This text is the second in a series of articles on the tasks of measurement technology in 100G networks.He deals with the physical layer measurement technology at 100G.The first part (LANline 6/2018, page 23) dealt with the development towards 100G and the associated general challenges for measurement technology.Further contributions will deal with transmission measurements at layers 2, 3 and 4 and with troubleshooting in 40G/100G transmission.100G is not equal to 100G.Because 100G on the long distance and line side uses different transmission methods: as OTN (Optical Transport Network), DWDM (Dense Wavelength Division Multiplex) and as native Ethernet.Understandably, each of these techniques requires different measurements at the physical level.Two transmission technologies compete with each other in the wide area network: the conventional multi-wavelength DW-DM technology (n × 10 GBit/s, n × 40 GBit/s) and the single-wavelength full-rate 40G/100G technology.On the client side, multi-? signals (similar to CWDM) carry the 100G signal in four sub-streams.Proprietary 100G procedures have emerged for the long-distance network side.With manufacturer-specific modulation and signal processing, these only require one wavelength for transmission.Currently, however, 100G transmission is already happening coherently on a broad front with just one wavelength, even in the metro environment.A method called "Polarization Multiplex-QPSK" is used.It is nothing new to transmit the 100G signal on ten or four wavelengths (10 × 10 GBit/s or 4 × 25 GBit/s) via DWDM in long-distance traffic.With regard to PMD (polarization mode dispersion), the DWDM technology with n × 10 Gbit/s places higher demands on a glass fiber than the coherent single-wavelength transmission method PM-QPSK.Via FFT (Fast Fourier Transformation) and with fast signal processors, it is possible for a coherent 100G transponder to calculate and compensate for the PMD distortions.This is not possible with 10G incoherent transmission (DWDM).100G Ethernet on the client side is often transmitted using CFP2 or QSFP28 (LR4 version).This is done with a fiber on four wavelengths, each transmitting 25 GBit/s.Today's BERT devices (Bit Error Rate Tester) display the individual levels and the resulting total level at the transceiver output.The four wavelengths in the image below that lie in the second optical window do not correspond to the CWDM or DWDM grid, but start at a spacing of 2.0 GHz/4.5 nm, starting at around 1295 nm.The transceiver self-test integrated in 100G transmission testers gives users the opportunity to test transceiver properties and functions and also enables automatic level testing of individual lasers.The examination includes the sum level at the output, the behavior in the event of a clock offset, the skewing (runtime differences of the lanes) and the bit error behavior under high loads.All this with selectable BERT pattern (say 231-1, 223-1?).Figure 2 shows an example measurement result.The benefit of the test is to identify faulty transceivers without having to spend a long time troubleshooting the optical fiber or a supposedly incorrect system configuration.In LAN backbone applications and in data centers, 100G transmission often takes place via Ethernet on multimode.The 100G transceivers CFP2, CFP4 and QSFP28 use either multiple fibers (100GBase-SR10, 100GBase-SR4) or multiple wavelengths (100GBase-SWDM4).In addition to the multimode standards, there are also parallel fiber standards for singlemode, such as PSM4.According to the MLG standard (Multi-Link Gearbox), the systems then transmit 10 × 10 GBit/s over 4 × 25 GBit/s.This standard from the OIF committee (Optical Internetworking Forum) arose because 10-lane transmissions are hardly ever installed at 100G.The MLG process has mappings and control functions on suitable transceivers and the system host (system interface card).The ribbon fiber MPO/MTP connector has found widespread use with the SR4 and SR10 interfaces in addition to the duplex LC connector in the newly emerging data centers.In the connector, twelve or 24 (or 16 or more) fibers are physically connected in one plug-in process without an air gap and without dirt particles.It is therefore strongly recommended to clean and visually inspect the connector before plugging it in.One should not rely on the fact that with MPO multimode connections an RS-FEC (Reed-Solomon Forward Error Correction) corrects the errors that occur with little contamination and a small increase in attenuation.An operator should check the end face thoroughly, because the dirt can migrate so unfavorably that a fiber becomes blind and even RS-FEC is no longer of any help when it is plugged in again just once.To be on the safe side, installers should inspect MPO connectors with a fully automated MPO video microscope.Since the distances in the data center are manageable, an operator can wire with copper lines that are fully compliant with EN-50173.Although EN-50173 prescribes the laying of fiber optic cables for the primary and secondary levels, copper and fiber optics are permitted in the tertiary level, for example in a corridor or for horizontal cabling.If servers are next to each other in the rack or in the same room, they can be connected with 100G up to a length of seven meters (100GBase-CR10) with copper cables.Twinax cables, also known as DAC (Direct Attach Copper) cables, are used for this purpose.They are terminated on both sides with QSFP28 or QSFP+ and plugged directly into the switch ports.100GBase-CR10 and 100GBase-CR4 are the most well-known standards.If a connection fails, the DAC cable may be damaged.The test uses 100 Gbps transmission testers with dual port capability, avoiding unnecessary cable swaps.Meters should also have an Optical Self-Test menu that allows you to choose whether to test an AOC cable, a DAC cable, or an optical transceiver.Cable testers and cable certifiers from the 1G/10G world are advancing into higher and higher frequency regions (up to 2,500 MHz).Depending on the device type and accuracy class level V, they can also be used for category 7A and category 8 cabling.This enables them to assess whether twisted pair cabling is suitable for 40G or 100G.However, this only applies if the route to be tested has suitable connection technology and high-quality test and reference cable connections.New tasks for measurement technology: LinkAs is well known, the solutions for 100G on multimode fibers as well as for CWDM4 and PSM4 also for single mode use an RS-FEC error correction method just below layer 2 in the interface chipset on the physical media dependent sublayer.Its purpose is to ensure that the transmission has the desired maximum linear expansion.Because attenuation budgets are getting smaller and smaller with increasing bandwidth, and contamination of MPO connectors is not uncommon.Of course, the technician in the field is interested in knowing how many errors are corrected with RS-FEC.The image on the left shows how today's testers display the error rate before and after FEC.A neuralgic point in the transmission path in the 40G/100G environment are the transceivers and the AOC, DAC or MPO breakout cables used in the data center.Since these components are very expensive, unnecessary replacement should be avoided.For this reason, today not only transmission measurement functions, but also physical component tests are integrated into the transmission testers.Thomas Friedrich is a consultant and focal account manager at Viavi Solutions and part-time lecturer for communication technology at the Baden-Württemberg Cooperative State University, www.viavisolutions.com.Read the latest news from the world of LANline regularly in our LANline newsletterVisit an event from our high-quality event series: LANline Tech ForumDatacenter SymposiumLANline Workshops© 2022 WEKA FACHMEDIEN GmbH.All rights reserved.