly suffered from stimulated Brillouin scattering (SBS) noise, which practically impaired obtained carriers. That could be compared to the modulation of a carrier signal, and such carriers cannot be used for data modulation.

During this experiment, we did not measure the micro-rod dispersion. However, it is important to analyze the effect of different curvature radii on the dispersion of the micro-rod resonator to see if it will negatively affect the OFC generation. Therefore, we added a numerical calculation of a micro-rod dispersion based on fabricated micro-rod parameters such as diameter, material, and curvature radius in this work. Figure 2 shows the fabricated micro-rod resonator characteristics simulated in the COMSOL Multiphysics software (finite element method) with the 2D axisymmetrical module, electromagnetic waves, frequency domain physics, and eigenfrequency study. The simulations for dispersion calculations are done iteratively, following the method laid out in [22], and it takes 10 iterations for the dispersion to converge.

 figure: Fig. 2.

Fig. 2. The simulations of the silica micro-rod resonators with D = 700 µm and different curvature radii. The simulation consists of a silica micro-rod resonator and air surrounding it. (a) The mesh for the simulation domains (the approximate mode area has a finer mesh, (b) the cross-section of the fundamental TE mode with the normalized absolute value of the electrical field at a wavelength of about 1550 nm. (c) The simulated dispersion for silica micro-rod resonators with curvature radii 50, 100, 150, and 200 µm.

Download Full Size | PDF

We compare four resonators with the same diameter of 700 µm but different curvature radii r. The results in Fig. 2 show that the curvature radius does not greatly impact the resonator's dispersion. The dispersion parameter slightly increases when the curvature radius is decreased. All the resonators are in an anomalous dispersion regime at the pumping laser's wavelength of 1550 nm.

3. Experimental setup of silica micro-rod WGMR-based Kerr-OFC light source for high-speed short-reach optical interconnects

3.1 Experimental setup of the designed silica micro-rod WGMR-based Kerr-OFC light source

The setup used for the generation of micro-rod WGMR-based Kerr-OFC is shown in Fig. 3. The tapered fiber (TP) is fabricated from a non-zero dispersion-shifted fiber (NZ-DSF), as shown in our previous work [23], and the silica micro-rod WGMR is enclosed in a box for dust and airflow prevention. The humidity inside the enclosure box is reduced and maintained below 20% using a silica gel desiccant. In addition, both the pumping source and enclosure box are positioned on a vibration isolation system breadboard table to minimize external low-frequency vibrations.

 figure: Fig. 3.

Fig. 3. Experimental setup illustrating the developed silica micro-rod WGMR-based Kerr-OFC as a light source for optical communications.

Download Full Size | PDF

A continuous wave laser (CW, Agilent 81989A) with a linewidth of 100 kHz, optical output power of +6 dBm at λ=1555.46 nm and relative intensity noise (RIN) of -145 dB/Hz serves as the pumping source. An optical power splitter (PS) with a 50/50 coupling ratio divides the light from the laser into two equal optical power paths – the clockwise pump and the counter-clockwise pump. Clockwise and counter-clockwise optical power paths at the same wavelength (λ=1555.46 nm) pump the micro-rod WGMR from both sides. First, the light in each optical path is sent through 5 dB fixed optical attenuators (to ensure the appropriate amplifier optical input power) and subsequently sent to an erbium-doped fiber amplifier (EDFA, at the counter-clockwise pump: Keopsys PS-CUS-BT-C, and at the clockwise pump: Spectra RED5018) with a fixed output power (up to +23 dBm). Then, the amplified optical signals pass through polarization controllers (PCs) placed before optical circulators (OCs) to align the polarization state of the pumping light and resonator mode for maximizing the coupling efficiency. The circulator is used to prevent light back-scattering. Back-scattering can cause CW laser instability and a drop in EDFA amplification efficiency. The clockwise and counter-clockwise light is injected into opposing ends of the tapered fiber via optical circulator 1 (OC1) and optical circulator 2 (OC2). The comb light on the return port of OC1 is sent to an optical spectrum analyzer (OSA, Anritsu MS9740A, 0.03 nm resolution) to monitor and measure the peak powers of generated OFC carriers. The clockwise and counter-clockwise pumps, injected into micro-rod WGMR from both sides, ensure the necessary build-up circulating intensity to introduce Kerr-OFC generation. When single pump is used, part of its power converts into micro-rod heating, and the pump from the second direction is required to compensate for this power loss. First, we tried using two counter-propagating laser method at different wavelengths demonstrated to obtain a soliton regime [24]. Our lasers did not have the required mutual long-term frequency stability. We received increased stability of comb generation when we tried to pump the micro-rod from opposite directions with a single laser light divided into two directions. Each direction has an EDFA amplifier that amplifies the light power to a fixed level of +23 dBm. We attribute a positive effect to an increased microresonator thermal locking range. A single pump from two directions has been used in microcomb lidar [25] and quantum chaos [26].

Finally, the circulator OC2 separates the generated OFC from the counter-clockwise pump and is used as the output of the OFC-generated signal. An optical coupler with a 10/90 coupling ratio captures the output carrier spectrum using OSA-2 (Advantest Q8384, 0.01 nm resolution). Figure 4 shows the entirely generated Kerr-OFC comb and the zoomed-in inset of generated comb carriers near the pump wavelength corresponding to a carrier (0). Kerr-OFC light source generated optical carriers depicted as (0): λ = 1555.46, (+1): λ = 1556.18, (+2): λ = 1556.9, (+3): λ = 1557.62, (+4): λ = 1558.34, (+5): λ = 1559.06 nm and (+6): λ = 1559.78 nm (see Fig. 4) are used to demonstrate the transmission of 50 and 60 Gbaud NRZ-OOK and 50 Gbaud PAM-4 modulated signals via a 2 km standard single mode fiber (SMF) link. We have chosen these OFC optical carriers for further data transmission experiments as they provide the highest peak power levels higher than -15 dBm (4, -2.7, -7.3, -9.6, -11.6, -14.8, and -14.1 dBm, respectively) compared to others. The spacing between carriers is 89 GHz (∼0.72 nm). The tone-to-noise ratio (TNR) of these carriers is 52.9 dB, 46.5 dB, 41.8 dB, 39.4 dB, 36.8 dB, 34.2 dB, and 35.1 dB, respectively. As shown in Fig. 4 zoomed-in part, optical carriers beyond (+6) have similar performance; however, during experiments, we observed that these carriers were not stable enough to modulate data on them. In addition, the EDFA used in DCI setup could not amplify carriers beyond (+6) as they do not provide enough input power; even carrier (+5) is close to the amplified spontaneous emission (ASE) noise level. A solution to scale data transmission to optical carriers beyond (+6) would be using EDFA with lower input power levels as a preamplifier stage.

 figure: Fig. 4.

Fig. 4. The spectrum of silica micro-rod WGMR-based Kerr-OFC light source output, generated in WGMR (Res2, R = 700 µm, r = 250 µm and Q-factor of 2.6 × 107), with 89 GHz (∼ 0.72 nm) mode spacing between comb carriers.

Download Full Size | PDF

Further we discuss the parameters of optical frequency comb as spectrum shape, OFC state, conversion efficiency, power variation, and stability of 7 chosen optical carriers for data transmission. Compared to [20], the obtained OFC in Fig. 4 is asymmetric – a dip in the spectrum at the pumping light's left side (shorter wavelength). This can be explained by mode-crossing when a mode with a different polarization exists nearby this wavelength, and the power of the pumping light is divided between modes. Most of the power is taken by the other mode. We obtained a simple Turing pattern OFC by coupling 250 mW pumping light power while tapered fiber and micro-rod were hard-touching. For most of the experiment, noise performance stays close to EDFA ASE noise, except for an optical carrier +1, where the noise comes from the stimulated Brillouin scattering. IM/DD schemes in optical networks do not obligatory require a stable phase (mode-locked state) of the carrier that soliton provides, and a simple Turing comb can be used.

The OFC is obtained in the second micro-rod with a conversion efficiency of ∼20% (250 mW input power, and OFC has around 50 mW of optical power in the whole bandwidth). This result is close to [9], where authors achieved similar conversion efficiency but with lower pump power in a 75 µm diameter toroid. However, most mode-locked combs obtained in integrated resonators have conversion efficiency at the scale of a few percent [27].

Another factor to consider is at least 20 dB variation in the power of different comb carriers (Fig. 4) and how to achieve flat-top OFC light source as reported in [28]. The reason behind the large variation of the comb carrier power is the Turing pattern state. It is characterized by the decreasing optical power of the comb carrier when moving away from the pumped wavelength. To ensure a similar BER performance on all optical carriers for WDM applications, it is important to manage the flatness of the optical frequency comb. Otherwise, the difference in optical power is challenging. One possible solution to this issue is obtaining an OFC in a soliton state that ensures a more flat-top spectrum. Despite improved TNR and stability of the soliton carriers, generating a soliton comb requires stabilizing the power of light within the resonator, usually done with an auxiliary laser. However, an auxiliary laser complicates the setup [8]. The soliton regime exists quite far detuned from the cold cavity resonance, which leads to lower out-coupled power. The wavelength of the mode gets detuned from the pumping frequency due to the thermo-optical effect. MI combs can have even better SNR than soliton and for MI comb lines near the pump have flatter power [29]. However, such a comb cannot be used for data transmission as the power of carriers is strongly varying. In addition, IM/DD schemes in optical networks do not obligatory require a stable phase (mode-locked state) of the carrier that soliton provides, and a simple comb can be used. The OFC generated in the modulation instability state would ensure a flat-top OFC envelope. Another possible solution, if using the presented OFC, is to employ optical amplification for each carrier or use an inline wavelength selective switch (WSS) to flatten the comb carriers before data modulation.

The optical power budget (OPB) must be considered to determine the rated distance with a specific link budget to transmit and receive data [30]. However, the optical signal-to-noise ratio (OSNR) is another fundamental aspect of achieving high-quality wavelength division multiplexed (WDM) transmission. The minimum ratio must be at least 20 dB [30]. OFCs possess the inherent problem of long-term stability – the power level fluctuates for a comb carrier, which causes fluctuations in OSNR and transmitted signal quality. We measured the stability of the generated OFC from the second ∼90 GHz (89 GHz) micro-rod over 20 hours (Res-2, see Fig. 5), while coupling between the taper and the rod was hard-touching. It is hard-touching throughout the whole experiment.

 figure: Fig. 5.

Fig. 5. Measured micro-rod WGMR–based Kerr-OFC light source performance for comb carriers (0) to (+6). (a) The obtained OFC for a second micro-rod WGMR (D = 700 µm and r = 250 µm) ∼90 GHz (89.4 GHz), where side peaks at the side of the carrier (+1) come from SBS, which here acts as a noise. However, the amplitude of the highest side peak is around 32.3 dB lower than the carrier itself. (b) The captured power stability over 20 hours. Optical carrier power drop at the second working hour is related to the room temperature change of 6 degrees. The temperature inside the enclosure box is adjustable to the temperature of the outside environment.

Download Full Size | PDF

We have observed that comb carriers (0) to (+6) of the second micro-rod WGMR were relatively stable for 20 hours and had a different performance due to the resonator's ability to support multiple spatial modes. Let's analyze events happening at hours 0-4, at around the 14th hour and 18th hour. The first event is explained by temperature changes that can affect the power of the comb. As we do not have an ideal environment in a lab and temperature is not actively stabilized within the enclosure box, the temperature inside the box is variable in relation to the outside temperature of 23.0 ± 1°C during the day. However, during the night, the heating in a room is reduced by 6 degrees, which explains the first fluctuations between hours 0 and 4th. Sudden spikes at the 14th and 18th working hours are related to a small mechanical disposition between the micro-rod and tapered fiber. Mechanical disposition can happen even from the tiniest airflow that cannot be excluded without an ideal outside environment. Mechanical disposition makes the mode within the micro-rod start circulating geometrically in a different position, resulting in power spikes. Despite these power fluctuations, if the outside environment is stable or controlled, the power of a comb carrier remains relatively stable (4th–14th hour, 14th–18th hour, and 18th–20th hour). Therefore, we can say that in the short term, the stability of the OFC comes down to the capability of stabilizing temperature and excluding perturbations from the outside environment. In the long term, a microresonator's integrity and, therefore, parameters degrade and become unusable. Please note that the (+4) carrier's power fluctuations trend (green curve) differs principally from others. The reason is the SBS effect, which causes fluctuations in power. In the case of this experiment, when pumping from both sides, SBS can occur in counter-propagating and co-propagating directions. At the 2nd working hour, SBS peaks besides carrier (+1) intensify, taking power from other carriers. In addition, at this moment, carrier (+4) also boosts, which is caused by the SBS effect shifting to the wavelength (integer number of 11 GHz) of the carrier (+4). This is also the case for the green curve between the 14th and 20th hour. The opposite happens between the 4th and 14th hours. SBS peaks (including the one at the wavelength of OFC carrier +4) give the power to other carriers, which gain power. Finally, the measurements were stopped at the 20th hour as the micro-rod heats up and some OFC carriers become affected more by SBS, and further measurement is not reasonable.

Mode hopping can occur during the environmental temperature change. However, we did not observe mode hopping in this experiment with micro-rod nor in our previous experiments with microsphere resonators. The pumping laser we used is thermally stabilized and lasing continuously without interruption. In addition, it can be explained by the thermal-locking of a resonator mode to pump laser wavelength. Consequently, the wavelength remains stable during the experiment, and FSR does not change. The only thing that can change FSR is the heating of the resonator. As it absorbs pumping power in the form of heat, the diameter of a micro-rod increases. But compared to a micro-rod diameter of 700 µm, the increase in diameter can be considered negligible.

Based on the relative power stability periods, when the outside environment is stable, the generated Kerr-OFC optical carriers of the second micro-rod WGMR can be further used for data transmission.

3.2 Experimental setup of high-speed IM/DD short-reach optical interconnects enabled by the designed silica micro-rod WGMR-based Kerr-OFC light source

The experimental setup is shown in Fig. 6. The micro-rod WGMR-based Kerr-OFC light source (i.e., Kerr-OFC) and its optical carriers (0) to (+6) are used to demonstrate 50 and 60 Gbaud NRZ-OOK and 50 Gbaud PAM-4 modulated signals over a 2 km SMF link. We have chosen NRZ-OOK and PAM-4 modulation format as DCI can reduce complexity benefiting from IM/DD compared to complex coherent modulation formats [31]. PAM-4 has already been employed with OFC [32], but with quantum dot mode-locked laser comb rather than micro-rod combs. The generated Kerr-OFC carriers are sent to the optical band-pass filter (OBPF, Santec OTF-350) with a 3-dB bandwidth of 35 GHz to separate one optical carrier at a time. Optical couplers with a 10/90 coupling ratio before and after the OBPF capture the Kerr-OFC comb spectrum by OSA-1 and filtered comb carrier by OSA-2. An EDFA (Amonics AEDFA-CL-18-B-FA) with a fixed optical output power of 23.5 dBm pre-amplifies the filtered optical carriers before launching them into a Mach–Zehnder modulator (MZM, Photline MX-LN40) with 40 GHz 3 dB bandwidth, 20 dB extinction ratio, and 9 dB insertion loss. A polarization controller PC is placed before the MZM to reduce polarization-dependent loss. At this point, the measured carrier power is around 5 dBm, adjusted to this level for optimal input power to modulator. Moreover, the RIN of optical carriers here can be considered negligible as it has not impaired the capability of external high-speed data modulation even for a carrier (+1), where SBS peaks are the only contribution to RIN. These side peaks are 20 GHz away from a carrier (+1), which can be seen as an optical modulation of a carrier signal. In our previous experiment with the microsphere, we observed side peaks at a 30 MHz distance from a carrier that corresponded to a microresonator's mechanical resonance and impaired the carrier's use in data transmission. But we have not seen such behavior in micro-rods, which is an advantage of micro-rod compared to microspheres.

 figure: Fig. 6.

Fig. 6. Experimental demonstration of high-speed optical interconnect up to 100 Gbps/λ enabled by a micro-rod WGMR-based Kerr-OFC light source.

Download Full Size | PDF

The high-speed NRZ-OOK and PAM-4 signals are generated offline using a 215-1 pseudorandom binary sequence (PRBS). The signal is up-sampled and filtered using a root-raised-cosine (RRC) filter with a roll-off factor of 1. This value is chosen as timing recovery with a smaller roll-off factor does not work due to the stability issues with the individual carriers. Afterward, the encoded signal is loaded into the electrical arbitrary waveform generator (EAWG). Frequency domain pre-equalization up to 30 GHz is used to compensate amplitude–frequency distortions and limited bandwidths (BWs) of a 65 GSa/s EAWG (Keysight M9502A, 25 GHz). We have chosen 50 and 60 Gbaud as these are the maximum baud rates we can achieve with 65 GSa/s AWG as it has a bandwidth of only 25 GHz. Additionally, we assume that if 60 Gbaud works for this comb, then it will give a performance margin for the more common 40 Gbaud baud rate at 100 GHz channel spacing.

The electrical amplifier (EA-1, 38 GHz, 29 dB gain) amplifies the generated electrical signal at the output of EAWG, which is fed into the MZM. Then, the modulated optical signal is transmitted over 2 km of the SMF link and passed through a variable optical attenuator (VOA). A splitter with a 10/90 splitting ratio and a power meter (PM) are used to control optical power at the receiver. The high-power 50 GHz InGaAs photo receiver (PIN, Lab Buddy, DSC10H-39) with a sensitivity of +4 dBm at BER of 10−12 and responsivity of 0.5 A/W performs the received signal’s optical-to-electrical conversion. An electrical amplifier (EA-2, 25 GHz, 16 dB gain) amplifies the electrical signal, and finally, a digital storage oscilloscope (DSO, Keysight DSAZ334A, 80 GSa/s, 33 GHz) samples and captures it for offline DSP processing. As you can see from the developed DCI setup, each newly generated carrier is filtered out before modulation. The experimental results in the next section are obtained for seven single-channel data-center interconnect systems. Consequently, no signal crosstalk is expected due to the single signal transmission through the fiber, which can be considered an advantage compared to an array of discrete lasers. Decorrelated telecom data on the neighboring carriers are needed to simultaneously test several comb carriers for possible crosstalk due to power differences between carriers and RRC filter overlap. In future research, this analysis will be performed by adding a decorrelation stage or using an additional modulator to modulate different data sequences on even and odd optical carriers.

4. Experimental results and discussion

Using the experimental setup shown in Fig. 6, we compare performance limits in terms of achievable data rates up to 100 Gbps/λ in the IM/DD short-reach optical DCI system based on the Kerr-OFC light source for NRZ-OOK and PAM-4 modulated data transmission. The received and sampled signal is processed offline using a DSP routine that consists of a low-pass filter (LPF) with a normalized bandwidth of 1.2, clock recovery, post-equalization, and a BER counter. The overall goal of using LPF equalization and RRC filter with a roll-off factor of 1 is to optimize the achieved performance of transmitted signal performance with this comb. The choice of LPF bandwidth (LPFBW) is made by evaluating the obtained BER as a function of the LPFBW/Baud rate for NRZ-OOK and PAM-4 signals in terms of different normalized LPFBW values within the range of 0.9 to 1.6. The LPFBW value is identified using the results shown in Fig. 7. The figure shows that the lowest BER for the worst-case scenario using a Kerr-OFC optical carrier (+6) is achieved with 1.2 normalized LPFBW. Therefore, it ensures the best possible performance after further processing.

 figure: Fig. 7.

Fig. 7. BER versus normalized LPFBW/Baud rate for the IM/DD short-reach optical DCI system based on Kerr-OFC light source. The worst-performing optical carrier (+6) provides transmission with 1.2 of normalized LPFBW for NRZ-OOK modulated signals at (a) 50 Gbaud/λ, (b) 60 Gbaud/λ, and (c) for PAM-4 modulated signals operating at 50 Gbaud/ λ.

Download Full Size | PDF

In addition to the previously mentioned DSP functions, we use adaptive equalization (EQ), e.g., post-equalization, to improve the received signal quality: overcome intersymbol interference (ISI) and bandwidth limitations of electrical components [30]. Therefore, we use a decision feedback equalizer (DFE) with 15 feed-forward taps (FFT) and 7 feedback taps (FBT) for 50 and 60 Gbaud NRZ-OOK transmission. However, for 50 Gbaud PAM-4 transmission, the number of FFT and FBT taps used varies for each carrier – (0, + 4, + 5–55 FFT and 15 FBT; + 1–85 FFT and 55 FBT; + 2–23 FFT and 16 FBT; + 3–10 FFT and 11 FBT). The chosen number of taps maximally improves BER performance. A total of 1.2 million bits are used for BER counting.

Further we analyze IM/DD short-reach optical DCI system performance with and without post-equalization. We also show BER curves and eye diagrams of the received NRZ-OOK and PAM-4 signals after transmission over the 2 km SMF link section. The hard-decision forward error correction (HD-FEC) with 7% overhead (OH) and pre-FEC BER threshold at 3.8 × 10−3, and the soft-decision forward error correction (SD-FEC) with 20% OH and pre-FEC BER threshold at 4.0 × 10−2 are considered for quality analysis of received NRZ-OOK and PAM-4 signals. The data transmission is considered successful if all errors can be corrected below FEC thresholds. We show 50 and 60 Gbaud NRZ-OOK data transmission (Figs. 8(a-d)) as well as 50 Gbaud PAM-4 transmission (Figs. 8(e-f)). Figure 8 demonstrates different BER trends for each optical carrier, although all optical carriers are amplified to the same fixed output power before modulation. Different trends are related to each comb carrier's different signal-to-noise (SNR) ratios, where noise floor sums from two ASE-generating components – EDFA in the Kerr-OFC light source and EDFA in the data transmission setup. Therefore, noise significantly increases, leading to increased SNR values between the comb carriers. Consequently, it increases the BER values of received signals.

 figure: Fig. 8.

Fig. 8. BER versus ROP for the IM/DD short-react optical DCI system operating on Kerr-OFC light source transmission for NRZ-OOK modulated signals at 50 Gbaud/λ (a) without and (b) with EQ, and at 60 Gbaud/λ (c) without and (d) with EQ. PAM-4 modulated signals at 50 Gbaud/λ (e) without and (f) with EQ. EQ for figures (b) and (d) uses 15 FFT and 7 FBT. EQ for the figure (f) uses 55 FFT and 15 FBT for carriers 0, + 4; + 5; 85 FFT and 55 FBT for carrier +1; 23 FFT and 16 FBT for carrier +2, and 10 FFT and 11 FBT for carrier +3).

Download Full Size | PDF

As one can see in Fig. 8(a), the 50 Gbaud NRZ-OOK signal transmission is possible without post-equalization for Kerr-OFC carriers (0) at λ = 1555.46 nm to (+6) at λ = 1559.78 nm. The worst-performing data channel (based on the BER value) is the carrier (+6), where the 20% SD-FEC BER threshold is achieved at -18 dBm of received optical power (ROP). The best performance is shown with the carrier (0) – CW pumping source, which corresponds to the peak with the highest Kerr-OFC light source output optical power of 4 dBm and the highest TNR value of 52.8 dB. The 50 Gbaud NRZ-OOK signal transmission beyond the 7% HD-FEC threshold at 3.8 × 10−3 is possible without post-equalization for Kerr-OFC carriers (0) at λ = 1555.46 nm to (+4) at λ = 1558.34 nm. As one can see in Fig. 8(b), the post-equalization slightly improves BER performance compared to the previous case without the post-equalization and enables 50 Gbaud NRZ-OOK signal transmission for Kerr-OFC carriers (0) at λ = 1555.46 nm to (+6) at λ = 1559.78 nm below the 20% SD-FEC limit.

As shown in Fig. 8(c), the 60 Gbaud NRZ-OOK signal transmission is possible for Kerr-OFC carriers (0) at λ = 1555.46 nm to (+4) at λ = 1558.34 nm without post-equalization with BER performance below the 20% SD-FEC limit. The carrier (+4) is the worst-performing data channel, where the 20% SD-FEC BER threshold is reached at -18 dBm of ROP. As in the previous case, for 60 Gbaud NRZ-OOK transmission, the best-performing Kerr-OFC pumping source carrier is (0). The results in Fig. 8(d) show that post-equalization significantly improves BER performance providing 60 Gbaud NRZ-OOK signal transmission for Kerr-OFC carriers (0) to (+5).

Finally, without post-equalization, 50 Gbaud (100 Gbps/λ) PAM-4 signal transmission can be realized with the optical carrier (0) at -17 dBm of ROP, providing signal BER below the 7% HD-FEC limit, see Fig. 8(e). Carriers (+1) and (+2) achieve BER value below the defined 20% SD-FEC limit at -16 dBm of ROP. The received signal BER for other carriers stays above the defined 20% SD-FEC limit. The post-equalization improves performance, enabling the transmission of 50 Gbaud PAM-4 with (0) to (+2) carriers. In the case of post-equalization, the achieved BER for the optical carrier (0) is below the 7% HD-FEC limit of 3.8 × 10−3 at -19 dBm of ROP, see Fig. 8(f). However, for optical carriers (+1) to (+2), the minimal BER value is achieved below the 20% SD-FEC limit of 4 × 10−2 at -16 dBm of the ROP. Received eye diagrams of the 50 and 60 Gbaud NRZ-OOK and 50 Gbaud PAM-4 signals after transmission over 2 km SMF for Kerr-OFC carrier (+1) are given in Fig. 9. The values of extinction ratio between optical levels 1 and 0 that we can obtain at 3.5 peak-to-peak voltage (Vpp) of MZM modulator are around 7-8 dB. If we consider 50 Gbps NRZ-OOK eye diagrams, the extinction ratio is around 6 dB, but for 60 Gbps NRZ-OOK, it is around 5.3 dB. The higher are bitrates the extinction ratios are lower.

 figure: Fig. 9.

Fig. 9. Received NRZ-OOK signal eye diagrams with and without post-equalization for carrier (+1) captured at ROP of -12.5 dBm in the (a) 50 Gbaud/λ and (b) 60 Gbaud/λ case, and (c) PAM-4 eye diagram in the 50 Gbaud/λ case.

Download Full Size | PDF

5. Conclusions

We experimentally investigated the applicability of an in-house-built silica micro-rod resonator-based Kerr-OFC light source for high-speed IM/DD short-reach optical interconnects. The newly proposed scheme exploits a 700-µm micro-rod WGMR with a curvature radius of 250 µm to generate a Kerr-OFC. The resulting optical frequency comb has optical carriers evenly spaced at 89.4 GHz. Seven of them having peak power levels (≥-15 dBm) and TNR above 20 dB are used to demonstrate NRZ-OOK and PAM-4 modulated data transmission over a 2 km long DCI. NRZ-OOK data transmission with a data rate of 50 Gbaud and 60 Gbaud with post-equalization using 15 FFT and 7 FBT is achieved on seven (0, + 1, + 2, + 3, + 4, + 5, + 6) and five (0, + 1, + 2, + 3, + 4) optical carriers below the 20% SD-FEC limit, respectively. However, 100 Gbps/λ (50 Gbaud) PAM-4 transmission is possible by optical carrier (0) without post-equalization and on four (0, + 1, + 2, + 3) optical carriers with post-equalization. For the PAM-4 transmission case, the number of taps varies for each carrier - 55 FFT and 15 FBT for carrier (0); 85 FFT and 55 FBT for carrier (+1); 23 FFT and 16 FBT for carrier (+2), and 10 FFT and 11 FBT for carrier (+3). In this case, the BER value on the optical carrier (0) without post-equalization is below the 7% HD-FEC limit. The post-equalization for optical carriers (+1, + 2, + 3) improves the BER value below the 20% SD-FEC limit of 4 × 10−2. These experimental results support the implementation of the IM-DD configuration over coherent, especially in intra- and inter-data center networks where ultralow latency and cost-effectiveness remain crucial factors. Future research is necessary to analyze the simultaneous transmission of multiple signals on different wavelengths through the fiber - multi-wavelength WDM application. To ensure a similar BER performance on all optical carriers for WDM applications, it is important to manage the flatness of the optical frequency comb. Otherwise, the difference in optical power is challenging. One of the possible solutions that can solve this issue is to obtain an OFC in a soliton state. The second is to employ optical amplification for each carrier or use an inline wavelength selective switch to flatten the comb before data modulation. The idea is partially already used in this manuscript, where we use fixed output EDFA for each filtered carrier.

Funding

European Regional Development Fund (1.1.1.2/VIAA/4/20/659, 1.1.1.5/19/A/003); Rīgas Tehniskā Universitāte (Doctoral Grant programmes); H2020 European Research Council (Starting Grant CounterLIGHT 756966); H2020 Marie Skłodowska-Curie Actions (Innovative Training Network "Microcombs" 812818); Max-Planck-Gesellschaft.

Disclosures

The authors declare no conflicts of interest.

Data availability

Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.

References

1. S. T. Ahmad, P. D. Lakshmijayasimha, A. K. Anandarajah, C. Browning, and P. M. Anandarajah, “Active Demultiplexer-enabled Directly Modulated DMT Transmission Using Optical Frequency Combs for Data Center Interconnects,” J. Lightwave Technol. 39(17), 5468–5473 (2021). [CrossRef]  

2. S. Fujii, S. Tanaka, T. Ohtsuka, S. Kogure, K. Wada, J. Kumazaki, S. Tasaka, Y. Hashimoto, Y. Kobayashi, T. Araki, K. Furusawa, N. Sekine, S. Kawanishi, and T. Tanabe, “Dissipative Kerr soliton microcombs for FEC-free optical communications over 100 channels,” Opt. Express 30(2), 1351–1364 (2022). [CrossRef]  

3. H. Hu and L. Oxenløwe, “Chip-based optical frequency combs for high-capacity optical communications,” Nanophotonics 10(5), 1367–1385 (2021). [CrossRef]  

4. T. Salgals, S. Spolitis, S. Olonkins, and V. Bobrovs, “Investigation of 4-PAM modulation format for use in WDM-PON optical access systems,” in Proceeding of Progress In Electromagnetics Research Symposium Spring (PIERS) (2017), pp. 2450–2454.

5. M. Mazur, M. G. Suh, A. Fülöp, J. Schröder, V. Torres-Company, M. Karlsson, K. J. Vahala, and P. A. Andrekson, “High Spectral Efficiency Coherent Superchannel Transmission With Soliton Microcombs,” J. Lightwave Technol. 39(13), 4367–4373 (2021). [CrossRef]  

6. T. Salgals, J. Alnis, R. Murnieks, I. Brice, J. Porins, A. V. Andrianov, E. A. Anashkina, S. Spolitis, and V. Bobrovs, “Demonstration of a fiber optical communication system employing a silica microsphere-based OFC source,” Opt. Express 29(7), 10903–10913 (2021). [CrossRef]  

7. S. Zhang, J. M. Silver, X. Shang, L. Del Bino, N. M. Ridler, and P. Del’Haye, “Terahertz wave generation using a soliton microcomb,” Opt. Express 27(24), 35257–35266 (2019). [CrossRef]  

8. S. Zhang, J. M. Silver, L. Del Bino, F. Copie, M. T. M. Woodley, G. N. Ghalanos, A. Ø. Svela, N. Moroney, and P. Del’Haye, “Sub-milliwatt-level microresonator solitons with extended access range using an auxiliary laser,” Optica 6(2), 206–212 (2019). [CrossRef]  

9. P. Del’Haye, A. Schliesser, O. Arcizet, T. Wilken, R. Holzwarth, and T. J. Kippenberg, “Optical frequency comb generation from a monolithic microresonator,” Nature 450(7173), 1214–1217 (2007). [CrossRef]  

10. J. Pfeifle, A. Kordts, P. Marin, M. Karpov, M. Pfeiffer, V. Brasch, R. Rosenberger, J. Kemal, S. Wolf, W. Freude, T. J. Kippenberg, and C. Koos, “Full C and L-Band Transmission at 20 Tbit/s Using Cavity-Soliton Kerr Frequency Comb Source,” in Conference on Lasers and Electro-Optics (CLEO) (2015), paper JTh5C.8.

11. E. A. Anashkina, M. P. Marisova, T. Salgals, J. Alnis, I. Lyashuk, G. Leuchs, S. Spolitis, V. Bobrovs, and A. V. Andrianov, “Optical Frequency Combs Generated in Silica Microspheres in the Telecommunication C-, U-, and E-Bands,” Photonics 8(9), 345 (2021). [CrossRef]  

12. M. Palomo, J. N. Kemal, M. Karpov, A. Kordts, J. Pfeifle, M. H. P. Pfeiffer, P. Trocha, S. Wolf, V. Brasch, M. H. Anderson, R. Rosenberger, K. Vijayan, W. Freude, T. J. Kippenberg, and C. Koos, “Microresonator based solitons for massively parallel coherent optical communications,” Nature 546(7657), 274–279 (2017). [CrossRef]  

13. T. Tanabe, S. Fujii, and R. Suzuki, “Review on microresonator frequency combs,” Jpn. J. Appl. Phys. 58(SJ), SJ0801 (2019). [CrossRef]  

14. S.B. Papp, P. Del’Haye, and S. A. Diddams, “Mechanical Control of a Microrod-Resonator Optical Frequency Comb,” Phys. Rev. 3(3), 031003 (2013). [CrossRef]  

15. T. J. Kippenberg, A. L. Gaeta, M. Lipson, and M. L. Gorodetsky, “Dissipative Kerr solitons in optical microresonators,” Science 361(6402), eaan8083 (2018). [CrossRef]  

16. J. S. Levy, A. Gondarenko, M. A. Foster, A. C. T. Foster, A. L. Gaeta, and M. Lipson, “Review on microresonator frequency combs,” Nat. Photonics 4(1), 37–40 (2010). [CrossRef]  

17. K. Cognee, Hybridization of open photonic resonators, (University of Amsterdam, University of Bordeau2020).

18. R. Niu, S. Wan, S. M. Sun, T. G. Ma, H. J. Chen, W. Q. Wang, Z. Lu, W. F. Zhang, G. C. Guo, C. L. Zou, and C. H. Dong, “Repetition rate tuning of soliton in microrod resonators,” arXiv, arXiv:1809.06490 (2018). [CrossRef]  

19. L. D. Bino, J. M. Silver, M. T. M. Woodley, S. L. Stebbings, X. Zhao, and P. Del’Haye, “Microresonator isolators and circulators based on the intrinsic nonreciprocity of the Kerr effect,” Optica 5(3), 270–282 (2018). [CrossRef]  

20. D. C. Cole, E. S. Lamb, P. Del’Haye, S. A. Diddams, and S. B. Papp, “Soliton crystals in Kerr resonators,” Nat. Photonics 11(10), 671–676 (2017). [CrossRef]  

21. A. Pasquazi, M. Peccianti, L. Razzari, D. J. Moss, S. Coen, M. Erkintalo, Y. K. Chembo, T. Hansson, S. Wabnitz, P. Del’Haye, X. Xue, A. M. Weiner, and R. Morandotti, “Micro-combs: A novel generation of optical sources,” Phys. Rep. 729, 1–81 (2018). [CrossRef]  

22. S. Fuji and T. Tanabe, “Dispersion engineering and measurement of whispering gallery mode microresonator for Kerr frequency comb generation,” Nanophotonics 9(5), 1087–1104 (2020). [CrossRef]  

23. T. Salgals, J. Alnis, O. Ozolins, A. V. Andrianov, E. A. Anashkina, I. Brice, R. Berkis, X. Pang, A. Udalcovs, J. Porins, S. Spolitis, and V. Bobrovs, “Silica Microsphere WGMR-Based Kerr-OFC Light Source and Its Application for High-Speed IM/DD Short-Reach Optical Interconnects,” Appl. Sci. 12(9), 4722 (2022). [CrossRef]  

24. S. Zhang, J. M. Silver, T. Bi, and P. Del’Haye, “Spectral extension and synchronization of microcombs in a single microresonator,” Nat. Commun. 11(1), 6384 (2020). [CrossRef]  

25. M. G. Suh and K. J. Vahala, “Soliton microcomb range measurement,” Science 359(6378), 884–887 (2018). [CrossRef]  

26. M. T. M. Woodley, L. Hill, L. D. Bino, G. L. Oppo, and P. Del’Haye, “Self-Switching Kerr Oscillation of Counterpropagating Light in Microresonators,” Phys. Rev. Lett. 126(4), 043901 (2021). [CrossRef]  

27. X. Xue, P. H. Wang, Y. Xuan, M. Qi, and A. M. Weiner, “Microresonator Kerr frequency comb with high conversion efficiency,” Laser Photonics Rev. 11(1), 1600276 (2017). [CrossRef]  

28. J. Z. Huang, Z. T. Ji, J. J. Chen, W. Q. Wie, J. L. Qin, Z. H. Wang, Z. Y. Li, T. Wang, X. Xiao, and J. Jun, “Ultra-broadband flat-top quantum dot comb lasers,” Photonics Res. 10(5), 1308–1316 (2022). [CrossRef]  

29. A. Lukashchuk, J. Riemensberger, A. Tusnin, J. Liu, and T. J. Kippenberg, “Chaotic micro-comb based parallel ranging,” arXiv, arXiv.2112.10241 (2021). [CrossRef]  

30. A. Udalcovs, T. Salgals, L. Zhang, X. Pang, A. Djupsjöbacka, S. Spolitis, V. Bobrovs, S. Popov, and O. Ozolins, “Optical Power Budget of 25+ Gbps IM/DD PON with Digital Signal Post-Equalization,” Appl. Sci. 10(17), 6106 (2020). [CrossRef]  

31. A. Fülöp, M. Mazur, A. L. Riesgo, O. B. Helgason, P. H. Wang, Y. Xuan, D. E. Leaird, M. Qi, P. A. Andrekson, A. M. Weiner, and V. T. Company, “High-order coherent communications using mode-locked dark-pulse Kerr combs from microresonators,” Nat. Commun. 9(1), 1598 (2018). [CrossRef]  

32. S. Pan, H. Zhang, Z. Liu, M. Liao, M. Tang, D. Wu, X. Hu, J. Yan, L. Wang, M. Guo, Z. Wang, T. Wang, P. M. Smowton, A. Seeds, H. Liu, X. Xiao, and S. Chen, “Multi-wavelength 128 Gbit s-1 λ-1 PAM4 optical transmission enabled by a 100 GHz quantum dot mode-locked optical frequency comb,” J. Phys. D: Appl. Phys. 55(14), 144001 (2022). [CrossRef]  

Data availability

Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.

Cited By

Optica participates in Crossref's Cited-By Linking service. Citing articles from Optica Publishing Group journals and other participating publishers are listed here.

Alert me when this article is cited.


Figures (9)

Fig. 1.
Fig. 1. (a) The fabrication of fused quartz micro-rod WGMRs using CO2 laser machining, where the spindle stage is used to rotate and a galvanometer is used to position the laser beam on the ZnSe lens. (b) The captured image of a fabricated micro-rod with a 700- µm diameter and 520- µm gap between WGMRs (Res1 to Res5). Images of micro-rod WGMRs with curvature radius of 250 µm, 250 µm, 200 µm, 175 µm, and 150 µm from the first resonator (Res1) to the fifth resonator (Res5). (c) The captured images of individual micro-rod resonators (Res1 to Res5) indicating different micro-rod curvature radii with white circles.
Fig. 2.
Fig. 2. The simulations of the silica micro-rod resonators with D = 700 µm and different curvature radii. The simulation consists of a silica micro-rod resonator and air surrounding it. (a) The mesh for the simulation domains (the approximate mode area has a finer mesh, (b) the cross-section of the fundamental TE mode with the normalized absolute value of the electrical field at a wavelength of about 1550 nm. (c) The simulated dispersion for silica micro-rod resonators with curvature radii 50, 100, 150, and 200 µm.
Fig. 3.
Fig. 3. Experimental setup illustrating the developed silica micro-rod WGMR-based Kerr-OFC as a light source for optical communications.
Fig. 4.
Fig. 4. The spectrum of silica micro-rod WGMR-based Kerr-OFC light source output, generated in WGMR (Res2, R = 700 µm, r = 250 µm and Q-factor of 2.6 × 107), with 89 GHz (∼ 0.72 nm) mode spacing between comb carriers.
Fig. 5.
Fig. 5. Measured micro-rod WGMR–based Kerr-OFC light source performance for comb carriers (0) to (+6). (a) The obtained OFC for a second micro-rod WGMR (D = 700 µm and r = 250 µm) ∼90 GHz (89.4 GHz), where side peaks at the side of the carrier (+1) come from SBS, which here acts as a noise. However, the amplitude of the highest side peak is around 32.3 dB lower than the carrier itself. (b) The captured power stability over 20 hours. Optical carrier power drop at the second working hour is related to the room temperature change of 6 degrees. The temperature inside the enclosure box is adjustable to the temperature of the outside environment.
Fig. 6.
Fig. 6. Experimental demonstration of high-speed optical interconnect up to 100 Gbps/λ enabled by a micro-rod WGMR-based Kerr-OFC light source.
Fig. 7.
Fig. 7. BER versus normalized LPFBW/Baud rate for the IM/DD short-reach optical DCI system based on Kerr-OFC light source. The worst-performing optical carrier (+6) provides transmission with 1.2 of normalized LPFBW for NRZ-OOK modulated signals at (a) 50 Gbaud/λ, (b) 60 Gbaud/λ, and (c) for PAM-4 modulated signals operating at 50 Gbaud/ λ.
Fig. 8.
Fig. 8. BER versus ROP for the IM/DD short-react optical DCI system operating on Kerr-OFC light source transmission for NRZ-OOK modulated signals at 50 Gbaud/λ (a) without and (b) with EQ, and at 60 Gbaud/λ (c) without and (d) with EQ. PAM-4 modulated signals at 50 Gbaud/λ (e) without and (f) with EQ. EQ for figures (b) and (d) uses 15 FFT and 7 FBT. EQ for the figure (f) uses 55 FFT and 15 FBT for carriers 0, + 4; + 5; 85 FFT and 55 FBT for carrier +1; 23 FFT and 16 FBT for carrier +2, and 10 FFT and 11 FBT for carrier +3).
Fig. 9.
Fig. 9. Received NRZ-OOK signal eye diagrams with and without post-equalization for carrier (+1) captured at ROP of -12.5 dBm in the (a) 50 Gbaud/λ and (b) 60 Gbaud/λ case, and (c) PAM-4 eye diagram in the 50 Gbaud/λ case.
Select as filters


Select Topics Cancel
© Copyright 2024 | Optica Publishing Group. All rights reserved, including rights for text and data mining and training of artificial technologies or similar technologies.