Modal launch conditions—what the standards say

Differing modal launch conditions can result in inconsistent measurements of insertion loss. Fortunately, standards-based ways to measure and control modal launch are now in place.

Content Dam Etc Medialib New Lib Lw Print Articles Lw Volume 26 Issue 9 0909lwapp01f2

By David Robinson


Differing modal launch conditions can result in inconsistent measurements of insertion loss. Fortunately, standards-based ways to measure and control modal launch are now in place.

It has been known for many years that modal launch conditions have a major effect on the attenuation and bandwidth measured in multimode cabling systems. As the bandwidth of LANs is pushed ever higher to meet demand, loss budgets are squeezed ever tighter. This has led to an increasing amount of attention being paid to minimizing the insertion loss in multimode cables and connector components.

For example, the IEEE 10-Gbps shortwave specification over 300 m (10GBase-S) specifies a maximum channel loss of 2.6 dB, which includes 1.5 dB for connector loss. So in a link with two connections, the maximum allowed loss on each connection is 0.75 dB.

The loss in a connection can vary by several tenths of a decibel, depending on the modal distribution launched. There is a requirement, therefore, to specify the degree of mode-filling so that a) loss measurements are representative of typical transmission sources and b) repeatable measurements can be obtained.

The biggest contributor to discrepancies in loss measurements is that different light sources may have very different modal launch conditions. Fortunately, standards bodies and equipment suppliers have responded to this challenge, providing launch specifications that can ensure that measurement accuracy to at least 0.08 dB may be obtained, which is approximately one-tenth of the maximum loss allowed per connector.

Dealing with insertion loss

Multimode fiber cable has a major advantage over singlemode in short-haul applications, where its larger core diameter and higher numerical aperture mean it is much easier to get light in and out of the cable and it is much more tolerant to imperfect splicing, connectorization, and general cleanliness and handling. There are also cost benefits in installing multimode networks instead of singlemode ones, although not in the cost of the cable itself but in system equipment.

Insertion loss is arguably the most fundamental parameter of an optical fiber cable. It has been shown in a number of studies that controlling modal launch conditions is by far the most significant contributor to variations in insertion loss measurements in multimode fiber cables and connectors. If you measure the loss of a cable using two light sources with different modal launch conditions you will get a different answer. If the sources have the same launch conditions you will, within acceptable limits, get the same result.

An experiment, in which the insertion losses of 11 concatenated patch cords were measured, demonstrated this point. Each patch cord was 3 m long and of 50-µm graded-index fiber; the connector types used were a combination of FC, SC, and ST. The insertion loss was measured using a variety of light sources with different modal distributions. The loss results, shown in Fig. 1 as green diamonds, range from 0.5 dB, for an axial singlemode launch, to 5.0 dB for an offset singlemode launch, where only high-order modes were excited. Some light sources were commercial test sets and others were empirical.

The loss measurements were then repeated using the same light sources but this time with a commercially available mode controller inserted between the light source and the patch cords under test, so the launched modal distributions were the same for all of the sources. In this case the insertion loss measurements, shown in Fig. 1 as purple squares, varied by only about 0.2 dB.

It is important to point out here that this says nothing about the accuracy of the insertion loss measurement. In a light source/power meter (LSPM) type measurement, the accuracy is determined by a number of factors such as stability of the light source, wavelength accuracy, quality of the reference patch cords, noise in the detector electronics, etc. If due care is taken, the measurement accuracy can be very good. The difference in loss measured here is real; the cable loss does actually depend on the modal distribution in the fiber.

Modal distribution and standards

So what exactly do we mean by “modal distribution”? Due to the electromagnetic properties of light, optical power traveling in a multimode fiber is restricted to certain discreet paths, or modes. In a typical 50-µm multimode fiber, at 850 nm, there are around 380 individual modes that can support light transmission. These modes may be categorized into 19 mode groups, where each mode in a group has the same propagation constant, or transmission velocity.

If every possible transmission mode in the fiber is occupied, the fiber is referred to as “fully filled.” However, in practice this never happens. In any practical fiber, light transmitting in so-called high-order modes is following a path that forces it close to the outside of the fiber core, making it susceptible to losses as the fiber is bent, or at small offsets, such as at connectors or splices. By contrast, the “low-order” modes stay close to the center of the fiber core and are much less lossy. So using this simple analysis, it’s clear that if the light launched into the fiber contains a lot of power in the high-order modes, by the time the light exits the fiber it will have lost more power than a fiber where only low-order modes are present.

Therefore the solution is really quite simple. The loss measured depends on the balance of low- and high-order modes present at the input to the cable. So, if all parties use the same launch distribution, loss measurements immediately become much more consistent. This leads to fewer discussions among fiber makers, cable makers, and system installers, and fewer problems at the user’s premises.

Of course, to standardize modal distributions we need to be able to accurately measure them. There are a number of ways to define and measure modal distribution, but over the last few years the telecom industry has moved toward a parameter called “encircled flux.” Encircled flux has been used for several years for the characterization of fiber-pigtailed vertical cavity surface-emitting lasers (VCSELs) and is described by IEEE standards such as IEEE 802.3aq. More recently, the TIA (TIA/EIA-455-203) and IEC (IEC 61280-1-4) have described the measurement technique in some detail.

As the name suggests, the method involves measuring the light intensity at a planar surface parallel to the optical axis of the fiber, for example, at the surface of a connector. This 2D map of the optical power is processed to calculate the optical power, or flux, contained in circles of increasing diameter, starting from the optical center of the fiber (see Fig. 2).

To successfully implement this technique, high-quality imaging optics and light detection equipment are needed, as well as traceable calibration of the optical magnification. Fortunately, standards-compliant equipment is now commercially available, which enables accurate, real-time measurements to be made (see photo).

It should be noted that the aerospace industry generally prefers a different definition of modal launch conditions. Sometimes called a “limited phase space” (LPS) definition, this definition requires the spot size and numerical aperture of the launched light to be, for example, 85% of the theoretical values of the fiber. Work is ongoing to compare the two methods. LPS launches are also used by some fiber manufacturers who typically measure many kilometers of fiber wound onto a small reel.

International standards are now in place that define modal launch requirements. For example, IEC recently introduced a new version of IEC 61280-4-1, which defines in detail the encircled flux required in the measurement of cable systems. It defines the modal launch conditions as a series of upper and lower limits at various radial positions.

While cable system and test equipment manufacturers work hard to make their measurement equipment compliant, at the sharp end installers and users need quick and easy ways to ensure they are doing the right thing when testing and commissioning installed systems. To take care of this, services are available that can measure the encircled flux of test sources. If the sources aren’t compliant, mandrels may help as would commercially available modal controllers, which will produce the right launch conditions from any source.

So, in summary, a major step has been taken by the industry to ensure that multimode-fiber-based systems can continue to provide cost-effective options in short-haul, high-bandwidth applications. An international effort over a number of years has resulted in the release of appropriate standards, and commercially available equipment for the measurement and control of modal launch conditions is now available. This will lead to much improved agreement between loss measured in different locations and using different equipment.

David Robinsonis managing director of Arden Photonics Ltd. (

Links to more information

LIGHTWAVE:Certifying MMF for 100G Ethernet Transmission
LIGHTWAVE: Take Control of Your Link-Loss Budget
LIGHTWAVE:CABLING INSTALLATION & MAINTENANCE: Meeting Reach Objectives for 40/100-GbE Channels

More in Design & Manufacturing