Internet Explorer는 TI.com에서 지원되지 않는 브라우저입니다. 최상의 환경을 위해 다른 브라우저를 사용하십시오.

비디오 시리즈

Precision Labs 시리즈: FPD-Link

주로 자동차에서 디지털 오디오/비디오를 비디오 화면(인포테인먼트)으로 전송하거나 운전자 지원 카메라(ADAS)에 사용되는 FPD-Link 시리얼라이저 및 디시리얼라이저(SerDes)를 사용하여 설계하는 방법을 알아보세요.

Video Player is loading.
Current Time 0:00
Duration 0:00
Loaded: 0%
Stream Type LIVE
Remaining Time 0:00
 
1x
  • Chapters
  • descriptions off, selected
  • subtitles off, selected

      발표자

      Hello. I'm Casey McCrea. In this training module, I will introduce you to FPD-Link, a core technology in automotive and industrial video applications worldwide. FPD-Link is a multi-protocol physical layer technology that aggregates data from various industry standard protocols and transfers it over cables.

      Its most common use is for digital audio and video transfer in automotive applications, such as video screens or driver assistance cameras. But FPD-Link dealing can also be utilized for other types of data transfers, such as lidar, radar and more. Some examples are shown in these vehicle diagrams. And for further explanation of the acronyms here, see the appendix slide, which is downloadable on the video page. In this video, I will discuss the fundamentals of FPD-Link technology, including parallel data interfaces, concepts of serialization and deserialization, the high speed serial channel itself, and bidirectional communication.

      FPD-Link technology acts as a bridge between protocol-based data interfaces, which require multiple signaling conductors to transfer high bandwidth data. Some common examples of these interfaces include TMDS, MIPI DFI, LVDS, or parallel LVCMOS. These standards all define different physical layers for transferring audio, video, or other digital video sources.

      However, generally, these standards are designed for transferring video only over short distances, which may include PCB trace or flex cable. In many cases, automotive or industrial applications demand that this high bandwidth data be transferred over much longer distances of 10 or more meters using low cost and low weight coaxial or shielded-twisted pair cabling.

      This then presents a challenge, not just because of the signal loss introduced with such a transmission channel, but due to the introduction of EMI and EMC considerations, which are common to automotive and industrial applications, where external noise sources can interfere with the cable as it passes to its destination. Because FPD-Link can interface to many different parallel data formats, this also means the technology may be used as a format converter, where the source interface may not necessarily match the sync interface.

      FPD-Link applications typically utilize either coaxial or twisted pair cables to carry the information between serializer and deserializer. Coaxial cables tend to have lower insertion loss characteristics when compared with twisted pair cables due to their electromagnetic construction, and they also typically come at lower cost when compared with other cable types.

      Twisted pair cables, on the other hand, benefit from the fact that signals are carried differentially across them, meaning that they are typically more immune to the effects of single-ended electromagnetic interference. Twisted pair cables generally come in three main variants-- STP, which stands for Shielded-Twisted Pair; STQ, which stands for Star-Quad; or UTP, which stands for Unshielded-Twisted Pair.

      Shielded-twisted pair cables provide even stronger immunity against external noise interference by surrounding the twisted pair wires with a thin conductor. Due to the harsh electromagnetic environment of automotive applications, most systems utilize STP or STQ cable, rather than UTP.

      Coaxial cables are often also utilized for transferring DC power to a remote device along the same conductor as the high speed data, since the outer shield can act like a return path for DC current, and the AC content can be filtered from the DC power, which is also on the line. This scheme is commonly referred to as Power over Coax or PoC. Most FPD-Link III devices can support both coax or twisted pair configurations, to ensure flexibility in various applications.

      The FPD-Link high speed forward channel is used to send the serialized video, audio, or other data to an endpoint device with minimal latency. In order to achieve this, the serializer must reformat its incoming data and embed the data clock so that it can be output using less conductors. By utilizing a proprietary echo cancelation technique, FPD-Link SerDes also allow for full duplex communication over one physical conductor.

      As high speed data transfers from the serializer to the deserializer in the forward direction, low speed data is also transferred back to the serializer simultaneously and without time multiplexing. The FPD-Link serializer and deserializer devices automatically establish this bidirectional channel by canceling out their own transmitting signals at each end of the link continuously.

      The back channel is typically operated at a significantly lower speed than the forward-channel data in order to facilitate proper separation on both sides, and may include information about the sync device, touch interrupts, control signals, status information, and more. Using the simultaneous back-channel communication, I2C access or GPIO transfer can also be enabled across the link in either forward or reverse directions.

      In order to compensate for channel insertion loss-- which may be significant, depending on the speed of operation and type or length of cabling used-- FPD-Link deserializers utilize multiple equalization techniques in order to recover high frequency signal content and mitigate the effects of intersymbol interference, reflections, or external noise influence.

      Consider eye diagram shown on the left side. The signal eye is sufficiently open when measured at the output of the serialized r device. However, once that high speed data travels through a 10-meter STP cable, the effects of insertion loss and ISI have significantly reduced the eye quality, to the point where an unequalized receiver could not properly recover that data.

      After passing through the adaptive equalizer on the deserializer, you can see that the eye has regained both vertical height and horizontal width. First, the FPD-Link deserializers utilize a Continuous Time Linear Equalizer, or CTLE circuit, which acts like an amplifier to high frequency signal content that is attenuated more rapidly across a lossy channel.

      The CTLE has several different gain stages to account for varying channel loss, depending on the type and length of cabling used in the system. Next, the FPD-Link deserializers utilize a Clock Domain Recovery, or CDR circuit, to further improve the signal quality in the time domain by reducing the effects of jitter caused by ISI, reflections, or external noise. For more details on how to interpret eye diagrams like these, see the Precision Labs video, "What is an Eye Diagram?"

      FPD-Link transfer works by combining input data into packets or frames in order to be transferred serially at higher speed. Depending on the device family, the forward-channel frame size may vary from 28 to 40 bits, and uses 8b/10b style encoding. For an FPD-Link forward-channel signal, the signaling frequency in gigahertz is half of the effective baud rate in gigabits per second.

      For example, a 4.2 gigabits per second forward channel would utilize 2.1 gigahertz fundamental carrier frequency. The forward-channel frame can be broken down into four main categories. Payload data makes up a majority of the frame. This is the high bandwidth portion of the data which could be comprised of video pixel information, audio data, or other data types, including input from radar, lidar, or more.

      There are two embedded clocking bits per frame, which help establish timing for the high speed signal and establish frame boundaries. DCA and DCB bits are used for both DC balancing the signal and also for conveying encoding information across multiple consecutive frames to establish the link. DC balancing helps to ensure that the AC-coupled channel will not experienced the effects of baseband wander caused by excessive 1 or 0 symbols in the data. And finally, control bits are used to convey side band information, like I2C, GPIO, SPI, or CRC.

      FPD-Link III devices utilize a continuous back-channel communication scheme, which allows full duplex communication on the same conductor or conductor pair. The back channel contains I2C, GPIO, status, or other information to be carried across the link to the remote partner. One main difference between the forward channel and back-channel communication is that the back channel is Manchester encoded.

      This transition-based encoding scheme allows lower data rate symbols to be passed through the AC-coupled link without the need for separate clocking channels. From an analog signaling perspective, it is important to note that for Manchester-encoded signals, the symbol rate is actually equal to the data rate. For example, if the back-channel rate is set for 10 megabits per second, the back-channel frequency will also equal 10 megahertz. These frequency spectrums are important to understand for the purposes of EMC optimization, as well as for power filtering or things like power over coax schemes.

      All right, let's take a little quiz and test your FPD-Link knowledge. For an FPD-Link forward-channel frequency of 3.36 gigabits per second per lane, what is the fundamental carrier frequency for that signal per lane? Is it a, 3.36 gigahertz; b, 96 megahertz; c, 1.68 gigahertz; or d, 6.72 gigahertz?

      The answer is c, 1.68 gigahertz. And remember why. The FPD-Link forward channel uses NRZ signaling, which results in two bits per clock period. Any time we express an FPD-Link forward-channel frequency in gigabits per second, the carrier frequency for that signal will be one half that rate in gigahertz.

      All right, number two. For an FPD-Link III back-channel rate of 20 megabits per second, what is the fundamental carrier frequency for this signal-- 10 megahertz, 20 megahertz, 5 megahertz, or 40 megahertz? So remember, this is a bit of a trick question. The answer is actually 20 megahertz. And why is this? Because the back channel for FPD-Link III is Manchester encoded, unlike the forward channel, which means that the baud rate in megabits per second will equal the fundamental carrier frequency in megahertz.

      And finally, a true or false. FPD-Link III is a bidirectional half-duplex protocol. The answer is false. And why is this? FPD-Link III is a bidirectional protocol, yes. But remember that it is actually a full duplex protocol, because data can be transferred in both forward and reverse directions at the same time without the need for time multiplexing. To wrap things up, you can access additional technical resources and find FPD-Link products to fit your application by following the links provided here to ti.com. Thank you very much for taking the time to watch this video, and see you next time.

      View series

      Precision Labs 시리즈: FPD-Link