CN203733105U - Optical fiber reflection memory card and optical fiber reflection memory network - Google Patents

Optical fiber reflection memory card and optical fiber reflection memory network Download PDF

Info

Publication number
CN203733105U
CN203733105U CN201420009177.3U CN201420009177U CN203733105U CN 203733105 U CN203733105 U CN 203733105U CN 201420009177 U CN201420009177 U CN 201420009177U CN 203733105 U CN203733105 U CN 203733105U
Authority
CN
China
Prior art keywords
module
card
data
random access
access memory
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired - Lifetime
Application number
CN201420009177.3U
Other languages
Chinese (zh)
Inventor
陆兆辉
刘文利
刘勇
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
CETC 8 Research Institute
Original Assignee
CETC 8 Research Institute
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by CETC 8 Research Institute filed Critical CETC 8 Research Institute
Priority to CN201420009177.3U priority Critical patent/CN203733105U/en
Application granted granted Critical
Publication of CN203733105U publication Critical patent/CN203733105U/en
Anticipated expiration legal-status Critical
Expired - Lifetime legal-status Critical Current

Links

Landscapes

  • Multi Processors (AREA)

Abstract

Disclosed are an optical fiber reflection memory card and an optical fiber reflection memory network. The optical fiber reflection memory card comprises a PCI interface control module, a master control module, a memory module, a serializing/deserializing module and a photoelectric module. The master control module is used for receiving high-speed parallel data sent back by the serializing/deserializing module and provides high-speed data and a control interface for the memory module, the memory module is used for storing local data and data received from the network in corresponding addresses according to different node numbers, the serializing/deserializing module is used for serializing multi-channel data and deserializing the received serial data, the photoelectric module is used for converting high-speed serial electric signals sent by the serializing/deserializing module into optical signals, and meanwhile the optical signals fed by an external optical fiber network are restored into the high-speed serial electric signals and then are fed into the serializing/deserializing module. The PCI interface control module is used for being communicated with an external computer through a PCI bus.

Description

Fiber reflection RAM (random access memory) card and fiber reflection internal memory net
Technical field
The utility model relates to a kind of digital reflex RAM (random access memory) card and Reflective memory network, and particularly a kind of low transmission delay that can be used for real-time Communication for Power, high transfer rate, single mode long distance have fiber reflection RAM (random access memory) card and the fiber reflection internal memory net of determinacy and predictability from, transmission delay.。
Background technology
Flight simulator is a kind of typical distributed real-time flight simulation system.Flight simulator not only wants to carry out the various sections object synthetic flight trainings such as traffic patter, large course line, stunt, but also can simulate the flight situation under the special feelings of the major part of aircraft and different flying condition, for the ease of fly simulation training and teaching, be also provided with the functions such as memory, reproduction.Flight simulator generally comprises the part compositions such as aircraft analog capsule cabin, computer generated image system, virtual image display system, Three-degree-of-freedom motion platform system, sound system, control desk and computer network.
In order to reach real effect, network service is vital.If system is used a traditional Ethernet (Ethernet) network connection system each module, be thisly connected with some shortcomings.For example, in order to reach real-time demonstration and real-time response, there are a large amount of data to need to calculate, need a large amount of data of high-speed transfer, also very harsh to the requirement of data delay.And traditional network cannot meet the requirement of real-time.
Traditional network technology, as Ethernet, there is the shortcoming of following several respects in FDDI etc. in application in real time:
1) uncertainty: determinacy for great majority based on OSI(open system interconnection (OSI)) seven layer protocol model compatible networks can not accomplish;
2) postpone unpredictable: for real-time Data Transmission, the output of a real-time system is likely as the input of another real-time application system, the time of data input has great timeliness meaning in this case, but conventional network techniques is seldom paid close attention to transmission delay minimization problem;
3) unreliable: a lot of real-time simulation test macros need absolute data integrity, can't stand data error.If wrong an appearance, it must be corrected immediately, and conventional network techniques cannot ensure such high reliability;
4) do not support cross-platform: real-time simulation test macro is often made up of a large amount of incompatible disposal systems, has plenty of industrial PC, has plenty of workstation, the use embedded system again having.Therefore, network communication module need to possess the ability that the system of different platform is supported, i.e. cross-platform ability, and conventional network techniques does not possess this ability.
Utility model content
In order to overcome above-mentioned shortcoming of the prior art, the utility model provides a kind of fiber reflection RAM (random access memory) card and fiber reflection internal memory net.
The utility model solves the technical scheme that its technical matters adopts:
On the one hand, the utility model provides a kind of fiber reflection RAM (random access memory) card, comprise pci interface control module, main control module, memory module, the serial/module of unstringing and optical-electric module, described main control module is for receiving the high-speed parallel data that the serial/module of unstringing is passed back, also for memory module provides high-speed data and control interface, and meet the system bus interface agreement under PCI structure; Described memory module is stored in its corresponding address for the data that receive by local data with from network according to different node numbers; Described serial/the module of unstringing, for realizing the serialization of multichannel data and the serial data of reception being unstringed; Described optical-electric module is used for realizing photoelectricity, electric light conversion, convert light signal to for high speed serialization electric signal that the serial/module of unstringing is sent here, then the light signal of simultaneously external fiber network being sent into is reduced into high speed serialization electric signal sends into the solution serializer circuit of the serial/module of unstringing; Described pci interface control module is for communicating by pci bus and outside computing machine, and the instruction from computing machine is sent to main control module.
According to above-mentioned fiber reflection RAM (random access memory) card, the described serial/module of unstringing comprises sendaisle and receiving cable, wherein scrambler, serializer, transmitter and clock generation circuit composition sendaisle; Demoder, deserializer, receiver and clock recovery circuitry composition receiving cable.
According to above-mentioned fiber reflection RAM (random access memory) card, described fiber reflection RAM (random access memory) card adopts 10 layer circuit board structures, and described 10 layers is respectively top layer, 1-the 8th middle layer and bottom, and wherein the 1st, the 5th and the 8th middle layer is stratum, the 4th middle layer is bus plane, and other layer is all signals layer.
According to above-mentioned fiber reflection RAM (random access memory) card, described memory module adopts DDR SDRAM as storage chip, and wherein the signal of DDR can be divided into control signal group, clock group, address bus group and data bus group according to its transferring content similarity; Wherein the spacing of same group is less, not on the same group keep larger spacing, and different groups is at different PCB layer cablings.
According to above-mentioned fiber reflection RAM (random access memory) card, the spacing between data signal line and address and control line has at least 4 times of live widths.
According to above-mentioned fiber reflection RAM (random access memory) card, described DDR is divided into multiple synchronization group, in the time connecting up, synchronous signal wire cloth is become to identical length, needs to carry out isometric line design.
According to above-mentioned fiber reflection RAM (random access memory) card, the described serial/module of unstringing has transmitting parallel data line and receives parallel data line, and these two kinds of data lines are dispersed in different layers and connect up and have larger spacing.
According to above-mentioned fiber reflection RAM (random access memory) card, the signal wire of same type has identical length.
According to above-mentioned fiber reflection RAM (random access memory) card, the HSSI High-Speed Serial Interface of the described serial/module of unstringing and optical-electric module, the clock of memory module and high speed distributed bus all use difference transmission lines, in the time of difference cabling, two differential signal correspondences are as far as possible close, and two signal wires of differential pair must keep isometric.
On the other hand, the utility model provides a kind of fiber reflection internal memory net, comprise many computing machines that adopt annular or star fashion to connect, fiber reflection RAM (random access memory) card described in any one above is wherein all installed in every computing machine being connected, each RAM (random access memory) card connects by optical fiber, and each RAM (random access memory) card logically shares a sector address.
Beneficial effect
The difference of Reflective memory network and Ethernet and FDDI is first in the realization of data communication, Ethernet and FDDI adopt the method for software to realize the data communication between each adjacent layer in osi model, whole process will take the CPU time of machine, and packet is generally larger, therefore the communication mode of Ethernet and FDDI is conducive to improve bandwidth and efficiency, but the little and sudden strong real-time Data Transmission of the amount of being unfavorable for; And Reflective memory network adopts hardware approach to realize data communication, communication does not take CPU time completely, and its packet is very little, is conducive to paroxysmal data transmission, can meet real-time application.The determinacy of Reflective memory network is with predictability and support that look-at-me transmission is the major advantage that Reflective memory network is compared with FDDI with Ethernet.The transmission mechanism of the CSMA/CD (csma/collision detection) of Ethernet has determined the determinacy that it is difficult to ensure transmission delay, FDDI Internet Transmission also has determinacy, but it does not support to interrupt transmission, " look-at-me is very important function in real-time system, can be used for the synchronous and event response in system to have limited its application in real-time system.In addition, Reflective memory network is compared with Ethernet has more reliable perfect hard error diagnosis and treatment mechanism " comprehensive above reason; Reflective memory network is one real-time network fast; it can be used for connecting various computing machines; can effectively solve the real time problems of data transmission in real-time system, and field is therefore widely used.
Brief description of the drawings
Below in conjunction with drawings and Examples, the utility model is further illustrated.
Fig. 1 is circuit working schematic diagram of the present utility model.
Fig. 2 is pcb board layer distribution plan of the present utility model.
Embodiment
In flight Simulation System, in order to shorten the response time delay of whole system as far as possible, avoid the unpredictability of network service, improve the fidelity of flight Simulation System, have now and in many flight Simulation System, all adopted advanced real-time network technology.The real-time system of being got up by real-time network technique construction is a kind of computer system that can carry out calculating or processing capacity and outside asynchronous event is made to response within definite time.The most important feature of real-time network is exactly determinacy and the predictability of its communication, and in real-time network, between each node, the time of data transmission is determined in other words, predictable.Network internal storage net be exactly a kind of at a high speed. real-time, deterministic real-time network, can solve preferably the real time problems of data transmission in flight simulation real-time system, improve transfer rate and the error correcting capability of network data.
Reflective memory network is a kind of real-time network of high speed, and its allows to adopt the computing machine of different bus structure and different operating system to share real-time data with the speed of determining.
Fiber reflection internal memory net is actually all installs a fiber reflection Memory communication board in every computing machine being connected, and each reflective memory cartoon is crossed optical fiber and connected.These cards form a continuous network structure, between reflective memory card, communicate by self-defining procotol, and each reflective memory is stuck in and shares in logic a sector address.In reflective memory network, reflective memory is distributed in (in reflective memory card) in each computing machine physically, shares in logic same section of memory address.Any computing machine can be accessed easily shared reflective memory as access common memory.Because reflective memory network has adopted the procotol of simplifying, so there is very high transmission speed.For the computing machine in reflective memory network, do not use the transmission problem of considering between reflective memory card, thereby greatly alleviated the work load of computing machine.Once any computing machine in reflective memory network is write data to reflective memory, the data of rewriting, at once with very high speed, are transferred to the shared drive that copies of other network node by high speed fibre network.That is to say, on the transmitting physical of data, exclusively enjoy high-speed dedicated network completely.Whole data Replica process does not have bus collision, there is no protocol overhead consuming time, does not have established data to destroy yet.Communication speed between each processor is the same with the speed of access memory fast, thereby has ensured the real-time of whole system.
Reflective memory plate uses simple read-write mode, the online data transmission of reflective memory is pure hardware operation, do not need to consider the communication protocol of network, on software, only need several line codes just can complete reading reflective memory plate, write operation, therefore with respect in the past based on TCP/IP or UDP/IP real-time network technology, it has improved system real-time Communication for Power greatly, the ability of data sharing and signal debugging, simplify the complicacy of flight simulator requirement of real time design, for the data communication of asynchronous system, share and synchronously provide easy, practical method, make the design of distributing emulation system structure and realize more simple.
Structure below in conjunction with accompanying drawing to the fiber reflection RAM (random access memory) card adopting in the utility model and PCB layout and high-speed wire-layout situation are elaborated.
Fiber reflection RAM (random access memory) card is mainly to adopt high performance programmable logic device (PLD) (for example FPGA) as overall control device, completes the functions such as the logic control to peripheral components, peripheral high speed interface, internal data current control.With reference to figure 1, fiber reflection RAM (random access memory) card mainly comprises pci interface control, main control module (for example FPGA), memory module, the serial/module of unstringing, optical-electric module.Described fiber reflection RAM (random access memory) card can also comprise and the corresponding configuration circuit of main control module.
Below the modules of composition RAM (random access memory) card is elaborated:
Pci bus technology is widely used at present, and design is got up very convenient, mainly contains two schemes and can realize pci bus interface, special pci interface chip design proposal and Design for Programmable Logic scheme.While adopting the design of special pci interface chip, deviser does not need to spend a large amount of energy and goes to understand the detailed operation principle of pci bus, only energy need to be concentrated on to the design that just can complete easily pci interface circuit on system.This scheme has the features such as simplicity of design, construction cycle be short.Design for Programmable Logic scheme is to adopt the programmable logic chip that meets pci bus specification to realize pci bus interface.Because current nearly all programming device factory commercial city provides the pci interface functional module of strict test, user only need carry out composite design and can realize to corresponding IP kernel the design of pci interface.Adopt programmable logic device (PLD) scheme great advantage to be that it is flexible, in general, typical PCI design does not really want to realize PCI all functions, only need select the PCI function needing when design.Therefore this design proposal has very strong dirigibility, is also convenient to function upgrading simultaneously.
By the comparison to two kinds of pci bus interface implementations above, can find out that the major advantage of the design proposal that adopts programmable logic device (PLD) is that the dirigibility of design is strong, therefore preferably realize the interface of pci bus and FPGA with IP kernel.Certainly it will be apparent to those skilled in the art that special pci interface chip design proposal and this two schemes of Design for Programmable Logic scheme can be used for the utility model.
In reflective memory card, main control module plays a part the most key, it not only needs to receive the high-speed parallel data that the serial/module of unstringing (SerDes chip) is passed back, also need to provide the required high-speed data of memory module and control interface, and meeting the system bus interface agreement under PCI structure, it is the hinge of whole communication board.Main control module preferably uses FPGA, more preferably gets the EP1S20 series of ALTERA company as main control chip, and this chip price is moderate, domestic stock in hand, and application is comparatively ripe, has referential technical information, meets system user demand.Certainly main control module also can be selected the common control chip of other kind, the various programming devices such as such as PAL, GAL, CPLD.
Memory module is reflective memory namely, and local data and the data that receive from network can be stored in its corresponding address by different node numbers.Storer is the main body of this module, so the selection of storer is even more important for the performance of this module.The utility model preferably adopts on the market the DDR SDRAM of maturation as storage chip, for example, can consider to utilize the chip of 4 32MB capacity to realize the memory requirement of 128MB.
Serial/the module of unstringing is SerDes module, for realizing the serialization of multichannel data and the serial data of reception being unstringed.Mainly formed by sendaisle and receiving cable: scrambler, serializer, transmitter and clock generation circuit form sendaisle; Demoder, deserializer, receiver and clock recovery circuitry composition receiving cable.For the requirement of transfer rate 2.125Gbps, preferably TLK2501 is as the serial/module of unstringing.
Optical-electric module is to realize photoelectricity, electric light conversion, has independently transmitting and drives and receive amplifying circuit, transmission-receiving function unification.Convert light signal to for high speed serialization electric signal that SERDES serial circuit is sent here, the light signal of simultaneously external fiber network being sent into is reduced into high speed serialization electric signal sends into the solution serializer circuit of SERDES.For the requirement of transmission range >=10km, transmission interface LC duplex form, the FTLF1319 of the preferred Finisar of the utility model company is as optical-electric module, its operation wavelength 1310nm, transmission range is greater than 10km, transfer rate 2.125Gbps, LC duplex, miniaturization SFF encapsulation.
Those skilled in the art obviously can design suitable storage chip, the serial/module of unstringing and the optical-electric module that need to select other kind.
Fiber reflection RAM (random access memory) card has comprised numerous high speed devices and high speed interface, therefore, in concrete PCB realization, will propose higher requirement, to guarantee that on circuit board, each device can normally move and not influence each other, each signalling channel can at a high speed and stably be transmitted data.Xia Ming is elaborated to the correlation circumstance of PCB layout and high-speed wire-layout.
Comprehensive device characteristics, wiring complexity and other factors, fiber reflection RAM (random access memory) card adopts 10 layer circuit board structures.It forms as shown in Figure 2.On board layer distributes, adopt three layers as stratum, each signals layer is separated.Adopt this structure to be because a lot of signal wire transmission is all high speed signal on this transmission board, as hierarchy adopting, adjacent between signals layer, so in the time of signal high-speed transfer, adjacent signals line between layers affects quite greatly, is far longer than the parallel adjacent traces of same layer.In addition,, owing to there is no reference planes, be difficult to the suitable matched impedance of design.And structure in employing Fig. 2, each signal is separated by stratum, has avoided crosstalking and radiation effect between adjacent layer, and impedance comparison is stable, is suitable for the transmission of HW High Way.Another feature of this structure is that stratum and bus plane are tightly adjacent, makes source current form a return flow path the shortest, has reduced the impact of electromagnetic radiation, and, due to the shortening of distance between bus plane and stratum, can produce the effect of capacitor, make signal environment more stable.
Before wiring, need to complete the layout of each device on circuit board.Whether the layout of system is reasonable, directly had influence on the complexity of wiring, heat radiation and EMI characteristic and the final performance of circuit board of each device.
In general should note following some:
(1) system layout should ensure the reasonable or optimum of wiring, can ensure reliably carrying out of wiring, need to have overall understanding and planning to the trend of signal and VDD-to-VSS spider lines.
(2) can whether printed board size conforms to demand size, meet PCB manufacture process requirement.Element has or not conflict on two dimension, three dimensions.Note the height of the physical size, particularly device of device.On PCI board, the height of device generally can not exceed 2cm.The circuit layout of many pcb boards all designs very beautifully, reasonable with wiring, but has neglected the accurate location of location connector, causes the circuit of design to dock with other circuit.
(3) component placement whether density in order, marshalling, whether all cloth is complete.In components and parts layout, not only to consider the trend of signal and the type of signal, the place that should be noted that or protect, also to consider the global density of device layout simultaneously, accomplish that density is even.
(4) can need the element of often replacing change easily, whether plug-in card interventional instrument be convenient.Should ensure the replacing of the components and parts of often changing and the convenience patching and reliable, whether conveniently adjust adjustable element.
(5) between thermal sensing element and heater element, whether there is suitable distance.Whether the place in needs heat radiation is equipped with heating radiator or fan, and whether air stream is unobstructed, should note the heat radiation of components and parts and circuit board.
(6) whether signal trend is smooth and easy and interconnection is the shortest, whether contradiction of plug, socket etc. and Machine Design.Whether the interference problem of circuit and the physical strength of circuit board and performance are considered to some extent.
(7), under the prerequisite that meets above-mentioned condition, should consider artistry and the aesthetic property thereof of circuit-board laying-out.
In the utility model, 70% signal wire is high-speed line, therefore, in the time of wires design, must consider feature and some restrictions of HW High Way, to improve signal transmission quality.In addition, due to DDR internal memory adopt be the single-ended level form of SSTL_2, and design clock frequency exceeded 100MHz, single line data transfer rate has exceeded 200Mbps.In addition, TLK2501 adopts the level form of LVTTL, and design frequency has also exceeded 100MHz.Therefore, for these high speed single ended line, in the time of wiring, should adopt some constraint measures, just can make signal stably transmit.
(1) DDR wiring constraint
DDR adopts double data rate (DDR) form, its data transfer rate is the twice of its transfer clock frequency, under two-forty, to ensure the stable transfer of signal, avoid signal reflex to improve transmission quality except adding terminal resistance, also must propose special requirement by the cabling on circuit board to DDR, need the wiring to DDR to retrain.In general, the content of constraint comprises live width, spacing and length of arrangement wire.
In general, DDR wiring does not have specific requirement to live width, but in the situation that speed is very high, need to consider the resistance matching problem of transmission line, and this just and wiring width generation relation.The utility model connects up at the general width that adopts 7mil of DDR part.Distance between two signals has larger impact to signal integrity, and DDR has also done stricter regulation in this respect.The signal of DDR can be divided into several groups according to its transferring content similarity, comprises control signal group (RAS#, CAS#, WR#, CS#, CKE etc.), clock group (CK, CK#), address bus group (A0-A13) and data bus group (DQ, DQS, DM).
In general, the spacing of same group can be less, not on the same group should keep larger spacing, particularly data signal line and address and control line should have enough spacing (generally at least 4 times of live widths), preferably can be at different PCB layer cablings, with the cross talk effects of avoiding the number of it is believed that of control signal demand pairs to be caused.Except signal live width and spacing, can the difference in length between signal wire receive to produce to signal and compare considerable influence.Therefore, when wiring, should synchronous data line cloth be become to identical length as far as possible, need to carry out isometric line design.DDR has generally been divided into 8 synchronization group, formed by 8 DQ and DQS, DM, and the read error causing in order to eliminate sequential deflection, 10 signal wires of same category answer cloth to become isometric.Be controlled at ± 0.5mm of error of the present utility model, though address and control line do not have strict isometric requirement, the cloth of also should trying one's best becomes isometric line, and should maintain a certain distance relation with clock signal input.
On pcb board, conventionally adopt the mode of cloth serpentine to realize isometric wiring.Serpentine is a kind of cabling mode often using in PCB layout, and its fundamental purpose is exactly in order to regulate time delay, meets system sequence designing requirement.Although serpentine can produce inductive effect, destroy signal quality, in actual design, in order to ensure that signal has enough retention times, or reduce the time migration between signal on the same group, often have to deliberately wind the line.
(2) TLK2501 wiring constraint
TLK2501 parallel data grabbing card adopt LVTTL level form, maximum clock frequency is 125MHz, employing be single data rate transmission form.Although TLK2501 parallel interface does not have the message transmission rate of ddr interface high, it still belongs to high speed signal, also needs its wiring constraint.
TLK2501 has transmitting parallel data line and receives parallel data line, if the signal wire that these two kinds of directions are different is separated by nearer on circuit board, will produce very large crosstalking, therefore should increase the spacing of two kinds of lines as far as possible, and they are dispersed in to different layers as far as possible and connect up.The signal wire of the same type of same chip also needs and corresponding clock synchronous, therefore also should realize length constraint, and the margin of error that the utility model adopts is herein ± 1mm.
(3) difference transmission lines design
In the utility model, the HSSI High-Speed Serial Interface of SerDes chip and optical transceiver module, the clock of DDR and high speed distributed bus all need to use difference transmission lines, differential signal is compared with common single-ended signal cabling, has the advantages that antijamming capability by force, effectively suppresses EMI and sequential registration.When difference cabling, two differential signal correspondences are as far as possible close, can strengthen like this their coupling, both can improve the immunity to noise, and the opposite polarity that can also make full use of magnetic field is offset electromagnetic interference (EMI) to external world.In general, differential signal line also needs the backflow of a reference horizontal plane of manufacturing as high-frequency signal.The most important thing is, two signal wires of differential pair must strictly keep isometric, and in the utility model, the length difference general control between differential pair is at ± 0.3mm.
Although exemplary embodiment of the present utility model is described in detail hereinbefore; those skilled in the art should understand at that time; above embodiment is described the preferred embodiment of the present invention; not scope of the present invention is limited; design under the prerequisite of spirit not departing from the present invention; various distortion and improvement that those of ordinary skill in the art make technical scheme of the present invention, all should fall in the definite protection domain of claims of the present invention.

Claims (10)

1. a fiber reflection RAM (random access memory) card, comprises pci interface control module, main control module, memory module, the serial/module of unstringing and optical-electric module, it is characterized in that:
Described main control module, for receiving the high-speed parallel data that the serial/module of unstringing is passed back, also for memory module provides high-speed data and control interface, and meets the system bus interface agreement under PCI structure;
Described memory module is stored in its corresponding address for the data that receive by local data with from network according to different node numbers;
Described serial/the module of unstringing, for realizing the serialization of multichannel data and the serial data of reception being unstringed;
Described optical-electric module is used for realizing photoelectricity, electric light conversion, convert light signal to for high speed serialization electric signal that the serial/module of unstringing is sent here, then the light signal of simultaneously external fiber network being sent into is reduced into high speed serialization electric signal sends into the solution serializer circuit of the serial/module of unstringing;
Described pci interface control module is for communicating by pci bus and outside computing machine, and the instruction from computing machine is sent to main control module.
2. fiber reflection RAM (random access memory) card as claimed in claim 1, is characterized in that: the described serial/module of unstringing comprises sendaisle and receiving cable, wherein scrambler, serializer, transmitter and clock generation circuit composition sendaisle; Demoder, deserializer, receiver and clock recovery circuitry composition receiving cable.
3. fiber reflection RAM (random access memory) card as claimed in claim 1 or 2, it is characterized in that: described fiber reflection RAM (random access memory) card adopts 10 layer circuit board structures, described 10 layers is respectively top layer, 1-the 8th middle layer and bottom, wherein the 1st, the 5th and the 8th middle layer is stratum, the 4th middle layer is bus plane, and other layer is all signals layer.
4. fiber reflection RAM (random access memory) card as claimed in claim 1 or 2, it is characterized in that: described memory module adopts DDR SDRAM as storage chip, and wherein the signal of DDR can be divided into control signal group, clock group, address bus group and data bus group according to its transferring content similarity; Wherein the spacing of same group is less, not on the same group keep larger spacing, and different groups is at different PCB layer cablings.
5. fiber reflection RAM (random access memory) card as claimed in claim 4, is characterized in that: the spacing between data signal line and address and control line has at least 4 times of live widths.
6. fiber reflection RAM (random access memory) card as claimed in claim 4, is characterized in that: described DDR is divided into multiple synchronization group, in the time connecting up, synchronous signal wire cloth is become to identical length, needs to carry out isometric line design.
7. fiber reflection RAM (random access memory) card as claimed in claim 1 or 2, is characterized in that: the described serial/module of unstringing has transmitting parallel data line and receives parallel data line, and these two kinds of data lines are dispersed in different layers and connect up and have larger spacing.
8. fiber reflection RAM (random access memory) card according to claim 7, is characterized in that: the signal wire of same type has identical length.
9. fiber reflection RAM (random access memory) card as claimed in claim 1 or 2, it is characterized in that: the HSSI High-Speed Serial Interface of the described serial/module of unstringing and optical-electric module, the clock of memory module and high speed distributed bus all use difference transmission lines, in the time of difference cabling, two differential signal correspondences are as far as possible close, and two signal wires of differential pair must keep isometric.
10. a fiber reflection internal memory net, comprise many computing machines that adopt annular or star fashion to connect, it is characterized in that: the fiber reflection RAM (random access memory) card as described in any one in claim 1-9 is all installed in every computing machine being connected, each RAM (random access memory) card connects by optical fiber, and each RAM (random access memory) card logically shares a sector address.
CN201420009177.3U 2014-01-07 2014-01-07 Optical fiber reflection memory card and optical fiber reflection memory network Expired - Lifetime CN203733105U (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201420009177.3U CN203733105U (en) 2014-01-07 2014-01-07 Optical fiber reflection memory card and optical fiber reflection memory network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201420009177.3U CN203733105U (en) 2014-01-07 2014-01-07 Optical fiber reflection memory card and optical fiber reflection memory network

Publications (1)

Publication Number Publication Date
CN203733105U true CN203733105U (en) 2014-07-23

Family

ID=51203163

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201420009177.3U Expired - Lifetime CN203733105U (en) 2014-01-07 2014-01-07 Optical fiber reflection memory card and optical fiber reflection memory network

Country Status (1)

Country Link
CN (1) CN203733105U (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104615483A (en) * 2015-02-12 2015-05-13 哈尔滨工业大学 Reflective memory card and access method for reflective memory net and reflective memory card
CN109890044A (en) * 2019-03-15 2019-06-14 上海微小卫星工程中心 Complicated Satellite system simulation real-time network communication system

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104615483A (en) * 2015-02-12 2015-05-13 哈尔滨工业大学 Reflective memory card and access method for reflective memory net and reflective memory card
CN109890044A (en) * 2019-03-15 2019-06-14 上海微小卫星工程中心 Complicated Satellite system simulation real-time network communication system

Similar Documents

Publication Publication Date Title
CN103761137A (en) Optical fiber reflection internal memory card and optical fiber reflection internal memory network
US9607673B1 (en) Methods and systems for pin-efficient memory controller interface using vector signaling codes for chip-to-chip communication
CN102799509B (en) High-bandwidth extendable complex logic verification system based on double FPGA (Field- Programmable Gate Array) chips
CN105612580B (en) Use the mass-storage system of standard controller component
CN105051706B (en) The equipment of operation for the low-power PHY with PCIE protocol stacks, method and system
US20100005214A1 (en) Enhancing bus efficiency in a memory system
US20100005218A1 (en) Enhanced cascade interconnected memory system
US7593288B2 (en) System for providing read clock sharing between memory devices
US11907140B2 (en) Serial interface for semiconductor package
CN105531766A (en) Load reduced memory module
CN106095334B (en) A kind of high-speed data acquisition storage system based on FPGA
US9672182B2 (en) High-speed serial ring
CN103605632A (en) Method and device for communication between AXI (advanced extensible interface) bus and AHB (advanced high-performance bus)
KR20150069006A (en) Low swing voltage mode driver
CN209489030U (en) Master control cabinet and power electronic control system
CN104991880B (en) A kind of FC AE ASM Communication Cards based on PCI E interfaces
CN107943733A (en) The interconnected method of parallel bus between a kind of veneer
CN203733105U (en) Optical fiber reflection memory card and optical fiber reflection memory network
Gopal et al. High-performance and small-form factor near-field inductive coupling for 3-D NoC
US20180275714A1 (en) Inductive coupling for data communication in a double data rate memory system
US20120191943A1 (en) Dynamic protocol for communicating command and address information
Mutnury et al. Analysis of fully buffered DIMM interface in high-speed server applications
CN207339846U (en) A kind of independent double-channel data Transmission system based on SRIO
CN103645999A (en) FPGA-based high-speed board card for achieving eight-channel receiving and transmitting serial ports
CN203673473U (en) High-speed board card for eight channel transceiving serial ports on basis of FPGA

Legal Events

Date Code Title Description
C14 Grant of patent or utility model
GR01 Patent grant
CX01 Expiry of patent term
CX01 Expiry of patent term

Granted publication date: 20140723