Articles
  • Register

Fiber Optic Technology - Part Five - Optical Fiber Testing

In part three and four of our Fiber Optic Technology series we have provided information for the installation and termination of fiber optic cables. In this final part of the series, we now look at the next part of the process, testing.

For every fiber optic cable plant, you will need to test for continuity, end-to-end loss and then troubleshoot the problems.

There are several ways of testing fiber optic cables:

  • Continuity testing with a visible light source - a LED or incandescent bulb in a fiber tracer or a higher power visible laser in a visual fault locator tells you if the fiber is continuous and if the connections from end to end is correctly set up (e.g. transmitter to receiver). A visual fault locator can also find problems like bending losses and broken fibers.
  • Visual inspection with a microscope allows checking connector ferrule ends for scratches, cracks, dirt and other contamination. It's always used in termination, but is also a valuable troubleshooting tool.
  • Insertion loss simulates the way the cable will be used by a transmission system, using a source (LED or laser) at the same wavelength(s) as the system and optical power meter, with two reference cables. This test is required by all network testing standards.
  • OTDR testing uses a unique property of fiber - backscatter - to create a "picture" of the fiber and find faults. It's generally used for splice verification and troubleshooting, but some users also test bare fibers in cables on spools before installing it - especially when it's long cables and installation is expensive.

In addition, there are several other parameters that may need testing:

  • Back reflection testing is very important for singlemode terminations as it can lead to laser instability and high optical background noise in a cable.
  • Bandwidth testing is becoming a problem as multimode fibers are used for multi-gigabit networks. Testing can be done in several ways in the factory but options are fewer in the field.


Testing fiber optic components and cable plants requires making several measurements with the most common measurement parameters listed below. Optical power, required for measuring source power, receiver power and, when used with a test source, loss or attenuation, is the most important parameter and is required for almost every fiber optic test. Backscatter and wavelength measurements are the next most important; and bandwidth or dispersion are of lesser importance. Measurement or inspection of geometrical parameters of fiber are essential for fiber manufacturers. And troubleshooting installed cables and networks is required.

Most test procedures for fiber optic component specifications have been standardized by national and international standards bodies, including TIA in the US and ISO/IEC internationally. Procedures for measuring absolute optical power, cable and connector loss and the effects of many environmental factors (such as temperature, pressure, flexing, etc.) are all covered in this testing process.

 In order to perform these tests, the basic fiber optic instruments are:

Inspection Microscope

Fiber Optic Power Meter & Test Sources

Optical Time Domain Reflectometer (OTDR)

Visual Fault Locators


Visual Inspection with Microscope

DINTEK's Handheld Fiber Inspection Probe

Cleaved fiber ends prepared for splicing and polished connector ferrules require visual inspection to find possible defects. This is accomplished using a microscope which has a stage modified to hold the fiber or connector in the field of view. 

Fiber Optic Inspection Microscopes vary in magnification from 30 to 400 power, with 30-100 power being the most widely used range. Cleaved fibers are usually viewed from the side, to see break-over and lip. Connectors are viewed end-on or at a small angle to find polishing defects such as scratches. 

Fiber Optic Power Meter

DINTEK's Handheld Optical Power Meter

Fiber Optic Power Meters measure the average optical power out of an optical fiber. Power meters typically consist of a solid state detector, signal conditioning circuitry and a digital display of power. 

To interface to the large variety of fiber optic connectors in use, some form of removable connector adapter is usually provided. Power meters are calibrated to read in dB referenced to one milliwatt of optical power. Some meters offer a relative dB scale also, useful for loss measurements since the reference value may be set to "0 dB" on the output of the test source.

Fiber Optic Test Sources

DINTEK's Handheld Optical Light Source

In order to make measurements of optical loss or attenuation in fibers, cables and connectors, you must have a test source as well as a Fiber Optic Power Meter. The test source must be chosen for compatibility with the type of fiber in use (singlemode or multimode with the proper core diameter) and the wavelength desired for performing the test. Most sources are either LED's or lasers of the types commonly used as transmitters in actual fiber optic systems, making them representative of actual applications and enhancing the usefulness of the testing. Some tests, such as measuring spectral attenuation of fiber, requires a variable wavelength source which is usually a tungsten lamp with a mono-chromator to vary the output wavelength.

Typical wavelengths of sources are 650 or 665 nm (plastic fiber), 820, 850 and 870 nm (short wavelength glass fiber) and 1300 or 1310 nm and 1550 nm (long wavelength). LED's are typically used for testing multimode fiber and lasers are used for singlemode fiber, although there is some crossover, especially in high speed LANs which use multimode fiber with lasers and the testing of short singlemode jumper cables with LED's.

Optical Time Domain Reflectometer (OTDR)

DINTEK's Optical Time Domain Reflectometer

The Optical Time Domain Reflectometer (OTDR) uses the phenomena of fiber backscattering to characterize fibers and installed cables, find faults and optimize splices. Since scattering is one of the primary loss factors in fiber (the other being absorption), the OTDR can send out into the fiber a high powered pulse and measure the light scattered back toward the instrument. 

The pulse is attenuated on the outbound leg and the backscattered light is attenuated on the return leg, so the returned signal is a function of twice the fiber loss and the backscatter coefficient of the fiber.

Visual Fault Locators

DINTEK's Pen-type Visual Fault Locator

Many of the problems in connection of fiber optic networks are related to making proper connections. Since the light used in systems is invisible, one cannot see the system transmitter light. By injecting the light from a visible source, such as a LED or incandescent bulb, one can visually trace the fiber from transmitter to receiver to ensure correct orientation and check continuity besides. 

The simple instruments that inject visible light are called Visual Fault Locators.

​Continuity testing is done with a visible light source - a LED or incandescent bulb in a fiber tracer or a higher power visible laser in a Visual Fault Locator. The low-powered fiber tracer - made from a flashlight or LED source - can be used to confirm that light can indeed be transmitted through the fiber and the proper connections between transmitter and receiver have been made.

The higher powered laser in a Visual Fault Locator can trace fibers over longer distances and even find breaks or locations of stress loss. In a break, the light lost can be seen through the jacket of simplex or zipcord cable and tight buffered fibers. It's the essential companion to the OTDR, since it finds faults near to the instrument, which is impossible with the OTDR.

Visual Fault Locators can also be used to optimize mechanical splices and prepolished/splice type connectors by adjusting the fibers to minimize the visible light lost.

Fiber Identifiers are a bit like Visual Fault Locators in that they are used to identify fibers, but can do this with light at the wavelength being used in the system. They use the fact that bending fiber causes loss. By inserting a tight but carefully-controlled bend in the fiber, it extracts light which is caught by a detector in the head of the instrument.

Most fiber identifiers can distinguish between high-speed network signals and a 2 kHz signal injected by a test source, so you can find fibers carrying a signal or identify a specific fiber into which you inject a test source.

Connector Visual Inspection 

You can visually inspect the polished end of a connector ferrule with a microscope to see that the ferrule is properly polished, there are no cracks in the fiber and that the tip is smooth and free of scratches. And of course, you can see dirt and any other contamination on the end of the ferrule that can affect light transmission through the connection.

There are many inspection microscopes available with magnifications of 100X to 400X, such as DINTEK's 250x Handheld Fiber Inspection Probe. Higher magnification may not be better, as it tends to make you more critical of scratches and imperfections. Lower magnification works just fine. A direct view at 100 times magnification with a low intensity light shining through the fiber core should look like the image below on the left. The bright dot in the center is the core of the fiber and the darker annular ring is the cladding. On this connector, notice the dark area to the left of the core, in the cladding. This appears to be a small crack in the fiber that only affects the cladding, not the core, so it is not a problem. If the crack had been in the core, we would not have seen a round dot for the illuminated core.

You should also look at the tip under the microscope at an angle if this is possible with the microscope you are using. Some microscopes actually hold the connector at an angle while others shine light from the side to get a similar effect. The angular view will highlight any surface irregularities better than the head-on view. It may look like the image shown below to the right. Now you can see some small amount of epoxy still on the end of the ferrule, which shows up as the dark, uneven ring around the fiber (the ring is caused by the convex end of the PC ferrule.) You can also see the dark area to the left of the fiber, which is the small crack we saw on the direct view, but is more obvious here. The core should be nice and smooth, an even gray color, with no big scratches. If you see large scratches, it may need repolishing or retermination.

Fiber Optic Power Meters

The power meter is a specialized light meter with adapters to allow connection to the fiber optic connectors on the cables and calibration at the wavelengths used in fiber optic systems. The power meter needs to be calibrated to be able to measure at appropriate wavelengths (850, 1310 or 1550 nm.) Most meters have measurement ranges for "dBm" for measuring absolute power and "dB" or relative power for measuring loss. If it has a dB scale, it will be able to set a "0 dB reference" for simplifying loss measurements. Some meters may have a linear range - usually milliwatts - used for measuring absolute power. dB is a measure of optical power on a log scale, simplifying measurements over a wide dynamic range.

Fiber optics uses power levels from +20 to -40 dBm, a range of 1,000,000 to 1! But that translates to 60 dB, an easier number to deal with. Absolute power is measured in dBm or dB referenced to 1 mw. Positive dBm means the power is greater than 1 mw, while negative numbers mean the power level is less than 1 mw. The ratios shown give you an idea of how dB relates to linear power in watts. A nice thing about dB is that loss is easily measured by subtracting the reference level for "0" dB from the measured value of the loss. That is, if you measure -20 dBm from the end of the reference cable, then -22 dBm when testing cables, the cable loss is 2 dB.

Optical Power Testing - FOTP-95

With data communication networks, we are primarily concerned with three fiber optic test procedures: FOTP-95 for measurements of optical power, FOTP-171 for testing patch cables and OFSTP-14 for testing the loss of the installed cable plant. Optical power is typically measured to check source power output at the transmitter or receiver power at its input. Optical power is measured with the power meter attached to the system cable or, when testing source output, a reference test cable.

Transmitter Power

The amount of light coupled into a fiber by a source is measured by attaching a patchcord to the source, either a known good system patchcord or a reference test cable. The cable used must have a connector that mates with the transmitter and a fiber size the same as the system cabling (50/125, 62.5/125 or SM) since the coupled power is highly dependent on the core size of the fiber. The meter connector adapter must be the same as the cable to allow connection.

Connect the meter, set the range on dBm or watts as appropriate and be sure to set the wavelength to the wavelength of the source, as the meter's calibration will be different due to the wavelength sensitivity of its detector!
Receiver Power

Receiver power is measured by removing the cable connected to the receiver input and connecting it to the power meter.

Set the meter range on dBm or watts as appropriate and be sure to set the wavelength to the wavelength of the source, as the meter's calibration will be different due to the wavelength sensitivity of its detector!

 Measure the power and record the results


Optical Power Testing - FOTP-171

Insertion loss testing simulates the way the cable will be used by the systems operating over it. A source, similar to the system source, is used for inserting light into the cable under test. A meter is used to measure the source output and the loss when the cable under test is added. Known-good reference cables are used to mate with the cable under test to insert light into the cable and allow testing loss of the connectors on the cable. A double-ended test like this measures the loss of the fiber and connectors on both ends, plus anything in the middle.

The test equipment for insertion loss testing includes a test source (L) and a meter (R), sometimes available as separate instruments and sometimes integrated into one instrument, called an OLTS (Optical Loss Test Set.) The source, similar to the system source, is used for inserting light into the cable under test. The meter is a specialized light meter used to measure the source output and the loss when a cable under test is added. A double-ended test like this measures the loss of the fiber and connectors on both ends, plus anything in the middle.

​A FOTP-171 test uses only a single launch reference cable to test the cable. This method allows testing a single cable from either end to find out if one connector is bad. Its main use is testing patchcords to ensure both connectors are good, but it can also be used to troubleshoot installed cables where one connector is suspected of being bad. The 0 dB loss reference is made by connecting the power meter to the output of the launch cable and measuring the power output. The cable under test is connected to the launch cable and the meter. The loss measured is only the loss of the mated connectors and any loss of the fiber in the cable, usually very small when testing patchcords this way.

The fact that the connector on the launch cable and the cable under test are mated directly to the meter, with its large detector, means that the connection loss to the meter is calibrated out of the loss test, allowing testing of only the connector mated to the launch cable.

Optical Power Testing - OFSTP-14

​OFSTP-14 is used for testing installed and terminated cable plants, where we want to test the connectors on each end and everything in between. So we use a meter and source with two reference cables - one on each end.

The big issue with this test method is how one sets the 0 dB reference.

Method A: with 2 cables (launch and receive cables)

This method sets the "0 dB reference" with the launch cable mated to the receive cable, so that one mated connector loss is included when setting the reference. Then when testing a cable with both launch and receive cables, the loss includes the loss of connectors on the cable under test and the loss of all the components in between, less the loss of the mated connectors included in the reference.

Method B: with 1 reference cable (the launch cable)

This method sets the "0 dB reference" with the power meter measuring the output of the launch cable directly, so that no connector loss is included when setting the reference. Then when testing a cable with both launch and receive cables, the loss includes the loss of both connectors on the cable under test and the loss of all the components in between.

Method C: 3 cables (launch, receive & "golden" cable)

This method sets the "0 dB reference" with the launch cable and the receive cable, plus a "golden" reference cable mated to them, so that two mated connector losses and any fiber loss in the third cable are included when setting the reference. Then when testing a cable with both launch and receive cables, the loss includes the loss of connectors on the cable under test and the loss of all the components in between, less the loss of the mated connectors included in the reference.

What is the reason for three different methods?  

Method A: with 2 cables (launch and receive cables)​

This method works if you need to test cables with connectors that can be mated to each other but do not mate with the test equipment, e.g. testing LC connector cables with test equipment with fixed ST connector interfaces.

Set your reference by connecting a hybrid cable (one with a connector that mates to the instrument on one end and the cable plant to be tested on the other, e.g. ST to LC) to the meter and another to the source. Mate them in the middle with an adapter and measure the power for your "0dB" reference. Once you set the reference value, do not remove the launch cable from the source as it may not couple exactly the same power, ruining the set reference. Meters usually have large detectors so disconnecting the meter cable is probably not a problem, but it is still recommended to not remove it either.

Note the reference includes one mated pair of connectors, so the loss you measure will be that much less than using Method B. ALWAYS report the method used whenever providing test results so the user can properly evaluate the measurements.

Method B: with 1 reference cable (the launch cable)

The reference is set using a single cable only connected between the meter and source, so no connector mating losses are included in setting the reference. Simply connect the meter and source with the reference cable and measure your "0dB" reference power. Set the meter to "0" if it has a zero loss set feature.

Once you set the reference value, do not remove the launch cable from the source as it may not couple exactly the same power, ruining the set reference. This method works when you have test equipment with adapters on the meter that match the connectors on the cables being tested and all connectors are the same. It can also be used if the meter has replaceable adapters for connectors that do not affect calibration, so you can set a reference with one connector and test with another by changing the adapter, eg using ST connectors to test FC, SC or other 2.5 mm ferrule connectors.


Method C: 3 cables (launch, receive & "golden" reference cable)
This method works if you need to test cables with connectors that can be mated to each other but do not mate with the test equipment, e.g. testing MT-RJ connector cables with test equipment which has fixed ST connector interfaces. Set your reference by connecting hybrid cables (one with a connector that mates to the instrument on one end and the cable plant to be tested on the other, eg ST to MT-RJ without pins) to the meter and another to the source. Using a compatible cable, eg MT-RJ with pins on either end, mate them in the middle with an adapter and measure the power for your "0dB" reference.

Once you set the reference value, do not remove the launch cable from the source as it may not couple exactly the same power, ruining the set reference. Meters usually have large detectors so disconnecting the meter cable is probably not a problem, but it is still recommended to not remove it either. Note the reference includes two mated pairs of connectors, so the loss you measure will be that much less than using Method B. ALWAYS report the method used whenever providing test results so the user can properly evaluate the measurements.

​Insertion Loss Testing difference with three reference methods

When you measure the loss of a cable plant using these three different methods, you will obviously get different losses, since Method B includes no mated connector losses, while Methods A and C include one and two mated connector losses respectively.

Below are carefully controlled results in a lab of the same cable plant tested with the same reference cables but using all three reference methods. As you see, the loss is highest with Method B and lowest with Method C, but notice the standard deviation, a measure of the reproducibility of the measurement, is much higher - worse - as you include the connectors. Thus the most repeatable measurement is Method B, also the highest loss. Method A yields a measurement of about 0.3 dB less loss (obviously the connectors on our reference cables were good - 0.3 dB mated loss) and Method C about 0.5 dB less. Again, all three methods are acceptable, as long as the method is included in the documentation.



After a fiber optic cable plant is installed, it may be used with a number of different types of fiber optic networks. Computer networks, telephone signals, video links, and even audio can be sent on the installed fibers. Each network type has a requirement for the performance of the fiber optic cable link. Most simply specify the maximum loss in the link that can be tolerated, a function of component specifications and installation quality. Others also specify the bandwidth performance of the fiber which is determined by the specifications of the fiber chosen.

Every fiber optic link has a maximum loss of a link over which it can work. That loss is determined by the output power of the transmitter coupled into the fiber and the sensitivity of the receiver, all expressed in dB. The loss of the fiber optic cable it uses must be less than the maximum loss for proper operation. While every link installed must meet some maximum loss to allow operation of the network intended to use it, different networks may have different link margins. Therefore we use a different approach. The loss of the link is considered acceptable if it is less than standard maximum values calculated from the characteristics of the link installation.

What causes the losses in the fiber optic cable? First - the fiber itself. The next loss factor is terminations. Splices are common in singlemode but rare in multimode networks Singlemode fiber is usually spliced with a fusion splicer which welds the two fibers together in an electric arc, with much lower losses.

The final loss factor is stress in installation. Fiber optic cable pulled with too much tension may be damaged. Each time you make a bend with a fiber optic cable, you put some stress in the fiber which can cause loss. Even cable ties tightened on the cable can cause loss. Stress loss should be zero!

Calculate the Loss Budget

Calculating the loss budget is the best way to estimate what loss we should be measuring. To calculate the loss budget, we figure what is the maximum loss with a normal installation. To begin, we need to know the approximate length of the link and the number of connectors and splices. For connectors, count the connectors on each end as one each (we'll mate them to reference connectors when we test them) and each mated pair used as a connection in the cable link as one also.

Consider a simple link as an example:

[Conn]--------------100m---------------[Conn]=Adapter=[Conn]----------------200m-------------[Conn]
Link length: 0.3 km (~1000 feet)
Connectors: 3 (one on each end and one patch connection in the middle)
Splices: none

To calculate the loss in the fiber optic cable, multiply the length times the attenuation at each wavelength:

0.3 km X 6.3 dB/km @ 850 nm = 1.05 dB loss and 0.3 km X 1.5 dB/km @ 1300 nm = 0.45 dB loss.

For the connector loss, 3 connectors X 0.75 dB = 2.25 dB. If we had splices, we would calculate the total loss the same way. Adding the loss of the fiber to the termination losses gives us the total loss. Thus our link should have a maximum loss at:

850 nm of 1.05 dB + 2.25 dB = 3.30 dB. At 1300 nm, the loss drops to 0.45 dB + 2.25 dB = 2.70 dB.

These numbers become "pass/fail" numbers for testing. When you test the link, you should have less than 3.3 dB loss at 850 nm and 2.70 dB at 1300 nm. If the loss is higher than that, you may have problems with the cable installation or the terminations, and must troubleshoot the installation.


Insertion Loss Troubleshooting

​The first step in troubleshooting is to check your reference cables. Are they clean and test with low loss in a single-ended test? Do they match the size of the fiber you are testing? We've gotten calls from people saying everything they test has 20 dB of loss but the connectors look great in a microscope! Well, that's what you get testing SM fiber with MM reference cables. A consistent 2-4 dB means you may be testing 50 micron fiber with 62.5 patchcords.

If that's not the problem, recheck your instruments to make sure the reference for "0 dB" has not changed. If the instruments and reference cables are OK, retest the questionable cable plant segment by segment, so you are only testing one fiber with connectors on each end. If all the fibers in a cable test bad, start looking for stress on the cable causing loss. It may also show up as loss at 1300 nm being equal to 850 nm in MM cable, although it should be much less (fiber loss is about 1 dB/km at 1300 nm and 3 dB/km at 850 nm). In SM cable, loss at 1550 nm is highly sensitive to stress loss so the difference to 1310 nm is reduced.

For any single fiber link, test single ended from both directions. A bad connector will have high loss mated to the reference connector but less effect when mated to the detector on the meter, so one way should have higher loss – and the bad connector is then the one mated to the launch reference cable.

Optical Return Loss in Connectors

​If you have ever looked at a fiber optic connector on an OTDR, you are familiar with the characteristic spike that shows where the connector is. That spike is a measure of the back reflection or optical return loss of the connector, or the amount of light that is reflected back up the fiber by light reflecting off the interface of the polished end surface of the connector and air. It is called Fresnel Reflection, and is caused by the light going through the change in index of refraction at the interface between the fiber (n=1.5) and air (n=1). For most systems, that return spike is just one component of the connector's loss, representing as much as 0.3 dB loss.

In high-bit rate singlemode systems, that reflection can be a major source of bit-error rate problems. The reflected light interferes with the laser diode chip, causes mode-hopping and can be a source of noise. Minimizing the light reflected back into the laser is necessary to get maximum performance out of high bit rate laser systems, especially the AM modulated CATV systems.

In any short singlemode link, back reflection can cause problems by reflecting back and forth many times in a link, creating "optical background noise" that confuses receivers. Problems with transmission, e.g. high bit-error-rate, in short singlemode links can often be traced to highly reflective connectors. Since this is primarily a problem with singlemode systems, manufacturers have concentrated on solving the problem for their singlemode components. Several schemes have been used to reduce back reflections, including reducing the gap between connectors to a few wavelengths of light, which stops the Fresnel Reflection. The usual technique involves polishing the end surface of the fiber to a convex surface or at a slight angle to prevent direct back reflections.

Testing optical return loss or back-reflection as it is sometimes called can be done with a meter and source, OCWR (Optical Continuous Wave Reflectometer) or an OTDR.

Optical Return Loss Test Procedure FOTP-107

​For the EIA FOTP-107 test procedure, one needs a calibrated coupler which can be used to inject a source into the test cable or pigtail and measure the light reflected back up the fiber, along with a standard power meter and laser source. The coupler split ratio must be calibrated to know how much of the return signal goes to the power meter and how much is diverted to the source side of the coupler to calculate the total amount of back reflection. Due to the dynamic range required to measure return losses in the range of -25 to -60 dB, a high power laser source is generally necessary, and the source must be stable enough to allow making accurate measurements over relatively long times required for the experiment. To measure the connector reflection, the far end of the cable must be terminated by surrounding the fiber in an index matching gel or epoxy.

To measure return loss, measure the amount of power transmitted to the end of the cable (P-out) and the power reflected back up through the coupler test port (P-back) with a fiber optic power meter. To calibrate out any crosstalk in the coupler or the back reflections of any intermediate connectors or splices, dip the connector end being tested in an index matching fluid (alcohol works well and isn't messy to clean up) and record the power at the coupler test port (Pzero). If the coupler split ratio is Rsplit (the fraction of the light that goes to the measurement port when transmitting in the back direction), the return loss is:

Some people have the impression that fiber has infinite bandwidth, but it's not true. In fact, the distance fiber can carry network signals depends as much on bandwidth as loss - sometimes more. 

There are several factors that affect the bandwidth of singlemode fiber, but the two major ones for singlemode fiber are chromatic dispersion, of the fact that light of different colors travels at different speeds in glass (the definition of index of refraction) and polarization mode dispersion, caused by the varying speeds of planes of polarization.

Equipment (expensive and complicated) are available to test these factors for long SM links, but it's beyond the scope of this article. In multimode fiber, you have chromatic dispersion for the same reasons as in SM fiber, but you also have modal dispersion, caused by the different path lengths light follows in the larger core.

 While these factors are tested in the lab by fiber manufacturers, field testing is not done. However, bandwidth testers for MM fiber may become available in the near future due to the high bandwidth requirements of networks like 10 GbE.

Using test instrumentation to test your fiber optic cable plant without understanding how they work can be disastrous. While today's instruments are accurate and easy to use, they all require adequate knowledge of their operation and "quirks" to get good data. 

As an example, we have seen several instances where users of OTDRs (Optical Time Domain Reflectometers) accepted the automatic results of the instruments without evaluating the displays (or perhaps not knowing how to interpret the displays.) The data was highly misleading and the consequences of the bad data was very costly. The reason was simply that the OTDR was being used outside of its normal operating parameters and the interpretation of the display is critical to understanding what is happening in the cable plant. Here we will examine the OTDR in detail and show examples of good and bad data. Then we will try to give you guidelines on using them.


When do you use OTDRs?

If you are installing an outside plant network such as a long distance network or a long campus LAN with splices between cables, you will want an OTDR to check if the fibers and splices are good. The OTDR can see the splice after it is made and confirm its performance. It can also find stress problems in the cables caused by improper handling during installation. If you are doing restoration after a cable cut, the OTDR will help find the location of cut and help confirm the quality of temporary and permanent splices to restore operation. On single mode fibers where connector reflections are a concern, the OTDR will pinpoint bad connectors easily.

OTDRs should not be used to measure cable plant loss. That is the job of the source and power meter, which duplicates the actual fiber optic link, as we described in the first part of this article and is documented by every standard ever written for cable plant loss. The loss measured will not correlate between the two methods; the OTDR cannot show the actual cable plant loss that the system will see.

The limited distance resolution of the OTDR makes it very hard to use in a LAN or building environment, where cables are usually only a few hundred feet long. The OTDR has a great deal of difficulty resolving features in the short cables of a LAN and is more often than not simply confusing to the user.

How does an OTDR work?
Figure 1

​Unlike sources and power meters which measure the loss of the fiber optic cable plant directly, the OTDR works indirectly. The source and meter duplicate the transmitter and receiver of the fiber optic transmission link, so the measurement correlates well with actual system loss. The OTDR, however, uses a unique phenomenon of fiber to imply loss.

The biggest factor in optical fiber loss is scattering. It is like billiard balls bouncing off each other, but occurs on an atomic level between photons (particles of light) and atoms or molecules. If you have ever noticed the beam of a flashlight shining through foggy or smokey air, you have seen scattering. Scattering is very sensitive to the color of the light, so as the wavelength of the light gets longer, toward the red end of the spectrum, the scattering gets less. Very much less in fact, by a factor of the wavelength to the fourth power - that's squared-squared. Double the wavelength and you cut the scattering by sixteen times!

You can see this wavelength sensitivity by going outside on a sunny day and looking up. The sky is blue because the sunlight filtering through the atmosphere scatters like light in a fiber. Since the blue light scatters more, the sky takes on a hazy blue cast.

Figure 2: OTDR Display

​In the fiber, light is scattered in all directions, including back toward the source as shown in Figure 1 above. The OTDR uses this "backscattered light" to make its measurements. It sends out a very high powered pulse and measures the light coming back. At any point in time, the light the OTDR sees is the light scattered from the pulse passing through a region of the fiber. Think of the OTDR pulse as being a "virtual source" that is testing all the fiber between itself and the OTDR as it moves down the fiber. Since it is possible to calibrate the speed of the pulse as it passes down the fiber, the OTDR can correlate what it sees in backscattered light with an actual location in the fiber. Thus it can create a display of the amount of backscattered light at any point in the fiber, as shown in Figure 2.

Figure 3: Increasing the Pulse Width

​There are some calculations involved. Remember the light has to go out and come back, so you have to factor that into the time calculations, cutting the time in half and the loss calculations, since the light sees loss both ways. The power loss is a logarithmic function, so the power is measured in dB. The amount of light scattered back to the OTDR is proportional to the backscatter of the fiber, peak power of the OTDR test pulse and the length of the pulse sent out. If you need more backscattered light to get good measurements, you can increase the pulse peak power or pulse width as shown in Figure 3.

Note on the display shown in Figure 2, some events like connectors show a big pulse above the backscatter trace. That is a reflection from a connector, splice or the end of the fiber. They can be used to mark distances or even calculate the "back reflection" of connectors or splices, another parameter we want to test in singlemode systems.

Test Setup

Generally, OTDRs are used for testing with a launch cable and may use a receive cable. The launch cable allows the OTDR to settle down after the test pulse is sent into the fiber and provides a reference connector for the first connector on the cable under test to determine its loss. A receive cable may be used on the far end to allow measurements of the connector on the end of the cable under test also.


Information in the OTDR Trace

​They say a picture is worth a thousand words, and the OTDR picture (or "trace" as they are called) takes a lot of words to describe all the information in it! Consider the trace in Figure 4.

Figure 4: Trace Information

The slope of the fiber trace shows the attenuation coefficient of the fiber and is calibrated in dB/km by the OTDR. In order to measure fiber attenuation, you need a fairly long length of fiber with no distortions on either end from the OTDR resolution or overloading due to large reflections. If the fiber looks nonlinear at either end, especially near a reflective event like a connector, avoid that section when measuring loss.

Connectors and splices are called "events" in OTDR jargon. Both should show a loss, but connectors and mechanical splices will also show a reflective peak. The height of that peak will indicate the amount of reflection at the event, unless it is so large that it saturates the OTDR receiver. The peak will have a flat top and tail on the far end, indicating the receiver was overloaded.

Sometimes, the loss of a good fusion splice will be too small to be seen by the OTDR. That's good for the system but can be confusing to the operator. It is very important to know the lengths of all fibers in the network, so you know where to look for events and won't get confused when unusual events show up (like ghosts, which we'll describe below.)

Reflective pulses can show you the resolution of the OTDR. You cannot see two events closer than is allowed by the pulse width. Generally longer pulse widths are used to be able to see farther along the cable plant and narrower pulses are used when high resolution is needed, although it limits the distance the OTDR can see.


Understanding the Physics (and Errors) of the Measurement

Don't let the title put you off, it's pretty basic. The amount of light scattered back to the OTDR for measurement is quite small, about one-millionth of what is in the test pulse, and it is not necessarily constant. This affects the operation and accuracy of OTDR measurements.


Overload Recovery

Since so little of the light comes back to the OTDR for analysis, the OTDR receiver circuit must be very sensitive. That means that big reflections, which may be one percent of the outgoing signal, will saturate the receiver, or overload it. Once saturated, the receiver requires some time to recover, and until it does, the trace is unreliable for measurement as shown in Figure 5 below.

The most common place you see this as a problem is caused by the connector on the OTDR itself. The reflection causes an overload which can take the equivalent of 50 meters to one kilometer (170 to 3000 feet) to recover fully, depending on the OTDR design, wavelength and magnitude of the reflection. It is usually called the "Dead Zone". For this reason, most OTDR manuals suggest using a "pulse suppresser" cable, which doesn't suppress pulses, but simply gives the OTDR time to recuperate before you start looking at the fiber in the cable plant you want to test. They should be called "launch" cables.
Figure 5: OTDR Launch Pulse and Launch Cable

Do not ever use an OTDR without this launch cable! You always want to see the beginning of the cable plant and you cannot do it without a launch cable. It allows the OTDR to settle down properly and gives you a chance to see the condition of the initial connector on the cable plant. It should be long, at least 500 to 1000 meters to be safe, and the connectors on it should be the best possible to reduce reflections. They must also match the connectors being tested, if they use any special polish techniques.


Ghosts
Figure 6: OTDR "Ghosts"

If you are testing short cables with highly reflective connectors, you will likely encounter "ghosts" like in Figure 6. These are caused by the reflected light from the far end connector reflecting back and forth in the fiber until it is attenuated to the noise level. Ghosts are very confusing, as they seem to be real reflective events like connectors, but will not show any loss. If you find a reflective event in the trace at a point where there is not supposed to be any connection, but the connection from the launch cable to the cable under test is highly reflective, look for ghosts at multiples of the length of the launch cable or the first cable you test. You can eliminate ghosts by reducing the reflections, using a trick we will share later.

On very short cables, multiple reflections can really confuse you! We once saw a cable that was tested with an OTDR and deemed bad because it was broken in the middle. In fact it was very short and the ghosted image made it look like a cable with a break in the middle. The tester had not looked at the distance scale or he would have noted the "break" was at 40 meters and the cable was only 40 meters long. The ghost at 80 meters looked like the end of the cable to him!


Backscatter Variability Errors

Another problem that occurs is a function of the backscatter coefficient, a big term which simply means the amount of light from the outgoing test pulse that is scattered back toward the OTDR. The OTDR looks at the returning signal and calculates loss based on the declining amount of light it sees coming back. Only about one-millionth of the light is scattered back for measurement, and that amount is not a constant. The backscattered light is a function of the attenuation of the fiber and the diameter of the core of the fiber. Higher attenuation fiber has more attenuation because the glass in it scatters more light. 

If you look at two different fibers connected together in an OTDR and try to measure splice or connector loss, you have a major source of error, the difference in backscattering from each fiber.
Figure 7: Loss Errors in OTDR Measurements

​To more easily understand this problem, consider Figure 7 showing two fibers connected. If both fibers are identical, such as splicing a broken fiber back together, the backscattering will be the same on both sides of the joint, so the OTDR will measure the actual splice loss. However, if the fibers are different, the backscatter coefficients will cause a different percentage of light to be sent back to the OTDR. If the first fiber has more loss than the one after the connection, the percentage of light from the OTDR test pulse will go down, so the measured loss on the OTDR will include the actual loss plus a loss error caused by the lower backscatter level, making the displayed loss greater than it actually is.

Looking the opposite way, from a low loss fiber to a high loss fiber, we find the backscatter goes up, making the measured loss less than it actually is. In fact, this often shows a "gainer", a major confusion to new OTDR users.

The difference in backscatter can be a major source of error. A difference in attenuation of 0.1 dB per km in the two fibers can lead to a splice loss error of 0.25 dB! While this error source is always present, it can be practically eliminated by taking readings both ways and averaging the measurements, and many OTDRs have this programmed in their measurement routines. This is the only way to test inline splices for loss and get accurate results.

Another common error can come from backscatter changes caused by variations in fiber diameter. A variation in diameter of 1% can cause a 0.1 dB variation in backscatter. This can cause tapered fibers to show higher attenuation in one direction, or we have in the past seen fiber with "waves" in the OTDR trace caused by manufacturing variations in the fiber diameter.


Overcoming Backscatter Errors

​One can overcome these variations in backscatter by measuring with the OTDR in both directions and averaging the losses. The errors in each direction cancel out, and the average value is close to the true value of the splice or connector loss. Although this invalidates the main selling point of the OTDR, that it can measure fiber from only one end, you can't change the laws of physics.


Resolution Limitations

​The next thing you must understand is OTDR resolution. The OTDR test pulse, as shown in Figure 8, has a long length in the fiber, typically 5 to 500 meters long (17 to 1700 feet). It cannot see features in the cable plant closer together than that, since the pulse will be going through both simultaneously. This has always been a problem with LANs or any cable plant with patchcords, as they disappear into the OTDR resolution. Thus two events close together can be measured as a single event, for example a connector that has a high loss stress bend near it will show up on the OTDR as one event with a total loss of both events. While it may lead you to think the connector is bad and try to replace it, the actual problem will remain.

Another place this problem shows up is in splice closures. An OTDR may show a bad splice, but it can actually be a crack or stress point somewhere else in the splice closure.

Figure 8: The OTDR Pulse Length limits its resolution

​There is a tool that will help here. A Visual Fault Locator as described earlier in this article, injects a bright red laser light into the fiber to find faults. If there is a high loss, such as a bad splice, connector or tight bend stressing the fiber, the light lost may be visible to the naked eye. This will find events close to the OTDR or close to another event that are not resolvable to the OTDR. Its limitation is distance too, it only works over a range of about 2.5 miles or 4 km.

The Visual Fault Locator is so valuable a tool that many OTDRs now have one built into them. If you are using an OTDR, you must have one to use it effectively.


Special Consideration for Multimode Fiber
Figure 9: OTDRs only see the middle of the Multimode Fiber Core

​Most OTDR measurements are made with singlemode fiber, since most outside plant cable is singlemode. But building and campus cabling are usually multimode fiber using light emitting diode sources for low and medium speed networks. The OTDR has problems with multimode fiber, since it uses a laser source to get the high power necessary to cause high enough backscatter levels to measure.

The laser light is transmitted by multimode fiber only in the center of the core (Figure 9) because its emission angle is so low. LEDs, however, are transmitted throughout the core of the multimode fiber, due to their wider radiation pattern. As a result of the OTDR light being concentrated in the center of the fiber, the loss of connectors is lower because the typical connector offset errors are not an effect. And even the fiber has lower loss, because the light in the center of the core travels a shorter path than the light at the outer edges of the core.

Several projects have tried to determine how to correlate OTDR measurements to source and power meter measurements, without success. Our experience is an OTDR will measure 6-7 dB of loss for a multimode cable plant that tests at 10 dB with a source and power meter.


Measuring Fiber, not Cable Distance

​And finally, OTDRs measure fiber not cable length. While this may sound obvious, it causes a lot of problems in buried cable. You see, to prevent stress on the fiber, cable manufacturers put about 1% more fiber in the cable than the length of the cable itself, to allow for some "stretch." If you measure with the OTDR at 1000 meters (3300 feet), the actual cable length is about 990 meters (3270 feet). If you are looking for a spot where the rats chewed through your cable, you could be digging 10 meters (33 feet) from the actual location!

​Auto-Test Programs

​Never simply attach an OTDR to the cable plant and hit the "auto-test" button! We know of applications where that was done that cost the installers and users big headaches - and bucks! ODTRs are not smart enough to make the decisions on setup and pass/fail themselves - they are easily fooled. If you do the setup correctly yourself, you can try "auto-test" and see if it gives reliable results, but never use it without knowledgeable operator oversight.

Stay Up To Date

Thanks for coming to use the services provided by logging in to the DINTEK website. Press the login button.

Once you are logged into the DINTEK Website. You will have access to additional content and services depending on the level of access that has been assigned to you.

Register