Skip to main content

What is an optical module

What is an optical module:

Optical Module (Optical Module), as an important part of optical fiber communication, is an optoelectronic device that realizes the functions of photoelectric conversion and electro-optical conversion.


To be precise, the optical module is a collective name for several types of modules, including: optical transmitting module Transmitter, optical receiving module Receiver, optical transceiver integrated module Transceiver and optical forwarding module Transponder. Generally, what we call optical module generally refers to an integrated optical transceiver module (the same below).


optical module working principle:

The optical module works at the physical layer of the OSI model and is one of the core components in the optical fiber communication system. It is mainly composed of optoelectronic devices (optical transmitters, optical receivers), functional circuits and optical (electrical) interfaces. Its main function is to realize the photoelectric conversion and electro-optical conversion functions in optical fiber communication.


Principle of Optical Module


Figure: Schematic diagram of optical module

Basic principle: The transmission interface inputs an electrical signal with a certain code rate. After being processed by the internal drive chip, the drive semiconductor laser (LD) or light emitting diode (LED) emits a modulated light signal of the corresponding rate. After transmission through the optical fiber, the receiving interface The optical signal is converted into an electric signal by the light detection diode, and the electric signal of the corresponding code rate is output after the pre-amplifier.


Key parameters of optical module

The key technical indicators of optical modules mainly include: transmit optical power, receive optical power, overload optical power, maximum receiving sensitivity, and extinction ratio.


Receiving optical power: Transmitting optical power refers to the optical power output by the light source at the transmitting end of the optical module under normal working conditions. It can be understood as the intensity of light in W or mW or dBm. Where W or mW is a linear unit, and dBm is a logarithmic unit. In communication, we usually use dBm to indicate optical power, and 0dBm optical power corresponds to 1mW.


Received optical power: Received optical power refers to the average optical power range that the receiving end component can receive under a certain bit error rate (BER=10-12) condition of the optical module. The upper limit is the overload optical power, and the lower limit is the maximum receiving sensitivity.


Overload optical power: Also known as saturated optical power, it refers to the maximum input average optical power that can be received by the receiving end component when a certain bit error rate (BER=10-12) is maintained at a certain transmission rate, unit: dBm .


It should be noted that the photodetector will experience photocurrent saturation under strong light irradiation. When this phenomenon occurs, the detector needs a certain time to recover. At this time, the receiving sensitivity decreases, and the received signal may be misjudged. It can cause bit errors, and it is also very easy to damage the receiving end detector. During operation, it is necessary to avoid strong light irradiation to prevent exceeding the saturated optical power.


Maximum Receiving Sensitivity: Receiving sensitivity refers to the minimum average input optical power that can be received by the receiving end components when a certain bit error rate (BER=10-12) is maintained at a certain transmission rate, unit: dBm.


In general, the higher the rate, the worse the receiving sensitivity, that is, the greater the minimum received optical power, and the higher the requirements for the receiving end of the optical module.


Extinction ratio: Extinction Ratio (EXT) is one of the important parameters used to measure the quality of optical modules. It refers to the ratio of the optical power P1 when the laser emits all "1" codes to the optical power P0 when all "0" codes are emitted under the condition of full modulation, and the unit is dB. The extinction ratio reflects the relative amplitude of the optical signal "1" level and "0" level. The factors affecting the extinction ratio in the optical module are the bias current (Bias) and the modulation current (Mod). The extinction ratio can be regarded as the value of EXT=Bias/Mod.


The value of the extinction ratio is not that the larger the optical module, the better, but the optical module with the extinction ratio that meets the IEEE 802.3 standard.


Detailed description of each parameter:



Common information

indicates the general information of the optical module.

Transceiver Type

indicates the type of optical module.

Connector Type

represents the interface type.

Wavelength (nm)

represents the wavelength of light waves.

Transfer Distance (m)

represents the transmission distance of light waves. 50um/125um indicates the diameter of the optical fiber, and OM2 indicates the grade of the optical fiber.

Digital Diagnostic Monitoring

indicates whether the diagnostic information of the optical module is monitored.


Vendor Name

represents the name of the optical module manufacturer. If the display content is "HUAWEI", it means that it is an optical module certified by Huawei data center switches; the other display content indicates that it is an optical module that is not certified for Huawei data center switches.

Vendor Part Number

indicates the manufacturer's number of the optical module.

Ordering Name

 indicates the external model of the optical module.

Manufacture information

represents optical module manufacturing information.

Manu. Serial Number

represents the production serial number of the optical module.

Manufacturing Date

represents the production date of the optical module.

Alarm information

indicates the alarm information of the optical module.

Diagnostic information

 indicates the diagnostic information of the optical module. If it is displayed as "-", it means that the optical module does not support obtaining this information or the information is not accurate.

Temperature (Celsius)

indicates the current temperature of the optical module.


represents the current voltage of the optical module.

Bias Current (mA)

 indicates the current current of the optical module. Note: If the interface supports splitting, when the interface is inserted into the optical module, the current current of each lane in the optical module will be displayed. The current current value of each lane should be within the range of Bias Low Threshold (mA)~Bias High Threshold (mA) to ensure the normal operation of the module.

Bias High Threshold (mA)

indicates the upper limit of the current of the optical module.

Bias Low Threshold (mA)

indicates the lower limit of the current of the optical module.

Current RX Power (dBm)

represents the current received power of the optical module. Note: If the interface supports splitting, when the interface is inserted into the optical module, the current received power of each lane in the optical module will be displayed.

Default RX Power High

Threshold (dBm) indicates the upper limit of the default received power of the optical module.

Default RX Power Low

Threshold (dBm) indicates the lower limit of the default received power of the optical module.

Current TX Power (dBm)

represents the current transmit power of the optical module. Note: If the interface supports splitting, when the interface is inserted into the optical module, the current transmit power of each lane in the optical module will be displayed.

Default TX Power High

Threshold (dBm) indicates the upper limit of the default transmit power of the optical module.

Default TX Power Low

 Threshold (dBm) indicates the lower limit of the default transmit power of the optical module.







Classification and packaging of optical modules

Classified by rate:

The common types of optical modules are as follows:


400GE optical module

200GE optical module

100GE optical module

40GE optical module

25GE optical module

10GE optical module

GE optical module

FE optical module

Classified by package type:

The higher the transmission rate, the more complex the structure, resulting in different packaging methods. There are SFP/eSFP, SFP+, SFP28, QSFP+, CXP, CFP, QSFP28, etc.


SFP (Small Form-factor Pluggable) optical module: small and pluggable. The SFP optical module supports LC fiber connectors.

eSFP (Enhanced Small Form-factor Pluggable) optical module: Enhanced SFP refers to SFP with monitoring functions of voltage, temperature, bias current, transmit optical power, and receive optical power. All current SFPs are equipped with it, so All eSFPs are collectively called SFP.

SFP+ (Small Form-factor Pluggable Plus) optical module: refers to the SFP module with increased speed. Because of the increased speed, it is sensitive to EMI. There are more skirts on the shell and the matching cage is relatively tightened.

XFP (10GB Small Form-factor Pluggable) optical module: "X" is the abbreviation of Roman numeral 10. All XFP modules are 10GE optical modules. XFP optical modules support LC fiber connectors. Compared with SFP+ optical modules, XFP optical modules are wider and longer.

SFP28 (Small Form-factor Pluggable 28) optical module: The interface package size is the same as that of SFP+, and it supports 25G SFP28 optical modules and 10G SFP+ optical modules.

QSFP+ (Quad Small Form-factor Pluggable) optical module: four-channel small form-factor pluggable optical module. QSFP+ optical modules support MPO fiber connectors, which are larger in size than SFP+ optical modules.

CXP (120 Gb/s eXtended-capability Form Factor Pluggable Module) optical module: It is a hot-swappable high-density parallel optical module standard, providing 12 channels each in the sending and receiving (Tx/Rx) directions, only applicable to Short-distance multi-mode link.

CFP (Centum Form-factor Pluggable) optical module: The size of length×width×height is defined as 144.75mm×82mm×13.6mm. It is a high-speed, hot-swappable, new type of optical module that supports data communication and telecommunications transmission. standard.

QSFP28 (Quad Small Form-factor Pluggable 28) optical module: The interface package size is the same as QSFP+, and it supports 100G QSFP28 optical modules and 40G QSFP+ optical modules.

Classified by mode:

Fiber is divided into single-mode fiber and multi-mode fiber. In order to use different types of optical fibers, single-mode optical modules and multi-mode optical modules have been produced.


Single-mode optical modules are used in conjunction with single-mode optical fibers. The single-mode fiber has a wide transmission frequency and large transmission capacity, which is suitable for long-distance transmission.

Multimode optical modules are used in conjunction with multimode optical fibers. Multimode fiber has the defect of modal dispersion, and its transmission performance is worse than single-mode fiber, but the cost is lower, and it is suitable for small capacity and short-distance transmission.


The center wavelength refers to the optical band used for optical signal transmission. At present, there are mainly three types of central wavelengths of commonly used optical modules: 850nm band, 1310nm band and 1550nm band.


850nm band: mostly used for short-distance transmission ≤2km


1310nm and 1550nm bands: mostly used for medium and long-distance transmission, transmission above 2km.

Classified by transmission distance:

According to the transmission distance of the optical module, it can be roughly divided into:


For short-distance optical modules, 2km and below are generally regarded as short-distance.

Medium distance optical module, 10~20km is medium distance.

Long-distance optical module: generally refers to an optical module with a transmission distance of more than 30Km.

The transmission distance of the optical module is limited, mainly because the optical signal will have a certain loss and dispersion during fiber transmission.


Loss is the loss of light energy due to the absorption, scattering and leakage of the medium when light is transmitted in the optical fiber. This part of the energy is dissipated at a certain rate as the transmission distance increases.

The generation of chromatic dispersion is mainly due to the unequal speed of electromagnetic waves of different wavelengths when propagating in the same medium, which causes the different wavelength components of the optical signal to arrive at the receiving end at different times due to the accumulation of transmission distance, resulting in pulse broadening and inability to distinguish the signal value.

Other optical modules:

Color light module:


The biggest difference between color optical modules and other types of optical modules is the central wavelength. The central wavelengths of general optical modules are 850nm, 1310nm and 1550nm. The colored optical module carries light of several different central wavelengths. Color optical modules are divided into two types: coarse-collected optical modules (CWDM) and dense-wave optical modules (DWDM). In the same waveband, there are more types of dense wave optical modules, so dense wave optical modules make fuller use of waveband resources. Lights with different central wavelengths can be transmitted without interference in the same fiber. Therefore, the light from the multiple color optical modules with different central wavelengths is combined and transmitted all the way through the passive combiner, and the remote end uses the splitter according to Different central wavelengths split the light into multiple paths, effectively saving fiber optic lines. Color optical modules are mainly used in long-distance transmission lines.


When using a long-distance optical module, the transmit optical power is generally greater than the overload optical power, so you need to pay attention to the fiber length to ensure that the actual received optical power is less than the overload optical power. If the length of the optical fiber is short, the long-distance optical module needs to be used with light attenuation. Be careful not to burn the optical module.


Optoelectronic Module


Photoelectric modules are usually called electrical modules, also known as light-to-electric modules, RJ45 modules. Unlike optical modules, electrical modules do not perform photoelectric conversion. Through the switch of the electrical module, the two optical interfaces can be connected with a network cable. At present, Huawei only provides GE electrical modules, the interface is RJ45 interface, using Category 5 network cable, support

Support 1000BASE-T (IEEE 802.3ab) standard, the maximum transmission distance is 100m.


Naming rules for optical modules

Naming rules for 100G optical modules:

There are mainly two key standard organizations for 100G optical modules, IEEE and MSA, which complement each other and learn from each other. The standards starting with 100GBASE are all proposed by IEEE802.3, and the naming rules are as follows:


Optical module naming rules


Figure: Optical module naming rules

The specific rules for each field are as follows:


The first end: XXX, which means rate and rate standard; 100 means 100GE.

m: Represents the transmission distance. Common distances are as follows:

KR: indicates that the transmission distance is 10cm, and K is backplane, which is the signal transmission distance between backplanes.

CR: indicates that the transmission distance is meter level, C means copper, high-speed cable connection.

SR: indicates that the transmission distance is 10m class, S class short, short distance transmission, generally multimode fiber.

DR: Indicates that the transmission distance is 500m. PSM4 is a 500-meter transmission, but it does not belong to the IEEE standard system.

FR: Indicates that the transmission distance is 2km. Usually CWDM single mode

LR: Indicates that the transmission distance is 10km, and L is Long. Single mode fiber

ER: Indicates that the transmission distance is 40km, and E means Extended.

ZR: Indicates that the transmission distance is 80km.

n: indicates the number of channels, indicating the number of SerDes channels occupied by 100GE.

4: Indicates that 4 SerDes channels are occupied, that is, 4*25GE.

10: Represents the occupied 10 SerDes channels, namely 10*10GE.

In addition to the above rules, there will generally be package types in the back.


Evolution of optical modules

The development of Ethernet has experienced rapid growth of 1Mbit/s, 10Mbit/s, 100Mbit/s (FE), 1Gbit/s (GE), 10Gbit/s (10GE) to 40Gbit/s (40GE), 100Gbit\s (100GE) Changes, with the rapid development of big data, smart cities, mobile Internet, cloud computing and other services, network traffic has shown exponential growth. The desire for continuous increase in bandwidth will require higher bandwidth rates, and optical modules will also develop rapidly.


In the current physical architecture network of mainstream data centers, the Spine-Leaf (Clos network architecture) architecture is generally followed. The 10GE interface is usually used as the access side server for interconnection, and the uplink on the Leaf side generally uses the 40GE interface. In large data centers, 25G has been widely used as mainstream access and 100G uplinks. In scenarios that require high computing and high bandwidth, using RDMA technology, GPU servers, etc. have used 100GE or even 200GE access. Data center switch interconnection is evolving to large-scale 400GE interconnection.


Clos network can refer to:


25GE standard:

The 25GE standard solution was born at the IEEE Beijing Conference in 2014. Microsoft took the lead in putting forward the 25GE project requirements for the development of TOR and Server interconnection scenarios. However, the IEEE conference rejected the reason that 25GE would decentralize investment in the industry and be unfavorable to industry development. The project requirements of the 25GE standard have been met. Since the 25GE solution can solve a series of problems such as CPU performance improvement and PICE bit bandwidth matching caused by the migration of server network cards from 10Gbit/s to 40Gbit/s, vendors such as Microsoft and Qualcomm independently established the 25GE Ethernet Alliance and invested in the research of 25GE solutions . IEEE does not want 25GE to become a de facto standard outside the organization standard, and passed the 25GE project in July of the same year. With the development of the network, facts have proved that the 25GE specification can cost-effectively expand network bandwidth, provide better support for a new generation of server and storage solutions, and will cover more interconnected scenarios in the future.


25GE standard technology is currently mainly used for server access in data centers. From the following aspects, why the access rate of the data center network is 25Gbit/s instead of the existing 40Gbit/s.


1. The natural advantages of technology realization.

​ When it comes to 25GE standard technology, you must mention SerDes (serializer/deserializer). SerDes is widely used in various circuits and optical fiber communication technologies, from PCIe used in computers to interconnections between network cards and internal chips in switches, all using SerDes connections. It can be said that all high-speed devices are connected using SerDes serial components and converted into final receiver data. The number of SerDes connections required for a switch port is called "Lane".


​ After years of technological development, SerDes speed can reach 25Gbit/s, that is to say, from the 25Gbit/s network card, through the switch to the other end of the 25Gbit/s network card, all end-to-end connections only need to use a 25Gbit/s The SerDes connection channel of the s rate is sufficient, while the 40GE port adopts the QSFP package type and is composed of 4 parallel 10GE links (each 10GE uses 12.5GHz SerDes) and requires 4 SerDes channels.


​ In addition, 100GE ports have become mainstream at the convergence layer and backbone layer. In the early exploration, there has been a mature IEEE 100Gbit/s Ethernet standard, which includes 4 channels of 25 Gbit/s electronic signals. It is realized by running 4 25Gbit/s channels (IEEE 802.3bj) on the optical fiber or copper cable pair, laying a certain foundation for the birth of the 25GE standard solution. Moreover, the 100GE port can be converted into four 25GE ports through a QSPF28 to SFP28 one-to-four cable. Compared with the port matching, it also has obvious advantages compared to 40GE.


2. Improved switch performance

​ If 40GE is a product of the 10Gbit/s rate era, then 25GE is the general trend of technology. Compared with the existing 10GE solutions, the single-channel 25GE improves the performance by 2.5 times. At the same time, 25GE has a higher port density than 40GE solutions connected to rack servers.


The technical parameters of 25GE standard and 40GE standard are as follows:

Technical parameters

Channel (Lane)

SerDes (Gbit/s)

Package type

 Connector type

Number of fibers/piece















      3. Smooth evolution of existing topologies, reducing costs

​ Capital expenditure (CapEx) is one of the key considerations for the adoption of any new data center technology. One aspect that companies care about most about data center deployment is cabling. Many engineers think that the most complicated and difficult part of data center management is wiring. Most engineers hope to not touch them after installing cables once.


​ 40GE switches use QSFP+ encapsulated optical modules. The connection inside the cabinet or adjacent cabinets can use QSFP+ DAC cables. For further connections, QSFP+ optical modules must be used with MPO optical cables for transmission. Modules generally use 12-core optical fiber. Compared with 10GE interface, the cost of two-core LC interface optical fiber is greatly increased, and it is completely incompatible. If you consider upgrading to 40GE based on the existing 10GE, all optical fiber cables must be discarded and MPO optical cables are used for rewiring.


​ The SFP28 package type optical module used on the 25GE switch is the same as the 10GE SFP package type optical module. It only uses a single-channel connection and is compatible with the existing topology of the LC connector type optical fiber. Compared with upgrading from 10GE to 40GE, if you upgrade to 25GE, it supports seamless migration from 10GE Ethernet without re-planning the topology and rewiring. After the equipment is upgraded or replaced with 25GE optical modules, it can be plug and play. Save worry and effort.


​ Compared with the 40Gbit/s solution for rack service connection, the 25GE standard solution can not only use the current 10GE topology to smoothly evolve, but also greatly reduce the purchase support of TOR switches and cables, reducing power, heat dissipation and The need for space.


100GE standard:

​ When it comes to the 100GE standard, one has to talk about the optical module standardization organization. In the era when the optical module industry was just getting started, the industrial chain was rather chaotic, and each manufacturer had its own package structure type, and the size and appearance were also varied. As an official organization, IEEE 802.3 working group has played a key role in the unification of optical module standards. Different from the official organization IEEE, MSA (Multi Source Agreement) can be regarded as an unofficial organizational form. As an enterprise alliance in the industry, it forms a consistent agreement for different optical module standards and defines a unified optical module structure package (encapsulation). Type, appearance size, pin assignment, etc.).


​ As early as 2006, the IEEE established a working group with the goal of studying the next-generation high-speed Ethernet 100GE standard, and in 2012 issued multiple standards for 100GE. In order to meet the requirements of 100GE uplink scenarios at different distances, there are more than 10 100GE standards defined by IEEE and MSA. As shown in the following table, the following are mainly several mainstream standards in data center networks.


setting agency


Optical fiber type

Transmission distance


IEEE 802.3

24-core MPO

multimode fiber , center wavelength 850nm

Multimode (OM2) fiber: 30m; Multimode (OM3) fiber: 100M; Multimode (OM4) fiber: 150M;


IEEE 802.3


single-mode fiber, center wavelength 1295.56~1309.14nm

single-mode (G.652) fiber: 10km



IEEE 802.3


single-mode fiber, center wavelength 1295.56~1309.14nm

   single-mode (G.652) fiber: 40km



IEEE 802.3

8/12 core MPO

 multimode fiber, center wavelength 850nm

 Multimode (OM3) fiber: 70m; Multimode (OM4) fiber: 100m;


100G PSM4


12-core MPO

single-mode fiber, center wavelength 1310nm

single-mode (G.652) fiber: 500km


100G CWDM4



single-mode fiber , center wavelength 1310nm

 single-mode (G.652) fiber: 2km



          The 100GBASE series standards are all formulated by IEEE 802.3, and the specific naming rules have been listed in the section naming optical modules.


​ The transmission characteristics of optical fibers and the manufacturing cost of optical modules determine different application scenarios. Multi-mode is often used for short-distance transmission, and single-mode is often used for long-distance transmission. From the previous summary, IEEE’s 100GBASE series standards are sufficient to cover long and short distance data center transmissions. 100GBASE-SR4 and 100GBASE-LR4 are the most commonly used standards defined by IEEE. However, in most data center internal interconnection scenarios, the distance supported by 100GBASE-SR4 is too short, and the cost of 100GBASE-LR4 is too high. The PSM and CWDM4 standards proposed by MSA perfectly solve the cost problem in mid-distance transmission scenarios.


​ CWDM4 uses optical devices MUX and DEMUX to multiplex 4 parallel 25Gbit/s channel peaks onto a 100Gbit/s optical fiber link. This is similar to LR4 with the following differences.


Different channel intervals:


CWDM4 defines a 20nm channel interval, while LR4 defines a 4.5nm LAC-WDM interval. The greater the channel spacing, the lower the requirements for optical devices, and the lower the cost.


Different lasers:


CWDM uses DML (Direct Modulated Laser), which is a single laser. The EML (Electro-absorption Modulated Laser) used by LR4 is a device composed of DML and EAM.


Different temperature control requirements:


Since the channel interval of LR4 is 4.5nm, the laser needs to be equipped with TEC (Thermo Electric Cooler, semiconductor thermoelectric cooler) Driver chip.


​ To sum up the above 3 points, the cost of 100GBASE-LR4 standard optical module is higher than that of 100G WDM4. In addition to CWDM, PSM4 is also an option for medium-distance transmission. The 100G PSM4 specification defines a point-to-point 100Gbit/s link with 8 single-mode fibers (4 transmissions and 4 receptions). Each channel is transmitted at a rate of 25Gbit/s, and each signal direction uses 4 independent channels with the same wavelength.


​ Because CWDM4 uses a wavelength division multiplexer, the cost of optical modules is higher than that of PSM4. But when sending and receiving signals, only two single-film fibers are needed, which is far less than the 8 single-film fibers required by PSM4. As the transmission distance increases, the cost of PSM4 increases.


​ The complete optical module solution not only includes the optical and electrical interface standards of the optical module, but also requires supporting structural packaging. As shown in the table below, the first proposed packaging format is CFP, but due to size issues, with the improvement of optical module integration, CFP can evolve to CFP2, CFP4, and then to the popular QSFP28. The overall development of optical modules shows a high rate , The trend of high density, low cost and low power consumption.


The following table shows the evolution trend of 100GE optical module packaging format:

Package Format

Power Consumption

Channel * Rate





10 10Gbit/s or 4 25Gbit/s

 Large size, high power consumption, long transmission distance



4 * 25Gbit/s

large size, high power consumption, long transmission distance



4 * 25Gbit/s

Small size and low power consumption



4 * 25Gbit/s

Small size, low power consumption


        ​ After several generations of development, the development of 100GE optical modules has matured. In response to some new technology applications and new development directions, new 100G MSAs have been established and standardized to promote the sustainable development of related industrial chains. For the network, we face endless challenges with higher bandwidth and lower latency.


400GE standard:

​The 400GE Ethernet standard is still under the responsibility of IEEE 802.3. Since 2013, the IEEE has achieved the establishment of the 400GE standard and started a study group to discuss the 400GE specifications periodically. After many technical competitions and program meetings, the 400GE and 200GE standards IEEE 802.3bs were officially released. The key technologies are the definition of hierarchical structure, FEC specification and physical optical interface transmission mechanism. The physical layer technical solutions and transmission distances mainly adopted by the 400GE standard are shown in the following table.


Transmission distance

Encoding method



16 * 25Gbit/s NRZ



4 * 100Gbit/s PAM4



8 * 50Gbit/s PAM4



 8 * 50Gbit/s PAM4




​ Among them, SR16 based on multimode fiber is basically unpopular, and DR4, FR8 and LR8 based on PAM4 electrical signal modulation technology have become the focus of attention. The area is based on the NRZ signal transmission technology commonly used in the previous 100GE standard (using high and low levels to represent digital logic signals 0 and 1), PAM4 uses 4 different signal levels for transmission, and each clock cycle can transmit 2bit Logical information (ie 00, 01, 10, 11). Therefore, under the same baud rate, the transmission efficiency of PAM4 is twice that of NRZ signals. Formally because of the efficient transmission efficiency of PAM4, IEEE standardized it as an electrical signal standard of the 400GE standard.


​ As mentioned earlier, the SerDes speed can reach 25Gbit/s, and the corresponding bit rate can be 50Gbit/s through PAM4 modulation, so the encoding technology in the IEEE 802.3 400GE/200GE interface is usually 50Gbit/s/lane PAM4 encoding technology.


Comprehensive comparison of 400GE optical module packaging formats:

Package format


 Power consumption/W

Electrical interface channel

Optical interface channel



107.5 41.5 12.5


 16 * 25 Gbit/s

 8 * 56 Gbit/s


107.8 22.6 13.0


 8 * 56 Gbit/s

8 * 56 Gbit/s


89.4 18.4 8.5


 8 * 56 Gbit/s

4 * 100Gbit/s


      ​ For the above three types of 400G optical module packaging structures, the market currently supports OSFP-DD relatively the most, and its ecosystem is also the most popular. The most common 400G optical module produced by mainstream optical module manufacturers is the OSFP-DD package. It also includes mainstream switches and mainstream commercial PHY chips that support OSFP-DD most widely. Both OSFP-DD and OSFP are electrical ports that support 8 50G PAM4 signals, and optical ports can support up to 8 parallel channels with low energy consumption. CFP8 electrical ports support 16 25G NRZ signals, and optical ports support up to 16 parallel channels with relatively low power consumption. high.


​ Compared with the 400GE standard, although 200GE started late, from the current situation, all manufacturers are more inclined to adopt the 4 * 56 Gbit/s PAM4 method, and the optical devices can use the existing 28Gbit/s devices. From an application point of view, the 200GE optical module is less difficult to implement, and the application in the data center scenario may be earlier than 400GE, which is the possibility of a window reserved for its own positioning. However, with the continuous in-depth research of single-wave 100GE technology, short-distance 400GE optical modules will be more favored in new scenarios in data.


Higher rate 800GE, 1.6T:

Faced with new applications, the speed of Ethernet is also increasing. As shown in the figure below, from the initial 10M, 100M to the recently standardized 400G, the interface speed has increased by 40,000 times. To further respond to the demand for doubling the switch capacity of the data center every two years, in 2018, the Ethernet Alliance has made it clear that in the next few years, it will launch the next generation of Ethernet speeds, 800G and 1.6T.


There are two ways to increase the bandwidth of optical modules:


Increase the bit rate of each channel


Increase the number of channels.


10G to 40G, the increase is the number of channels. From 40G to 100G, the baud rate of the single channel (10G->25G) is improved.



Popular posts from this blog

AMD's GPU technology enters the mobile phone chip market for the first time

In addition to the release of the Exynos2100 processor, Samsung also confirmed a major event at this Exynos event, that is, the custom GPU that they have worked with AMD for many years will soon appear and will be used on the next flagship machine. The current Exynos2100 processor uses ARM’s Mali-G78GPU core with a total of 14 cores, so the GPU architecture developed by Samsung will be the next Exynos processor, and the GPU will be the focus. This is probably the meaning of Exynos2100’s GPU stacking. The key reason. Dr. InyupKang, president of Samsung’s LSI business, confirmed that the next-generation mobile GPU in cooperation with AMD will be used in the next flagship product, but he did not specify which product. Samsung is not talking about the next-generation flagship but the next one, so it is very likely that a new Exynos processor will be available this year, either for the GalaxyNote21 series or the new generation of folding screen GalaxyZFold3. In 2019, AMD and Samsung reached

Apple and Intel want to join the game, what happened to the GPU market?

Intel recently announced that it will launch Xe-LP GPU at the end of this year, officially entering the independent GPU market, and will hand over to TSMC for foundry. At the 2020 WWDC held not long ago, Apple also revealed that it is possible to abandon AMD's GPU and use a self-developed solution based on the ARM architecture. It will launch a self-developed GPU next year. What happened to the GPU market? Why are the giants entering the game?    Massive data calls for high-performance GPU    Why has the demand for GPUs increased so rapidly in recent years? Because we are entering an era where everything needs to be visualized. Dai Shuyu, a partner of Aiwa (Beijing) Technology Co., Ltd., told a reporter from China Electronics News that visualization requires a large amount of graphics and image computing capabilities, and a large amount of high-performance image processing capabilities are required for both the cloud and the edge.    Aiwa (Beijing) Technology Co., Ltd. is an enterp

NVIDIA officially launches RTX 30 series mobile graphics cards

In the early morning of January 13, NVIDIA officially launched the RTX30 series of mobile graphics cards at the CES2021 exhibition. Ampere-based GPUs have also reached the mobile terminal, mainly including RTX3080, RTX3070 and RTX3060 models. In addition to improving game performance, the RTX30 series of mobile graphics cards have twice the energy efficiency of the previous generation, and support the third-generation Max-Q technology, mainly supporting DynamicBoost2.0 dynamic acceleration technology, WisperMode2.0 noise control, ResizableBAR (similar to AMD’s SAM technology) and DLSS. The third-generation Max-Q technology uses AI and new system optimization to make high-performance gaming laptops faster and more powerful than ever. These technologies include: ·DynamicBoost2.0: The CPU and GPU powers of traditional gaming notebooks are fixed, while games and creative applications are dynamic, and the requirements for the system will vary with the number of frames. With DynamicBoost2.0,