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

106.25-Gbps PAM-4 bidirectional optical sub-assembly module with in-line arrangement of optical and electrical interfaces

Open Access Open Access

Abstract

We successfully demonstrate a 106.25-Gbps PAM-4 bidirectional optical sub-assembly for optical access networks, including a driver amplifier and an electro-absorption modulated laser for a transmitter, a photodiode and transimpedance amplifier for a receiver, and an optical filter block. For its implementation, we propose design strategies providing an in-line arrangement of optical and electrical interfaces while ensuring optical alignment tolerance for easy assembly and reducing electrical crosstalk between the transmitter and receiver. Measured receiver sensitivity was <–11.4 dBm for the KP4 forward error correction limit during transmitter operation, and measured power penalty of 10-km single-mode fiber transmission was <0.9 dB.

© 2022 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement

1. Introduction

Several research groups on various optical networks, such as Ethernet and optical access networks, have been working to harness PAM-4 technology to meet the ever-increasing demand for data traffic [19]. Recently, the standard of bidirectional optical access networks has also employed the PAM-4 technology and signaling rate of Ethernet standard [9]. The optical access network requires a bidirectional optical sub-assembly (BOSA) module as a key component. In general, the electrical interfaces of a transmitter (Tx) and receiver (Rx) of the conventional BOSA are arranged perpendicular to each other [1012]. This is because the optical filter block separates optical Tx and Rx signals in different directions because of an angle of incidence of 45°. This arrangement causes the detour of the electrical signal paths for the connection between the BOSA and other electrical components inside the transceiver, as shown in Fig. 1(a). This may not be a big problem at the BOSA as far as low-speed binary signals are used, such as ∼10-Gbps non-return to zero (NRZ) [1012]. However, when using PAM-4 signals operating at >100 Gbps, the performance of the transceiver employing the conventional BOSA schemes may deteriorate owing to the long electrical paths. In general, the quality of the PAM-4 signal is much more sensitive to noise than that of the NRZ signal.

 figure: Fig. 1.

Fig. 1. Components placement of the transceivers using (a) a conventional BOSA and (b) a proposed BOSA.

Download Full Size | PDF

In this study, we successfully demonstrated a 106.25-Gbps PAM-4 BOSA integrated with a driver amplifier close to an electro-absorption modulated laser (EML). By applying the proposed structure to the BOSA, the electrical interfaces of the Tx and Rx were placed on one side of the module, and the optical and electrical interfaces were arranged in a straight line rather than in a bending path, as shown in Fig. 1(b). Thus, the scheme of the BOSA provides good signal integrity for high-speed electrical signals and easy placement of components inside the optical transceiver. In Section 2, we describe the proposed configuration and optical/electrical design strategies of the BOSA, including: (1) an optical rod is incorporated to improve the alignment tolerance at the Rx optical part; (2) the driver amplifier and the wide-band bias-T element are integrated close to the EML for high-speed operation; and (3) the signal line configuration of the Tx and Rx parts on the flexible printed circuit board (FPCB) are designed to reduce crosstalk between the Tx and Rx. Section 3 analyzes the electrical and optical performance of the BOSA, such as frequency responses and transmission penalties. The conclusions are presented in Section 4.

2. Design and implementation of 100-Gbps PAM-4 BOSA

2.1 Configuration of BOSA

 figure: Fig. 2.

Fig. 2. Conceptual scheme of the proposed BOSA.

Download Full Size | PDF

To arrange the electrical interfaces of the Tx and Rx on the same side of module, we have proposed the structure of the BOSA employing a zig-zag thin-film filter block for separating optical Tx and Rx signals [13]. This design concept can provide minimal electrical signal paths inside the optical transceiver to ensure signal integrity. Figure 2 shows the structure of the implemented BOSA operating with a 106.25-Gbps PAM-4 signal. In the Rx part, the input optical signal of the BOSA is collimated through the lens integrated into the receptacle [14], separated from the Tx optical signal through the optical filter block, and applied to the photodiode (PD) block. The PD block is composed of a pin-PD on an aluminum nitride submount, a trans-impedance amplifier (TIA) and a focusing lens [15,16]. The responsivity and bandwidth of the pin-PD (Albis’ PD40X1) are ∼0.7 A/W and ∼35 GHz, respectively. The bandwidth of the linear TIA (Inphi’s IN5662TA) is ∼35 GHz. In the Tx part, the optical signal from the EML is collimated through the lens and subsequently passes through the optical isolator and the optical filter block. The wavelength and bandwidth of the EML (Lumentum’s HL13B5CP00-L3) are 1331 nm and ∼35 GHz, respectively. The bandwidth of the driver amplifier (Inphi’s IN5630SE) is ∼45 GHz. In the optical filter block, the distance between the centers of the Tx and Rx paths was designed to be 3.3 mm for the integration of the driver amplifier and the bias-T element for the EML and the TIA for the PD. The wavelengths of the Tx and Rx optical signals were set as 1331 and 1271 nm, respectively. The large gap between the optical filter and PD blocks was filled with an optical rod for the coupling efficiency and alignment tolerance. The effect of the optical rod will be analyzed in detail later. The BOSA was connected to the evaluation board through the FPCB. At the optical interface of the BOSA, a receptacle with a lens was applied for the Tx and Rx optical signals [14,15]. The lens simultaneously operates as a focusing lens for the Tx side and a collimating lens for the Rx side.

2.2 Design of optical signal path

 figure: Fig. 3.

Fig. 3. (a) Experimental schematic to measure the alignment tolerances of the receptacle and collimating lens, (b) coupling efficiencies as a function of the X1- and Y1-axis displacements of the receptacle, and (c) coupling efficiencies as a function of the X2- and Y2-axis displacements of the collimating lens in front of the EML.

Download Full Size | PDF

In the aforementioned optical signal flow, the Tx optical signal power of the BOSA is strongly dependent on the position of the receptacle, optical filter block, and collimating lens. To confirm the feasibility of a BOSA’s manufacturing process with passive and active alignment processes, we have investigated alignment tolerances of the key components. In the case of the optical filter block, the alignment tolerance was well investigated in our previous studies [15,16]. To establish an efficient assembly process of the Tx optical part, the alignment tolerances of the receptacle and collimating lens were measured through the experimental setup shown in Fig. 3(a). Figure 3(b) shows the coupling efficiencies with the X1-and Y1-axis displacements of the receptacle. The coupling efficiency is defined as the ratio of the coupled optical power at the receptacle to the output optical power of the EML. The maximum coupling efficiency was approximately 65%. It was confirmed that the measured 1-dB alignment tolerances measured in the X1- and Y1-axis directions were ±55 and ±100 μm, respectively. The margin for the Y1-axis is larger than that of the X1-axis because the light output from the EML has a wide oval shape, which is elongated along the X1-axis. In the case of the Z1-axis, there was no significant loss over the displacement range of several hundred micrometers. Figure 3(c) depicts the coupling efficiencies of the X2- and Y2-axis displacements of the collimating lens. It was also confirmed that the measured 1-dB alignment tolerances were approximately ±2 μm in both the X2- and Y2-axis directions. The Z2-axis alignment margin was measured to be ±25 μm, which was larger than those of the X2- and Y2-axis. The alignment tolerances of the optical filter block reported from our previous studies were observed to be several hundred micrometers [15,16].

Considering the measured alignment tolerances, it was possible to fabricate the Tx optical part by the passive assembly of the receptacle and the optical filter block. In the case of the collimating lens in front of the EML, the active assembly process was necessary because of the very limited alignment tolerance. In the manufacturing process of the Tx optical part, the active alignment of the collimating lens was performed after the passive alignment of the receptacle and optical filter block.

 figure: Fig. 4.

Fig. 4. (a) Effect of an optical rod and (b) beam diameters of the collimating light as a function of propagation distance from the optical filter block with and without the optical rod. Insets of (b) show the measured beam shapes after 5-mm propagation with and without the optical rod, respectively.

Download Full Size | PDF

To increase the alignment tolerance for the efficient fabrication of the Rx optical part, we proposed the utilization of an optical rod between the optical filter and PD blocks. Figure 4(a) shows the effect of the optical rod. In general, the diameter of the collimated optical signal expands after the beam waist as it progresses through air. When an optical rod is placed between the optical filter block and beam analyzer, the optical signal is refracted between the air and optical rod, which has a higher refractive index than air (nair = 1), according to Snell's law, as shown in Fig. 4(a). Hence it reduces the beam diameter expansion to improve the coupling efficiency and alignment tolerance. To verify this, the diameter of the optical signal after the optical filter block was measured using a beam analyzer while changing its propagation length with and without the optical rod. The propagation length was defined as the distance between the optical filter block and beam analyzer. The optical rod used had a refractive index (nrod) of 1.414 and a length (Lrod) of 4 mm. The input and output facets of the optical rod were anti-reflected coated. Figure 4(b) shows that the increase in the diameter of the optical signal is mitigated by the optical rod. In the BOSA, the free space between the optical filter and PD blocks was ∼5 mm long. At a propagation of 5 mm, the beam diameter was measured to be ∼288 μm. In contrast, with the optical rod, the measured beam diameter was ∼268 μm. Therefore, it is observed that the alignment tolerance increases with the presence of the optical rod. It is expected that an optical rod with a higher refractive index or longer length will improve the alignment tolerance.

Tables Icon

Table 1. Measured alignment tolerances of the receptacle in the Rx part with and without the optical rod.

To make the BOSA's manufacturing process more reliable, we applied an optical rod between the optical filter and PD blocks, and measured the alignment tolerance of the receptacle in the Rx part. In the measurement, the PD block was used rather than the beam analyzer, as shown in Fig. 4(a). Table 1 lists the measured alignment tolerances of the receptacle in the Rx part with and without the optical rod. The 1-dB alignment tolerances of the X1- and Y1-axis were increased by 10 and 16 μm, respectively, using the optical rod. The coupling efficiency also improved by approximately 3%. This was because the diameter of the optical signal received at the PD block was suppressed by ∼20 μm by the optical rod, as shown in the insets of Fig. 3(b). Thus, the improvement of the alignment tolerance can be enhanced to ±10 μm through optimization of the process. It is expected that the optical rod can mitigate the manufacturing errors that may occur in passive assembly.

2.3 Design of electrical signal path

To guarantee good quality of the 106.25-Gbps PAM-4 signal in the BOSA, we integrated the driver amplifier into the module. For the integration, a wideband bias-T element was stationed between the EML and driver amplifier, as shown in Fig. 5(a). The implemented bias-T element was made of two bias-Ts with a π-shaped topology to form two DC ports (VDD and VEA) integrated into the Tx substrate. The first bias-T was used to provide the power supply voltage (VDD) for biasing the output driving stage in the driver chip, while the second bias-T provided the negative bias (VEA) voltage of the modulator in the EML chip. A 0.1-μF capacitor was used for a DC block, and two ferrite beads were used in series for high-frequency components on the DC feed, as shown in Fig. 5(a). The low-frequency components on the DC feed were placed outside the BOSA, consisting of a 47-μH inductor and a 560-Ω resistor in parallel. Figure 5(b) shows the measured characteristics of the bias-T element. The insertion and return losses were observed to be <2.5 dB and >12 dB, respectively, in the 40 GHz region. We also measured the DC port isolation, which includes the characteristics of the DC feed part placed outside the BOSA. The isolation between the radio frequency (RF) signal (P1 in Fig. 5(a)) port and the DC port (VDD or VEA) was >30 dB up to 40 GHz. We expect that the isolation in the high-frequency region of >40 GHz can be further improved by connecting bypass capacitors on the DC feed line located outside the BOSA.

 figure: Fig. 5.

Fig. 5. (a) Photograph of bias-T with two DC ports on the Tx substrate and (b) measured frequency responses: insertion loss, return loss, and DC port isolation.

Download Full Size | PDF

In the proposed scheme, the 106.25-Gbps PAM-4 BOSA was co-integrated with the optical transmitter and receiver in one package, and the high-speed electrical signals of the Tx and Rx were located on the same side of the module. Accordingly, the design of the signal integrity capable of minimizing the signal interference between the Tx and Rx should be considered. There were two critical paths in the BOSA structure for high-speed signals: (1) the signal path in the package and (2) the signal path in the FPCB for the outside electrical interface. The BOSA package was designed such that Tx and Rx grounds were separated to minimize noise coupling through the common ground path of the Tx and Rx. According to our three-dimensional (3-D) EM simulation results, the crosstalk between the optical transmitter and optical receiver within the package was calculated to be >50 dB up to 50 GHz. The FPCB was 9 mm long, and the separation between the Tx and Rx transmission lines was 3 mm. As shown in the inset (ii) of Fig. 6(a), the FPCB has been proposed to arrange the signal paths of the Tx and Rx on different sides of the FPCB. The proposed structure dramatically reduced the crosstalk between the Tx and Rx parts that occurred in the FPCB. Through 3-D EM simulation of the proposed structure, it was confirmed that the crosstalk occurring at the FPCB was improved by more than 60 dB compared to the conventional structure, as shown in the inset (i) of Fig. 6(a), where Tx and Rx signals were arranged on the same plane. Figure 6(b) shows a cross-sectional view of the fabricated FPCB for high-speed Tx and Rx signals. The grounds of the Tx and Rx parts were separated, and high-speed electrical signals for the Tx and Rx were arranged on the bottom and top sides of the FPCB, respectively.

 figure: Fig. 6.

Fig. 6. (a) Simulated RF characteristics of the conventional and proposed FPCBs and (b) photograph of the fabricated FPCB. Insets (i) and (ii) of (a) show the cross-sectional schematics of the conventional and proposed FPCBs, respectively.

Download Full Size | PDF

Figure 7 shows the frequency response characteristics calculated by the 3-D EM simulation from the structure in which the package and FPCB were connected. Because this was for the characteristic estimation of passive components at the Tx and Rx, active devices such as the driver amplifier and TIA were excluded. The insertion loss (blue curve) and return loss (red curve) of the signal path were calculated to be <1.9 dB and >20 dB up to 50 GHz, respectively. The crosstalk (green curve) between Tx (P1) and Rx (P4) was calculated to be >54 dB up to 50 GHz.

 figure: Fig. 7.

Fig. 7. Simulated RF characteristics of the FPCB with Package. Inset shows the simulation structure.

Download Full Size | PDF

2.4 BOSA implementation

The manufacturing process of the BOSA was divided into three steps as follows: (1) the receptacle and optical rod were passively placed in the package of the BOSA referring to the align marks; (2) the optical filter block was passively attached to the package because of sufficient alignment tolerance [15,16]; and (3) the collimating lens between the EML and optical isolator was finally placed through the active alignment to maximize the Tx optical power. The EML, PD block and optical isolator were passively assembled with the electrical components such as the driver IC and bias-T element. In particular, the EML was attached on the aluminum nitride substrate for the Tx and the optical isolator was mounted on the metal plate (i.e. CuW) for the Tx.

Figure 8(a) shows the inside view of the fully assembled BOSA. The coupling losses of the Tx and Rx for five samples of the BOSA were measured to be 1.8 dB ∼ 2.1 dB and 0.9 dB ∼ 1.1 dB, respectively. The effective responsivity of the Rx was ∼0.55 A/W. Figure 8(b) shows the fabricated BOSA and the evaluation board. The size of the BOSA was 18.4 mm × 8 mm × 5.4 mm. It is suitable for the QSFP28 form factor [17], which is widely utilized in 100 Gbps PAM-4 optical transceiver. The optical interface of the BOSA utilizes the receptacle integrated with a lens [14]. The BOSA was connected to an evaluation board using FPCBs. The FPCBs provided two functions: DC supply for Tx/Rx and transmission of high-speed Tx/Rx signals. The evaluation board was designed to separate the grounds of the Tx and Rx, and to include circuits for the power supply and control signals of the Tx and Rx, respectively. The multilayer evaluation board employed Rogers RO4350B material on the top layer for high-speed signaling and a laminated structure with two layers of FR4 material for power and control signals. The electrical interfaces of the multi-layered evaluation board, as shown in Fig. 8(b), were fabricated using 2.4-mm RF edge-connectors mounted with a single-layer return current path to alleviate signal distortion [4].

 figure: Fig. 8.

Fig. 8. (a) Inside view of the BOSA and (b) photograph of the 106.25-Gbps BOSA with an evaluation board.

Download Full Size | PDF

3. Experimental results and discussions

3.1 Electrical performances

 figure: Fig. 9.

Fig. 9. (a) Experimental setup and (b) EO and OE responses of the BOSA.

Download Full Size | PDF

Figure 9(a) shows the experimental setup using Keysight's Lightwave Component Analyzer (LCA, N4373D) for the electrical-to-optical (EO) and optical-to-electrical (OE) frequency responses of the BOSA. Figure 9(b) shows the measured results of the EO frequency responses of the Tx and the OE frequency responses of the Rx in the BOSA with the evaluation board. The 3-dB EO bandwidth (red curve) and electrical return loss (orange curve) of the Tx were measured to be >32 GHz and >9.7 dB, respectively. Through the opposite connection configuration of Tx, the 3-dB OE bandwidth (blue curve) and electrical return loss (green curve) of the Rx were measured to be >34.6 GHz and >8.9 dB, respectively.

 figure: Fig. 10.

Fig. 10. (a) Experimental setup and (b) electrical crosstalk between the Tx and Rx caused by the Tx, which is ON and OFF.

Download Full Size | PDF

The Tx of the BOSA, in which the driver chip was integrated, would operate a relatively large signal compared to that of the Rx, and this may cause unwanted electrical crosstalk from the Tx to the Rx. The electrical crosstalk was measured through the two electrical ports of the LCA with and without the Tx operation in the configuration shown in Fig. 10(a). The measured electrical crosstalk between the Tx and Rx during the Tx power ON or OFF were >21 dB and >40 dB up to 50 GHz, respectively, as shown in Fig. 10(b). The effect of the electrical crosstalk by the Tx on the Rx can be observed through the bit error ratio (BER) performance, which will be described later.

3.2 Optical performances

 figure: Fig. 11.

Fig. 11. (a) Experimental setup for measuring the performance of Rx in the BOSA and (b) BER performance of the Rx in the BOSA. Insets (i) and (ii) of (b) show the electrical eye diagrams at the output of the Rx while the Tx is not operated and operated at the maximum gain, respectively.

Download Full Size | PDF

Figure 11(a) depicts the experimental setup used to measure the performance of the Rx in the BOSA with and without Tx operation. An optical PAM-4 signal operating at 1271 nm was generated using a tunable laser and a high-speed Mach–Zehnder modulator with a 6-dB bandwidth of ∼60 GHz at the optical transmitter. The signaling rate was set to be 106.25Gbps [13]. The pattern length of the gray-coded PAM-4 signal was 215-1. The optical PAM-4 signal was inserted into the BOSA using an evaluation board. BERs were measured through offline processing while varying the input optical power. Offline processing was performed through data acquisition using a digital serial analyzer (DSA, Tektronix 73304D) [8]. The sampling rate and bandwidth of the DSA were 100 Gsymbol/s and 33 GHz, respectively. The structure of the digital signal processing (DSP) in the off-line processing was shown in the inset of the Fig. 11(a) [8]. At the start of the DSP, the received signal was resampled at the rate of two times of baud rate. The clock of the resampled signal was recovered by finding the timing offset that maximizes the standard deviation of the signals (i.e. maximum eye opening). The decision feedback equalizer (DFE) was applied to improve the signal quality. In the DFE, the number of taps for forward and backward were 13 and 1, respectively. Finally, BER was calculated after Gray decoding. The number of bits for the BER measurement was four million. To validate the effect of the activated Tx on the Rx, the electrical PAM-4 signal generator for driving the Tx was added to the experimental setup, as shown in Fig. 11(a).

Figure 11(b) shows the BER performance of Rx in the BOSA according to the Tx operating conditions. When the Tx was not in operation (blue curve), the receiver sensitivity was <–11.8 dBm at a BER of 2.4e-4, and KP4 forward error correction (FEC) threshold [1,2]. The inset (i) of Fig. 11(b) shows the measured electrical eye diagram at the received optical power of –12 dBm without the Tx operation. Under the optimum gain condition of the driver amplifier in the Tx, the receiver sensitivity was measured to be <–11.4 dBm and the crosstalk penalty caused by the electrical interference of Tx on the Rx was <0.4 dB (red curve). While the Tx was operating with the maximum gain of the driver amplifier, the crosstalk penalty was observed to be <1 dB (green curve). The inset (ii) shows the electrical eye diagram at the received optical power of –12 dBm with the Tx operation of maximum gain. The eye diagrams of insets (i) and (ii) in Fig. 11(b) were obtained through TDECQ equalizer.

 figure: Fig. 12.

Fig. 12. (a) Experimental setup for measuring the performance of Tx in the BOSA and (b) BER performance of the Tx in the BOSA.

Download Full Size | PDF

Figure 12(a) shows the experimental setup used to measure the performance of the Tx under Rx activation. The BER was measured through offline processing of the signal received by a reference ROSA (Ref. ROSA), while varying the input optical power. To measure the transmission characteristics of the Tx, the Ref. ROSA was implemented using the same PD block, TIA, and the package used in the BOSA. Off-line processing was performed through data acquisition using the DSA in the same way of Fig. 11(a) [8]. We also measured the performance of the Tx over a 10-km single-mode fiber (SMF) transmission. The inset of Fig. 12(a) shows the measured eye diagram of the Tx. The transmitter dispersion eye closure quaternary (TDECQ) and the extinction ratio of the measured eye diagram were 2.76 and 7.22 dB, respectively. Figure 12(b) plots the measured BER performances for the back-to-back operation and 10-km SMF transmission. After the 10-km SMF transmission, a power penalty of <0.9 dB was attributed to the chromatic dispersion of the fiber. The performance was consistent with the standards for a 106.25-Gbps PAM-4 signal [1,2].

4. Summary

This is the first implementation of the 106.25-Gbps PAM-4 BOSA integrated with a driver amplifier close to an EML. In the BOSA based on the proposed structure, unlike the conventional structure, the electrical interfaces of the Tx and Rx were placed on one side of the module, and the optical and electrical interfaces were arranged in-line. The BOSA is well suited for application in transceivers operating with high-speed multi-level signals such as the 106.25-Gbps PAM-4 signal, because it allows minimal length of the electrical signal paths inside the transceiver. We realized an efficient manufacturing process for the BOSA by analyzing the alignment tolerance of the receptacle and collimating lens in front of the EML. For an efficient manufacturing process, the BOSA employs an optical rod, which increases the alignment tolerance of the receptacle for the Rx part by >10 μm. The improvement can be increased by using the optical rod with a higher refractive index. For the 106.25-Gbps PAM-4 operation, we integrated the driver amplifier and wideband bias-T element into the BOSA. To suppress the crosstalk between the Tx and Rx electrical parts, we proposed an FPCB design to suppress the electrical crosstalk by >60 dB. We experimentally validated the performance of the BOSA. The measured EO bandwidth of the Tx and OE bandwidth of the Rx were >32 GHz and >34.6 GHz, respectively. While the Tx was operating at the optimum gain of the driver amplifier, the receiver sensitivity was measured to be <–11.4 dBm at the BER of 2.4e–4 (KP4 FEC threshold) and the power penalty caused by the effect of the activated Tx on the Rx was measured to be <0.4 dB. It was also confirmed that the Tx optical signal had a power penalty of <0.9 dB due to chromatic dispersion after 10-km SMF transmission. The proposed BOSA is expected to be applicable to high-speed optical access networks, including 100G and beyond.

Funding

Ministry of Science and ICT, South Korea (2019-0-00002).

Acknowledgments

We give special thanks to Dr. Kwangjoon Kim for valuable discussions and help in the preparation of the manuscript. This work was supported by the Institute of Information & communications Technology Planning & Evaluation (IITP) grant funded by the Korea government (MSIT) (No. 2019-0-00002, Development of Optical Cloud Networking Core Technology).

Disclosures

The authors declare no conflicts of interest.

Data availability

No data were generated or analyzed in the presented research.

References

1. IEEE P802.3bs, “Media access control parameters, physical layers and management parameters for 200 Gbps and 400 Gbps operation,” https://standards.ieee.org/standard/802_3bs-2017.html.

2. . IEEE P802.3cd, “Media access control parameters for 50 Gbps and physical layers and management parameters for 50 Gbps, 100 Gbps, and 200 Gbps operation,” https://standards.ieee.org/standard/802_3cd-2018.html.

3. 100G Lamda MSA, “400G-FR4 technical specification Rev. 2.0,” (2018), http://www.100glambda.com.

4. S. K. Kang, J. Y. Huh, J. Song, and J. K. Lee, “Verification of hybrid-integrated 400-Gb/s (4 × 100 Gb/s) CWDM4 ROSA using a bandwidth-improved multilayer board,” Opt. Express 28(22), 33094–33105 (2020). [CrossRef]  

5. N. Ohata, Y. Kawamoto, M. Binkai, T. Murao, H. Sano, Y. Imai, H. Itamoto, and K. Hasegawa, “A compact integrated LAN-WDM EML TOSA employing stripline with an aperture in the FPC,” J. Lightwave Technol. 38(12), 3246–3251 (2020). [CrossRef]  

6. F. Saliou, P. Chanclou, L. A. Neto, G. Simon, J. Potet, M. Gay, L. Bramerie, and H. Debregeas, “Optical access network interfaces for 5G and beyond [Invited],” J. Opt. Commun. Netw. 13(8), D32–D42 (2021). [CrossRef]  

7. H. H. Lee, K. Kim, K.-H. Doo, J.-Y. Oh, G.-Y. Moon, S.-H. Baek, and H. S. Chung, “Demonstration of high-power budget TDM-PON system with 50 Gb/s PAM4 and saturated SOA,” J. Lightwave Technol. 39(9), 2762–2768 (2021). [CrossRef]  

8. S. R. Moon, M. Sung, J. K. Lee, and S. H. Cho, “Cost-effective photonics-based THz wireless transmission using PAM-N signals in the 0.3-THz band,” J. Lightwave Technol. 39(2), 357–362 (2021). [CrossRef]  

9. IEEE P802.3cp, “Bidirectional 10 Gbps, 25 Gbps, and 50 Gbps optical access PHYs,” https://standards.ieee.org/standard/802_3cp-2021.html.

10. T.-T. Shih, M.-C. Lin, P.-H. Tseng, C.-Y. Li, T.-Y. Hung, Y.-J. Chiu, and W.-H. Cheng, “High-performance and low-cost 10-Gb/s bidirectional optical subassembly modules,” J. Lightwave Technol. 25(11), 3488–3494 (2007). [CrossRef]  

11. H. Wu, Z. Huang, Z. Yu, X.-S. Qian, B. Chen, and Y. Lu, “The efficiency improvement of a bidirectional optical subassembly,” in Asia Pac. Conference on Communications (APOC),2009, pp. 597–600.

12. K. Asaka, A. Kanda, A. Ohki, T. Kurosaki, R. Yoshimura, H. Sanjoh, T. Ito, M. Nakamura, and M. Yoneyama, “Low-cost and small BOSA employing impedance matching circuits for 10G-EPON,” in Opto-Electronics and Communications Conference (OECC), 2012, pp. 81–82.

13. J. Y. Huh, S. K. Kang, S. H. Yoo, J. Song, and J. K. Lee, “Bidirectional optical transceiver module,” US Patent 17/204341 (17 March 2021).

14. S. K. Kang, J. K. Lee, J. Y. Huh, and J. H. Lee, “Receptacle-collimator assembly and multi-wavelength optical receiver module,” US Patent 9 804 346 (31 October 2017).

15. J. Y. Huh, S. K. Kang, J. H. Lee, J. K. Lee, and S. Kim, “Highly alignment tolerant and high-sensitivity 100 Gb/s (4 × 25 Gb/s) APD-ROSA with a thin-film filter-based de-multiplexer,” Opt. Express 24(24), 27104–27114 (2016). [CrossRef]  

16. J. K. Lee, J. Y. Huh, S. K. Kang, and Y. S. Jang, “Analysis of dimensional tolerance for an optical demultiplexer of a highly alignment tolerant 4 × 25 Gb/s ROSA module,” Opt. Express 22(4), 4307–4315 (2014). [CrossRef]  

17. SFF Committee, “QSFP+ 28 Gb/s 4X Pluggable Transceiver Solution (QSFP28) Rev. 1.9,” (2015), https://www.snia.org/technology-communities/sff/specifications.

Data availability

No data were generated or analyzed in the presented research.

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 (12)

Fig. 1.
Fig. 1. Components placement of the transceivers using (a) a conventional BOSA and (b) a proposed BOSA.
Fig. 2.
Fig. 2. Conceptual scheme of the proposed BOSA.
Fig. 3.
Fig. 3. (a) Experimental schematic to measure the alignment tolerances of the receptacle and collimating lens, (b) coupling efficiencies as a function of the X1- and Y1-axis displacements of the receptacle, and (c) coupling efficiencies as a function of the X2- and Y2-axis displacements of the collimating lens in front of the EML.
Fig. 4.
Fig. 4. (a) Effect of an optical rod and (b) beam diameters of the collimating light as a function of propagation distance from the optical filter block with and without the optical rod. Insets of (b) show the measured beam shapes after 5-mm propagation with and without the optical rod, respectively.
Fig. 5.
Fig. 5. (a) Photograph of bias-T with two DC ports on the Tx substrate and (b) measured frequency responses: insertion loss, return loss, and DC port isolation.
Fig. 6.
Fig. 6. (a) Simulated RF characteristics of the conventional and proposed FPCBs and (b) photograph of the fabricated FPCB. Insets (i) and (ii) of (a) show the cross-sectional schematics of the conventional and proposed FPCBs, respectively.
Fig. 7.
Fig. 7. Simulated RF characteristics of the FPCB with Package. Inset shows the simulation structure.
Fig. 8.
Fig. 8. (a) Inside view of the BOSA and (b) photograph of the 106.25-Gbps BOSA with an evaluation board.
Fig. 9.
Fig. 9. (a) Experimental setup and (b) EO and OE responses of the BOSA.
Fig. 10.
Fig. 10. (a) Experimental setup and (b) electrical crosstalk between the Tx and Rx caused by the Tx, which is ON and OFF.
Fig. 11.
Fig. 11. (a) Experimental setup for measuring the performance of Rx in the BOSA and (b) BER performance of the Rx in the BOSA. Insets (i) and (ii) of (b) show the electrical eye diagrams at the output of the Rx while the Tx is not operated and operated at the maximum gain, respectively.
Fig. 12.
Fig. 12. (a) Experimental setup for measuring the performance of Tx in the BOSA and (b) BER performance of the Tx in the BOSA.

Tables (1)

Tables Icon

Table 1. Measured alignment tolerances of the receptacle in the Rx part with and without the optical rod.

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.