FOA Guide


Topic: Frequently Asked Questions About The

Optical Time Domain Reflectometer (OTDR)

Table of Contents:
The FOA Reference Guide To Fiber Optics


Frequently Asked Questions On OTDRS And Hints On Their Use

OTDRs, also known by their technical name optical time domain reflectometers, are valuable fiber optic testers when used properly, but improper use can be misleading and, in our experience, lead to expensive mistakes for the contractor. We have been personally involved in several instances where misapplication of OTDR testing has cost the contractor as much as $100,000 in wasted time and materials. Needless to say, it's extremely important to understand how to use these instruments correctly.


What Is The Difference Between OTDR Testing And Insertion Loss Testing?

An insertion loss test made with a light source and power meter is a simple test that is similar in principle to how a fiber optic link works. A light is placed on one end of the cable and a power meter measures loss at the other end, just like a link transmitter and receiver use the fiber for communications.
An OTDR, however, works like RADAR. It sends a pulse down the fiber and looks for a return signal from fiber backscatter and reflections from joints, creating a display called a "trace" or "signature" from the measurement of the fiber. From this trace, the OTDR calculates fiber length, attenuation and joint loss. So it does not “measure” loss directly, it implies it from the trace.


When is OTDR Testing Appropriate?

The answer to this question has several aspects.

Let’s start with “What are we trying to test?”

OSP: In a long outside plant cable with many splices, the OTDR is used to ensure that the cable has not been damaged during installation and each splice is properly made. Results are archived with other documentation to be available if restoration is necessary in the future. Later OTDR testing may be used for troubleshooting problems like finding locations of cable breaks caused by dig-ups. Generally, OSP networks are also tested with a LSPM (light source and power meter) also called an OLTS (optical loss test set.)
Premises: Premises cabling, however, has short cable runs and almost never includes splices, so the requirement of OTDR testing appears to be as an alternative to insertion loss testing with a light source and power meter, which in reality, is inappropriate.
We recommend that insertion loss testing be done even when OTDR testing is required by installation contracts. In our experience, OTDR testing of premises cabling systems is often required by users who do not really understand when OTDR testing is appropriate or even what an OTDR is. A knowledgeable contractor should be able to educate the user on proper testing requirements.
New international testing standards, however, accept both OTDR testing and insertion loss testing, even for short premises multimode cable plants. The differences in the measurement techniques used by OTDRs and a light source and power meter means that OTDR testing, especially on longer premises cable plants with higher loss, may not be comparable to measured insertion loss or the actual loss the communications system will experience.
The international standard that allowed OTDR as well as LSPM (light source and power meter) testing based that decision on tests performed on cable plants appropriate for 10G Ethernet which had losses of <2dB. In the real world, multimode cable plants in premises installations can have losses of 5-10 dB or more. OTDR tests will generally not correlate with LSPM tests (or often with other OTDRs!) which test the cable plant more like how the cable plant is used by communications equipment, and usually OTDR results are lower, setting up the network owner for a problem when the communications electronics are installed.
For this reason alone, we recommend that insertion loss testing be done even when OTDR testing is required by installation contracts.
Most of the technical calls we get at FOA regarding problems using OTDRs on premises cabling systems are caused by users who don't know what an OTDR is requiring them for testing their installation just because somebody told them to or they assume a bigger, more expensive instrument gives better data.

Next: What Can The OTDR “Test”?

OTDRs use an indirect measurement process, have poor length resolution and unique measurement errors that limit its accuracy in testing cable plants. It is not considered a replacement for insertion loss testing by knowledgeable fiber optic personnel.
From a more technical standpoint, the first and most important consideration for OTDR use is the length of the fibers to be tested. Most OTDRs are designed for long cable plants, especially singlemode OTDRs, and may be inappropriate for short cables. Some multimode OTDRs are now usable for short length multimode premises cables but only if they are properly set up before use and launch and receive cables have connections with low reflectance. Furthermore, on short cables or even relatively long ones with highly reflective events, “ghosts” caused by the high reflectance
OTDR measurement of joint lost are directional, dependent on the backscatter coefficient of the fiber, causing measurement to vary by 0.25 dB or more depending on direction.
If you are looking to test a cable plant to see if it will support a communications system’s loss budget, you do not want OTDR testing. If your network is short, the OTDR will not give you valid data.
The second most common tech question we get at the FOA is from people trying to use an OTDR when it’s inappropriate.

Do I Need Training To Use An OTDR?

"My OTDR manufacturer tells me its fully automatic and I just push a button and it gives me a PASS/FAIL result like my “Cat 5” tester. They say I don’t need any training."
Let’s just say that the majority of calls we get at the FOA involve OTDRs being used by people who are ignorant of their use, either trying to use them for cable plants that are too short or full of ghosts, their launch cables are too short, the setup wrong, or they simply don’t know how to interpret the OTDR trace. We have many examples including one instance where over $100,000 worth of cable was rejected and pulled out when it was perfectly OK but the OTDR user did not understand the trace. We have had calls from people trying to test 70m singlemode cables without a launch cable, MM cables with SM OTDRs and vice versa, and many more.
If you are using an OTDR without training, you are going to have big problems.

Why Do I Need A Launch Cable On The OTDR?
OTDRs are always used with a launch cable and may use a receive cable. The launch cable, sometimes also called a "pulse suppressor," has two major reasons for its use:
1. The launch cable allows the OTDR trace to settle down after the test pulse is sent into the fiber so you can analyze the beginning of the cable you are testing. The large event you see right in front of the instrument on the OTDR trace is caused by crosstalk within the instrument and reflectance from the connector on the face of the OTDR. The long recovery time from this overload pulse means the OTDR cannot make any useful measurements near the instrument itself. The launch cable has also been called a "pulse suppressor" because it allows time for the OTDR to settle down from this initial overload. If possible, singlemode OTDRs should have APC connectors on the front panel to reduce reflectance. Also a short connection cable attached to the OTDR before the launch cable that never gets removed from the OTDR prevents excess wear on the panel connector.
2, The launch cable  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.

What Is The Resolution In Length Of The OTDR?

Most OTDRs have a display range digitized to about 10-20,000 parts, so on a 20km range, the display resolution is 1m, or on a 2km range it would be 0.1m. The actual resolution of the OTDR is usually thought of as its ability to distinguish between two points on the cable, like intermediate patchcords or closely spaced splices. The actual resolution is determined by the width of the test pulse and the bandwidth of the OTDR receiver and is usually much longer than the display resolution. A 100ns pulse, for example, equals about 20m, but the depending on the shape of the test  pulse, the OTDR may not be able to distinguish events 2-3 times that length.
See the demonstration below for a way to prove this to yourself.

I Have Heard The OTDR Measures Fiber Length, Not Cable Length. How Do I Correct For That?

First, what are the sources of error? The OTDR uses the speed of light in the fiber (from the index of refraction) to calculate the length of the fiber. Also, most cables have 1-2% excess fiber (less on ribbon cables) to prevent fiber stress under cable tension. Some manufacturers of cable can provide that information for your testing. If you do not know the index of refraction/velocity of propagation or the ratio of excess fiber, you can estimate it or, if you have a long spool of cable, calibrate it.
Just measure the fiber on the spool of cable with the OTDR, then look at the cable jacket for length markings to get the actual length of the cable from the printed markings at each end of the cable. Use the OTDR’s calibration feature to set the index of refraction to the value that makes the OTDR read the same as the marked length of the cable.

Directional Results Can Be Confusing: I am testing a cable with OTDR. I have a limit 0.2 db loss per splice. I use bi directional analysis. In some fibers from A>B direction i have 0.25 loss but in B>A it doesnt show up that splice at all. I changed the pulse width but nothing happened. Any ideas?
You are seeing the directional differences. For a splice with 0.25 dB loss in one direction and 0 dB in the other, the average is (0.25+0)/2 = 0.13 dB loss. If you shoot in both directions and overlay, the software should recognize that there should be events in both directions, input a "0 dB" event and average accordingly. Most OTDRs also allow setting a threshold for detection of events and that must be set correctly to recognize events. There are many times a splice is undetectable in an OTDR trace due to good splices and the simple fact that the OTDR measurement technique itself is limited.

How can we differentiate a ghost from a real event?
A ghost will not have any loss, it will be at equal distance from a highly reflective event (look for repetition), tends to be in the middle of noise after the end of the cable.

What are good values to set a OTDR to for PASS/FAIL?
Splice threshold
Reflectance threshold
Slope threshold (slope is attenuation coefficient)
End threshold (depends on whether you 1) use receive reference cable which would be a normal connection loss or 2)the length of the cable and the noise floor of the measurement. Best to make sure the trace is not noisy to the end and have 2-3dB from the cable backscatter level to the noise floor.



Application

Splice

Connector

Reflectance

Attenuation

Singlemode long distance (>5-10km)

<0.1dB (fusion)

<0.3dB

>-40dB (that means -41dB or more)

0.4dB/km at 1310nm, 0.25dB/km at 1550nm

SM short links

<0.1dB (fusion)

<0.3dB

>-50dB (that means -51dB or more)

0.4dB/km at 1310nm, 0.25dB/km at 1550nm

MM

0.3dB (fusion or mechanical)

<0.5dB

>-30-35dB

3.0 dB/km at 850nm, 1dB/km at 1310nm

 


Uncertainty of OTDR Test Results
We received a call from a contractor who had tested a cable plant for an end user using an OTDR. The user had several others do the same test and got different results, not widely different, but different enough to make him wonder why. The same thing happens with OLTS testing too, but for slightly different reasons. This conversation inspired a short tutorial which follows:
Two Types of Measurement Errors
    Measuring a physical parameter always involves errors. Unfortunately you never know the real value to begin with, so all you can do is to estimate the error based on the possible sources of error in making the measurement. There are two types of errors, random and systematic.
    Random errors are what you see when you make a measurement multiple times and get a slightly different value each time. Hook up your instrument and make the measurement, disconnect and try again. Each time, the result will be slightly different. Generally one should make several measurements, average them and use the data to calculate the random error, called standard deviation, to understand the uncertainty of the measurement.
    Systematic errors are how the average measurement differs from the real value, which can be caused by some problem in setup or calibration. Unfortunately, it’s hard to determine the systematic error, but some possible ways exist sometimes.

Let’s look at  OTDR measurement uncertainty from both a random and systematic viewpoint.

Random Errors
    Testing loss with an OTDR requires a launch cable that connects to the fiber in the cable under test, taking a trace and analyzing the trace, either manually or using some preprogrammed auto-test function. This leads to several random errors in loss measurement which may include:
1.    Variation in loss of the connection of the launch cable to the cable under test caused by alignment variations each time it’s connected, dirt, temperature, etc.
2.    Changes in stress of the launch cable or cable under test which can be caused by temperature variations or physical movement of the cable.
3.    Changes in the mode power distribution of launched pulses  which can affect both multimode and singlemode cables (short SM may not be single-mode-it may take hundreds of meters!)
4.    Noise in the OTDR trace, with the effect greater effect with less averaging.
5.    How the user sets the markers on the trace for each measurement. This is affected by pulse width (risetime) and the reflectance from an event which can overload the OTDR and cause difficulties in determining where the fiber baseline is located.

Systematic Errors
    When you set up the OTDR, you have to make certain set-up decisions, including range, wavelength, fiber glass index of refraction, pulse width, number of averages, etc. that affect the measurement uncertainty for every measurement.
Length Measurement
1.    The range may affect the time base of the OTDR which is used (along with index of refraction) to calculate the length of the fiber.
2.    The index of refraction (n) is a direct expression of the speed of light in the fiber (V=C/n). Distance is calculated as “time X speed.” Most OTDR users use a generic value, but sometimes the actual value for the fiber being tested is known.
3.    Each cable has excess fiber, typically ~1%, to prevent stressing the fiber when pulled. The OTDR measures the length of the fiber, not the cable. It can be calibrated for the cable under test if one knows the length of the cable and uses that to calculate a cable-specific index of refraction.
4.    The pulse width may cause systematic errors in the measurement of length, since wider pulses have longer risetimes which make placing the markers consistently more difficult.
Loss
1.    Setting markers for measuring loss is affected by the width of the test pulse. Longer pulses have longer risetimes which make setting markers consistently more difficult. Wider pulses cause greater reflectance from connectors and affect both the shape of the reflected pulse and the shape of the return to the fiber baseline, causing uncertainty on how the markers are set. Noisy traces are wider, which can lead to systematic errors.
2.    Manually setting markers generally will introduce random errors, as the operator sets their location each time, but can introduce systematic errors due to the way the operator typically works.
3.    Auto-test programming may introduce systematic errors depending on the pulse width, reflectance, range, averaging, etc. Generally auto-test should not be used until it has been compared to manual analysis.
4.    Connectors on different launch or receive cables will change the measurements systematically.
5.    The length of the launch cable may affect SM or multimode measurements. A SM launch cable should be 500-1000 m long to ensure the test signals are singlemode. Multimode fiber will change mode power distribution with length.
6.    Any mode conditioning on a MM cable (e.g. mandrel wraps) will affect the modal conditioning on the downstream part of the test where the test pulse from the OTDR goes out from the OTDR. On the return backscattered light, the fiber modes will be fully filled.
7.    Instrument calibration will cause systematic errors. Few users ever calibrate OTDRs, but the time base and linearity of the amplifiers can affect the measurements.

Uncertainty of Results
So what can you expect? Length may vary by several percent on different OTDRs. Loss can vary by several tenths of a dB on short lengths and proportionally more on long cable plants for different OTDRs and at least as much with the

I’m Troubleshooting A Break In A Long Cable Run But I Don’t Know The Correction Factor For Fiber Vs Cable Length. What Can I Do?

Here is a perfect example of why you need cable plant documentation. If you have the data from the original design and testing, you should have the actual length of the cable plant. With that you can calculate the point of the break very closely. Here is an example:

Let’s say we have a 10km cable with a break around 6km from one end. From one end, the OTDR says the distance to the break is 6500m and from the other end it says it’s 4000m.


Total length of OTDR cable length: 6500+4000=10,500m

If the actual cable length is 10,000 m, the correction factor is:

Actual length/measured length = 10000/10500 = 0.952 = correction factor


Thus our 6500m measurement is actually 6500X0.952 = 6190m and from the other end it’s 3810m.


If you do not have documentation, try to calibrate the OTDR using a section where you can get real length data from the cable jacket.


Sometimes My Traces Show Big Reflections From The End Of The Cable But Sometimes It Shows None At All. Why?

The reflection on the end of the cable depends on the end of the fiber. If it’s broken and ragged, you will see practically no reflection, but a perfectly cleaved fiber will show a giant reflection peak.
Look at these three traces:
 
Cleaved fiber
Note how the cleaved fiber has a high reflectance, reaching saturation on the OTDR trace

 
broken fiber
The broken trace shows a small reflectance.

shattered fiber 
The shattered fiber shows virtually no reflectance.


How are OTDRs Calibrated? 

Calibration of OTDRs is a messy issue. There are many variables.

You can purchase OTDR calibration artifacts for calibrating your OTDR but as far as I know, they are generally not traceable to national standards labs. Using fibers to calibrate an OTDR introduces errors.

Two parameters of the OTDR need calibration: dB and length.

Calibration of the dB scale, used for measuring loss and attenuation coefficient (which is also dependent on length calibration) is complicated by the way the instrument is used. For loss, the measurement is very low in magnitude (~0.1dB) but fine in resolution (as low as 0.001dB), so nonlinearities on a small scale along the entire measurement range are the issue. Proper calibration would include the linearity of the entire measurement range which is virtually unknown.

For attenuation coefficient, the dB measurement is over a longer range and can be done with a calibrated artifact. But that calibration is wavelength dependent.
For length, it's a matter of time measurement in the OTDR - distance is calibrated from the index of refraction of the fiber or the group velocity of the test pulse in the fiber. This can be done with a calibration artifact - a fiber of known length and index of refraction - but again the calibration is wavelength sensitive.

You can get calibration artifacts from NPL in the UK, but you need to know the calibration wavelength of your OTDR and the characteristics of their artifact to make corrections.

Another method uses a electronic calibrator - take the pulse from the OTDR and trigger a delayed return pulse to calibrate the distance scale and a optical ramp to simulate the attenuation of a fiber. This removes the unknowns associated with using a fiber and has been championed by many scientific types.
Any OTDR manufacturers want to offer their wisdom?


Our thanks to FOA Master instructor Terry O’Malley (
http://www.fiberopticsolutions.biz/) for his advice and work creating the sample traces and the following exercise.



Demonstrating OTDR Length Measurement Capability

This testing exercise demonstrates that the OTDR is extremely accurate “unto itself”. That is; not in actual fiber length (IOR dependent) and defiantly not in sheath length but it has some important usage when it comes to troubleshooting.

OTDR  GNNettest 4000
Settings
Range 6kft
IOR 1.464
PW  10ns

Fiber Sections
Reel  1=   <600 ft.
Reel  2=  >1000 ft.
Reel  3=  <1000 ft.

TEST 1 & 2
When the reels are connected consecutively (1-2-3) the distance to the end Fresnel is within two (2) feet from either direction. Demonstrates the repeatability of distance measurements.

TEST 3
All three sections of fiber measured exactly the same length from both ends.
Demonstrates the repeatability of distance measurements.

TEST 4
Cutting off the far end fiber in 1 inch sections. On the third 1 inch cut off ( a total of 3 inches) the frenel jumped back towards the OTDR test end one foot. On the second set of fiber cuts it required 8 one (1 ) inch cuts to get “behind” a sampling point to again move the Fresnel back one sampling point in distance.
This demonstrates the resolution as it relates to sampling points and distance accuracy.

TEST 5
A ten (10) foot cords was attached to the far end and the distance reading to the far end remained the same as in test 1. 
This demonstrates that the 10 foot cord was “hidden” in the recovery of fiber reel 3’s end termini reflectance.

TEST 6
The 10 foot cord was then connected to the OTDR test cord at the near end. The system under test measured an additional 10ft exactly.
This demonstrates that the 10 foot cord length could be measured if not hidden in the end reflectance.



Want to learn more? Try the FOA Self-Study Program on OTDRs at Fiber U.

Return To The FOA OTDR Tutorial
Download the free FOA OTDR PC Simulator To Learn How To Use An OTDR
See the FOA Virtual Hands-On OTDR Tutorial
Return to the FOA Online Reference Guide Table of Contents  

Return to the FOA Guide Table of Contents

Return to the FOA home page 




 


(C)2008-2012, The Fiber Optic Association, Inc.