G.652D has zero dispersion at 1310nm and is typically used for CWDM applications.

G.655D has a larger core and a small but non-zero amount of dispersion at 1550nm, making it less prone nonlinear effects (e.g., four wave mixing) and more suitable for 1550nm DWDM applications.

I should add that G652D (such as SMF28) is also used extensively in 1550nm DWDM and is less costly than the G655D. The latter is more suitable for long-distance submarine type applications.

S. Elahmadi thanks for your further clarification. The issue is that a simple google always instruct you that G.652D is for CWDM only while it is widely deployed for DWDM. My point is to know the reason of this ambiguity and correct it somehow.

Abdul Aziz Khan : - It is misleading to suggest that G.652.D "is for CWDM only". Did you mean that "only G.652.D should be used for CWDM"?

I would also question your statement that "Theoretically G.655 is much better than G.652". There are some applications where G.655 fibre has advantages, but G.652 performs well in a wide range of circumstances. In many cases, and particularly for high capacity DWDM links, G652 will offer superior performance.

Fibres similar to type G.652 were originally intended for single channel operation with zero dispersion wavelength close to 1300 nm. As manufacturing techniques improved, lower attenuation became available at longer wavelengths in the 1525 to 1625 nm range, where Rayleigh scatter is weaker. Operation in these longer wavelength C and L bands was encouraged by the introduction of erbium-doped fibre amplifiers (EDFAs), enabling transmission over multiple fibre spans without the need for electrical regeneration.

In contrast, CWDM is used without optical amplification, typically using directly modulated lasers over relatively short distances, and operating at wavelengths between 1270 nm and 1610 nm on a 20 nm grid.

Note that older types of fibre meeting G.652.B may have relatively high attenuation at wavelengths around 1240 nm and 1380 nm, caused by hydroxyl (OH) groups chemically bonded within the silica glass matrix. This makes them unsuitable for operation at the shorter CWDM wavelengths.

https://spie.org/documents/publications/00%20step%20module%2007.pdf (figure 7-10, p 263)

The more recent G.652.D specification requires that attenuation is no greater than 0.4 dB/km at any wavelength between 1310 nm and 1625 nm. Other parameters such as mode field diameter and chromatic dispersion are broadly similar, but in places more tightly specified than for G.652.B.

https://www.itu.int/rec/T-REC-G.652-201611-I/en

As modulation speeds increased over time, fibre chromatic dispersion penalties became more important. G.653 fibres were introduced to support single mode transmission with zero chromatic dispersion in the 1550 nm band. These enabled higher modulation rates without the need for chromatic dispersion compensation, providing lower transmission losses than 1310 nm systems and compatibility with EDFA operation.

https://www.fia-online.co.uk/pdf/Whites/wp0704.pdf

Unfortunately when dense wavelength division multiplexing in the erbium amplifier bandwidth is required, a zero dispersion wavelength close to the operating wavelengths gives rise to four wave mixing (FWM) and other non-linear interactions. G.652 fibre is much less susceptible to FWM, but for 2.5 GBaud and faster systems, the maximum transmission distance is limited by intra-channel chromatic dispersion.

G.655 fibre is designed with a dispersion zero wavelength outside the erbium C-band transmission window. The dispersion between 1530 nm and 1565 nm is between 1 ps/nm/km and 10 ps/nm/km for G.655.C fibre, compared with up to 18.6 ps/nm/km for G.652 fibre at 1550 nm.

The minimum cable cut-off wavelength is 1450 nm, so G.655 may not be suitable for shorter CWDM wavelengths.

https://www.itu.int/rec/T-REC-G.655-200911-I/en

Unfortunately, while 1 ps/nm/km is sufficient to suppress four wave mixing in a short 2.5 Gbit/s system with 200 GHz channel spacing, non-linear WDM crosstalk becomes problematic at narrower channel spacing and higher symbol rates. The FWM noise penalty varies as the square of the per-channel signal power and inversely as the 4th power of the channel spacing. For the same signal power, FWM is 16 times stronger if the channel spacing is halved to 100 GHz. At higher symbol rates, 10 GBaud and higher, Cross Phase Modulation (XPM) is often the dominant penalty, but acceptable performance is possible with higher fibre dispersion combined with dispersion compensation.

Fibre dispersion compensation modules add to system costs, and later iterations of the G.655 specification mandated higher values for the minimum dispersion:

• >2.8 ps/nm/km at 1550 nm for G.655.D
• >6.06 ps/nm/km at 1550 for G.655.E

The system designer must choose a fibre dispersion specification which is a compromise between:

• Lower chromatic dispersion to reduce in-band signal distortion and avoid or reduce the expense of dispersion modules
• Higher chromatic dispersion to reduce non-linear inter-channel crosstalk due to FWM, XPM and stimulated Raman scattering (SRS).

More recent high capacity systems operate at high symbol rates, 10-30 GBaud, employ multi-level QAM modulation, and coherent detection.

In contrast to older IMDD (intensity modulation direct, detection) systems, coherent detection can recover both the amplitude and the phase of the optical waveform. With access to both in-phase and quadrature components of the optical field, distortion due to chromatic dispersion can be corrected using digital signal processing (DSP).

Coherent detection with DSP avoids the need for fibre dispersion compensation modules. Non-linear distortion and crosstalk are minimised by selecting fibre with high chromatic dispersion. Higher optical power can be launched into the fibre, and in such systems, G.652 fibre will usually offer superior performance to any of the G.655 variants.

A secondary advantage of G.652 fibres is that the effective area is typically around 80 µm2, compared with 70 µm2 for Corning LEAF, and 55 µm2 for OFS TrueWave REACH. This further reduces the non-linear penalty, allowing higher launch powers and improved signal to noise ratio.

https://www.corning.com/media/worldwide/coc/documents/Fiber/LEAF%20optical%20fiber.pdf

https://www.ofsoptics.com/wp-content/uploads/TrueWaveREACHFiber-124-web-4.pdf

https://www.corning.com/media/worldwide/coc/documents/Fiber/SMF-28%20Ultra.pdf

For trans-oceanic submarine systems, single mode fibres are available with even larger effective area and slightly higher dispersion than G.652 fibres.

For terrestrial systems, G.652.D is lower cost, can be connected to existing single mode fibre with few problems, and offers good performance. It is likely to be compatible with future generations of high capacity terminal equipment.

This may explain why your operators prefer G.652.D to G.655.D.

hi Alan Robinson quite detailed and to the point answer, many thanks for your time.

One confusion not directly related to the issue is that why google is still suggesting the old side of the story. Do you have or plan to have written an article over it, I will try to disseminate it as much as possible so that people may get both stories, old and new, and decide themselves.

Hi Abdul Aziz Khan: - Can you explain in more detail how my answer differs from the established view of DWDM performance? In particular:

• Where did Google lead you to an "old side of the story" which contradicts this?
• Have you any links with supporting technical details beyond a simple statement that G.655.D is superior?

A useful resource is the Handbook of Optical Fiber, Cables and Systems, published by the International Telecommunications Union in 2009. By now, this is a little dated and does not cover recent developments in QAM modulation formats and coherent systems. Having said this, the basic information remains valid.

https://www.itu.int/dms_pub/itu-t/opb/hdb/T-HDB-OUT.10-2009-1-PDF-E.pdf

In particular, you may find the following sections may of interest

• Chapter 7, Optical System Design
• Section 2.7 Penalty due to fibre non linearities
• Section 2.7.2 p 180, SRS
• Section 2.7.4 p 181, XPM
• Section 2.7.5 p 182, FWM
• Section 2.7.6 p 184, Maximum power threshold due to non-linear effects

My previous answer reflects my understanding of the state of the art around 2011. At that time I had been involved in developing commercial WDM transmission systems for 18 years. My responsibilities included analysis of non-linear propagation impairments and establishing robust and efficient link budgeting techniques. Since then I have not followed developments in detail, but I am not aware of any technical developments or changes to our understanding of the underlying physics which would change my conclusions.

Consider that Google is a search engine and advertising platform. The links it returns reflect accessibility and some measure of popularity. Judging technical merit is a more difficult task for an automated system.

Out of interest, I did search Google for differences between G652 and G655 fibres, and near the top of the hits found several blog sites of questionable reliability. Is this where you found your opinions?

Note that fomsn.com, fs.com, medium.com, thefoa.com, mefiberoptic.com and mjadom.com, typically display very similar graphics, and most include a table indicating that G652 fibre is unsuitable for long haul DWDM or data rates. This suggests to me that they may be more closely linked than is apparent, rather than expressing independent opinions.

The graph common to several, showing positive and negative dispersion fibres is highly incongruous to anyone familiar with single mode fibre design. Although there was a short period in the mid-1990s where negative dispersion fibres were investigated for their insensitivity to modulation instability, these fibres had positive dispersion slope, not negative as shown in the graph.

Sylvie Liu on https://medium.com/@sylvieliu66/single-mode-fiber-type-g652-vs-g655-fiber-fbbcc6db67ee is more specific. She states "G655 is an enhanced single mode fiber with the characteristic of elimination of FWM and low dispersion value". It absolutely does NOT eliminate FWM. This is not a matter of opinion or debate. Low dispersion is well known to greatly increase the magnitude of four wave mixing,

I note that the post is dated 9 May 2018, but states that G.655 has A, B and C subcategories. She does not mention the G.655.D and G.655.E versions from the 2006 release of the standard, or that the A and B subcategories were dropped. Such lack of attention to detail does not inspire confidence, and she reveals little understanding of non-linear Kerr effect crosstalk in WDM systems.

A more focused Google search found a more recent publication. I have not downloaded the full paper, but the abstract indicates a much more realistic understanding of DWDM propagation. It reports a mixed line rate comparison of G.652, G.652D, G.653, G.654, G.655 and LEAF fibres, with G.652 showing best performance.

Bajpai, R., Sengar, S., Iyer, S. et al. Performance investigation of MLR optical WDM network based on ITU-T conforming fibers in the presence of SRS, XPM and FWM. Int. j. inf. tecnol. 11, 213–227 (2019). https://doi.org/10.1007/s41870-018-0212-2

Article Performance investigation of MLR optical WDM network based o...

The benefits of high chromatic dispersion in suppressing WDM non-linearity was understood from the earliest demonstrations of FWM in optical fibres.

K.O.Hill et. al, "CW three-wave mixing in single mode optical fibers", J. Appl Phys 49, p 5098, 1979. https://aip.scitation.org/doi/abs/10.1063/1.324456

The impact of low dispersion on FWM was explored in more detail in the paper by N. Shibata, R. Braun & R. Waarts, "Phase-mismatch dependence of efficiency of wave generation through four-wave mixing in a single-mode optical fiber" IEEE JQE vol 23, #7, pp1205-1210, (1987), https://ieeexplore.ieee.org/document/1073489/

The analysis was extended to multi-span systems when erbium-doped fibre amplifiers were introduced. https://www.researchgate.net/publication/26861221_Phase-mismatching_characteristic_of_four-wave_mixing_in_fiber_lines_with_multistage_optical_amplifiers

As modulation rates increased from 2.5 to 10 Gb/s, cross phase modulation became the dominant WDM penalty for many systems. There is a rather more complex relationship between fibre dispersion, channel spacing, and dispersion compensation strategy for XPM impairments, discussed by Rongqing Hui, K.R. Demarest, & C.T. Allen, "Cross-phase modulation in multispan WDM optical fiber systems", JLT, 17, #6, pp 1018-1026, 1999.

https://ieeexplore.ieee.org/document/769302

In spite of their poorer DWDM performance, there can be valid reasons to deploy near-zero dispersion-shifted fibres to reduce the need and expense of dispersion compensation modules. Typically, this will be where cost of terminal equipment is more important than maximising capacity and the potential for future enhancements. However, stating that G652 fibre is unsuitable for 10 Gbit/s DWDM is demonstrably untrue.

The introduction of electronic dispersion compensation eliminated much of any potential cost benefit for reduced dispersion fibre in high capacity long-haul transmission, avoiding the need for fibre dispersion compensation modules. For example: Doug McGhan, Charles Laperle, Alexander Savchenko, Chuandong Li, Gary Mak, and Maurice O’Sullivan, "5120-km RZ-DPSK Transmission Over G.652 Fiber at 10 Gb/s Without Optical Dispersion Compensation", IEEE Photonics Technology Letters, Vol. 18, No. 2, January 15, 2006

https://www.researchgate.net/profile/Charles_Laperle/publication/3293938_5120-km_RZ-DPSK_transmission_over_G652_fiber_at_10_Gbs_without_optical_dispersion_compensation/links/02e7e51fbca2d815b6000000.pdf

Coherent detection and digital signal processing at the receiver gave further improvements in performance, even on legacy fibre with relatively poor polarisation mode dispersion.

C.Laperle, "WDM performance and PMD tolerance of a coherent 40-Gbit/s dual-polarization QPSK transceiver" JLT Feb 2008.

https://www.researchgate.net/publication/3244334_WDM_performance_and_PMD_tolerance_of_a_coherent_40-Gbits_dual-polarization_QPSK_transceiver

Note that the papers above used G652 fibre. This is not only because G652 fibre is very widely deployed, but also because it delivers significantly better WDM performance for this type of system.

Here is more recent work using ULA fibre with an even higher chromatic dispersion than G652 to improve the WDM performance

Article Erratum [regarding "High Spectral Efficiency 400 Gb/s Transm...

I have no plans at present to publish a more detailed justification. As I hope I have shown, there is a huge volume of work published in peer-reviewed journals supporting much of what I stated. Have you identified an audience who are unable to access such information, or do they simply need to know what questions to ask?

If you google the issue then you find G.655 as the preferred solution

Below is the link that comes as the first choice and they have a table, attached, that says G.652 is for CWDM and G.655 is for DWDM:

https://www.fomsn.com/optical-fibers-and-cables/ingrid/single-mode-fiber-comparison-g-652-vs-g-655/

Abdul Aziz Khan - Yes I found the link to the Fibre Optic Social Network page and read the contents.

As I said in my previous two answers, there are circumstances where G.655.D fibre is an appropriate choice. For moderate link lengths, especially with channel spacing 200 GHz or greater, it can have advantages.

However, for long-haul high capacity DWDM transport with 100 GHz or 50 GHz channel spacing, non-linear crosstalk from FWM and XPM will result in much poorer performance compared to G.652 fibre with appropriate dispersion compensation - either by fibre compensation modules or by electronic dispersion compensation.

The statement made in the last cell of the table, that G655 fibre has: "low dispersion value: overcoming nonlinear effect" is unsupported and simply incorrect.

If you do not accept the various references I supplied in my previous answer, here is a report from the UK National Physical Laboratory, our national standards institution, with a highly respected reputation as experts in metrology, including fibre optic metrology.

"The Synthesis of L-Band Wavelength References Via Four-Wave Mixing", Helen White & Colin Campbell, NPL report CETM 53, April 2003.

https://eprintspublications.npl.co.uk/2748/1/CETM53.pdf

Equations 1.3, 1.8 and 1.9 are equivalent to the equations presented in the papers I linked by Hill (1978), Shibata (1987) and Inoue (1992). They are similar to the equations I personally derived, and give predictions which were verified by laboratory measurements I supervised at Standard Telecommunication Laboratories in 1996.

The NPL report includes both measurements of commercially available fibres, and simulations of FWM efficiency over a range of wavelengths separations and fibre types. These confirm that a reduction in wavelength separation and a decrease in chromatic dispersion coefficient both lead to a strong increase in FWM power, contrary to the statements on the web site you linked.

Note also that Appendix A, page 35 of the NPL report shows the wavelength dependence of 4 types of transmission fibre, including "Submarine LEAF", an NZDSF or G655 type fibre, with negative dispersion at C-band wavelengths around 1550 nm. Unlike the image shown on the fomsn page, the dispersion slope is positive, not negative as they purport for (-D) NZDSF.

The "Specified Wavelength" range in your table is 1550 to 1625 nm. This is not what the standard states. G655.C is specified between 1530 nm and 1565 nm. G.655.D and G.655.E are both specified over the range 1460 nm to 1625 nm.

The "Division" entry in the table presumably refers to the G.655.A, G.655.B and G.655.C categories. They do not mention G.655.D, or G.655.E (which offers better suppression of FWM). They seem unaware than while G.655.A and G.655.B featured in the 2003 version of the standard, neither appear in the 2009 release of the standard, or even the previous 2006 release. For a blog post from September 2018, this shows poor attention to detail.

I take this all as further evidence that the claims expressed on fomsn.com cannot be trusted.

They may be unfamiliar with the published literature and long established industrial experience, or they might have decided to slant their presentation in favour of a more expensive product with a higher profit margin. Feel free to make your own assessment.