pdfcrowd.com open in browser PRO version Are you a developer?

Try out the HTML to PDF API

The Fiber Optic Association - Tech Topics
FAQs: Fiber Optic Testing

We get questions all the time, so we have started putting those we feel have general interest on this
web page. As it grows larger, we'll index it for convenience. Here's more general FAQs too.

Looking for more information? Try the FOA Online Reference Guide on Testing or Tech Topics
index

Topics covered in this FAQ:
FO Power Measurements
Fiber and Cable Loss
OTDRs
Bandwidth
Reflectance/Optical Return Loss
Fiber Optic Cleaning Procedures
pdfcrowd.com open in browser PRO version Are you a developer? Try out the HTML to PDF API
Optical Power Measurements
Standard: FOTP-95
More on power measurements.

What are the measurement units for power?
Optical power is measured in linear units of milliwatts (mW), microwatts (uW - really the greek letter
"mu"W), nanowatts (nW) and decibels (dB).
What is the difference between "dBm" and "dB"?
dB is a ratio of two powers, for example the loss in a fiber optic cable. When power is measured in
linear units (mW, uW or nW), dB is calculated on a log scale using this formula:

power (dB) = 10 log (power1/power2)

If we are measuring absolute power levels, the measurement is generally referenced to 1 milliwatt
(mW), is expressed as "dBm" and the calculation becomes:

power (dBm) = 10 log (power/1 mW)

Thus 1 mW = 0 dBm, 1 uW = -30 dBm, 1 nW = -60 dBm and two equal powers compared are 0dB
(eg. power being the same, there is no loss.)

What power level should a source have?
It depends on the type of source. When coupled into a good test cable, the source output power will
be in these ranges:
pdfcrowd.com open in browser PRO version Are you a developer? Try out the HTML to PDF API
LED: -10 to -25 dBm into 62.5/125 fiber
Telcom/LAN laser: 0 to -13 dBm into singlemode fiber, to +20 dBm with DWDM and fiber amplifier
systems
CATV Laser : +16 to 0 dBm into singlemode fiber
What power level should a receiver see?
It depends on the network and type of source. When measured at the end of the network cable, the
source output power will usually be in these ranges:
LAN/LED: -20 to -35 dBm into 62.5/125 fiber
Telcom/LAN laser: -20 to -45 dBm into singlemode fiber
CATV Laser : 0 to -10 dBm into singlemode fiber

How do you calculate a loss budget?
The loss budget is a calculation of how much attenuation a link should have. You compare that loss
to the dynamic range of the networking equipment to see if the range and link loss are compatible.
How accurate are fiber optic power meters?
All optical power meters which are calibrated to NIST (the US standards body) or any national
standards lab will measure optical power to an uncertainty of about +/- 0.2 dB or 5%. Therefore,
since every power meter has an uncertainty of +/- 0.2 dB, any two meters can differ by 0.4 dB in the
worst case (one at +0.2 dB and one at -0.2 dB) even if both are within their specification!
More information on calibration uncertainty.

Are more complex or higher priced FO power meters more accurate?
The high priced meters offer better dynamic range and more features, but not better absolute
measurement uncertainty.
pdfcrowd.com open in browser PRO version Are you a developer? Try out the HTML to PDF API
Why is the measurement uncertainty so high? That is because there are three to four calibration
transfers from the NIST absolute optical power standard before the meter is delivered to the
customer. The NIST standard has an uncertainty of about 1% itself and every transfer adds errors of
about 1%. More information on calibration uncertainty.

Why do most meters only offer calibrations at a few wavelengths?
NIST only offers calibrations at 850, 1300 and 1550 nm, so those meters that have calibrations at
other wavelengths have to extrapolate to those values, increasing the measurement uncertainty at
those wavelengths.
If my source is at a slightly different wavelength from the standard calibration wavelength, doesn't
that add to measurement error?
Perhaps, but the wavelength of most sources is not known by the person making the measurement.
If everyone uses meters calibrated at only a few specific wavelengths, everyone can be testing to
the same standard and will get more closely correlated mesurements on sources of unknown
wavelengths .
Fiber Optic Cable Loss (Insertion Loss With Light Source and Power Meter
Standard: FOTP-171 for cable assemblies
Standard: OFSTP-14 for the installed multimode cable plant, OFSTP-7 for the installed singlemode
cable plant
As of 2011, OFSTP-14 has been replaced by an international standard that as of mid-2011
is very controversial. First of all, it allows the use of either insertion loss testing with a light
source and power meter or OTDR testing. This was predicated on comparisons of OTDR
tests on cable plants for 10GbE of less than 2 dB loss. For other more typical multimode
links of 5-10 dB, the two methods will generally give divergent results. Since insertion loss
pdfcrowd.com open in browser PRO version Are you a developer? Try out the HTML to PDF API
is designed to test according to how the link will be used, it should be the test used for
longer links. The new standard also includes a new metric for measuring mode power
distribution in multimode fiber called Encircled Flux (EF). It is also controversial, but
international documents say the older source with mandrel wrap meets their requirements.
Here is a more detailed explanation of all the options in cable testing.

Why do you use a launch cable on the source?
You use a launch cable to set the proper test conditions for testing another cable. The launch cable
should match the fiber size and connector type of the cable you want to test, and be tested to insure
it is a low loss connector.

Why can't I just attach my cable I'm testing directly to the source?
Sources have a great deal of variety in how they launch light into the cable, which can cause
undesirable variations in loss measurements. Furthermore, the coupled power can vary considerably
with each insertion, depending on the alignment of the connector ferrule in the source output
connector.

What makes a launch or receive cable "good"?
A good launch or receive cable will have low loss - less than 0.5 dB loss when tested in a single-
ended FOTP-171 test.

Do I always need a laser source to test singlemode?
No, you can use a LED source for short SM patchcords or cables up to about 5 km long. Longer SM
cables will show higher loss with an LED due to the spectral width of the LED causing higher loss at
the upper and lower ends of its spectral output. FOTP-171 actualy calls for LED sources to test SM
patchcords to prevent problems with the interference caused by the coherent light of the laser.
pdfcrowd.com open in browser PRO version Are you a developer? Try out the HTML to PDF API

What is a receive cable?
The receive cable is used in a double ended test to measure the connector loss on both ends of the
cable.

When doing a double-ended loss test, why don't you set the reference with both launch and
receive cables connected together?
There are several reasons why you use the same reference method for single-ended and double-
ended tests.
If you want to measure the loss of connectors on both ends of the cable being tested, you need a
reference with the launch cable only.
A two cable reference removes one connector loss from the measurement, since you include it in the
reference. You can also reference with three cables, where you simply replace the middle cable with
the cable under test, but the loss removes two connectors from the test value, as they are included
in the reference. However all three methods are approved standard methods and the three cable
reference is sometimes the only way you can test cable plants with connectors like the MT-RJ that
cannot be directly connected to test instruments. Here is a more detailed explanation of all the
options in cable testing.
What happens if the launch and/or receive cables have bad connectors with say 3 dB loss.
When the zero reference is set, it will include the loss of two bad connectors. When you attach a
cable between them to test, you will measure erroneously high loss for one or both connectors on
the cable being tested, invalidating the measurement, so it is very important to keep all test
connectors clean.
How do I test cables with different connector types?
Use a two or three cable reference as necessary. More.
pdfcrowd.com open in browser PRO version Are you a developer? Try out the HTML to PDF API
What should my cable loss be? Are my measurements correct?
A basic guideline for loss measurements is to estimate the loss using these approximate values:
Connector loss: 0.5 dB per mated pair
Splice loss: 0.2 dB per splice
Fiber loss:
Multimode: 3 dB/km @ 850 nm, 1 dB/km @ 1300 nm
Singlemode: 0.4 dB/km @ 1300 nm, 0.3 dB/km @ 1550 nm
For example: a 1.5 km multimode link with 3 connections @ 850 nm:
Loss = (1.5 km X 3 dB/km) + (3 conn X 0.5 dB)
Loss = 4.5 dB + 1.5 dB = 6 dB

You should also read this more detailed explanation and this page about "loss budgets".
How Long a Fiber Can My Meter and Source Test?
That depends on the output power of the source and the sensitivity of the meter. For example, one
of our LED sources will have a maximum output into 62.5/125 fiber of about -15dBm. Your meter
should be used at power levels above about 10 dB higher than its minimum spec. A meter can easily
read to -45 dBm (min spec is -55 dBm), giving us a range of 30 dB (-45 dBm from -15 dBm gives us
30 dB). At 850 nm and a loss of 3 dB/km, that's 10 km of fiber, less our connector and splice loss,
and at 1300 nm,a loss of 1 dB/km, it's 30 km less connector and splice loss, both lots longer than
any networks operating on multimode fiber.
For singlemode testing, lasers can give you 0 to -10 dBm output, giving a range of 35 to 55 dB,
corresponding to over 100 km of fiber, even approaching 200 km at 1550 nm!
How Can An Insertion Loss Test Show "Gain", Not Loss!
When an insertion loss test shows gain not a loss, it is usually a problem with setting the "0dB" loss.
pdfcrowd.com open in browser PRO version Are you a developer? Try out the HTML to PDF API
If the reference cables are dirty when setting the "0dB" reference and then cleaned before testing (or
the dirt falls off), the measurement may show a positive gain not a negative loss measurement. It's
also possible it may indicate a problem with the instrument - source drifting
Optical Time Domain Reflectometers (OTDRs)
Reference: Understanding OTDRs, Also see OTDR FAQs

Can't I use an OTDR to test cable loss?
Well, Yes and No. The OTDR will measure the loss in the cable plant, but using a technique based
on backscatter signals that indirectly measure loss, unlike a source and power meter that measure
loss directly. The OTDR measurement technique doesn't correllate well with the source and meter.
Since the source and meter tests loss just like the transmission link, all standard cable plant tests
specify using a source and meter to measure loss. You must always use them to test the cable plant
loss. Even the outside plant singlemode test standard, OFSTP-7, says you should accept only the
source/meter results.

Why do I use an OTDR?
Use the OTDR for troubleshooting. If you have a cable break, especially in the outside plant, the
OTDR is the best way to find it. You can also use it to verify splice loss (but test both ways and
average to get a reliable measurement) or find problems with back reflection (optical return loss).

How do I see close features with an OTDR?
The blind spot of an OTDR caused by crosstalk from the test pulse can be overcome by using a
pdfcrowd.com open in browser PRO version Are you a developer? Try out the HTML to PDF API
"pulse suppressor", a long (1 km is normal) length of cable to allow the OTDR to settle down after
the initial pulse.

Bandwidth Measurements

Do I need to test bandwidth?
Generally no. Most systems are specified for use with a minimum bandwidth fiber and most fiber is
much better than minimum specification. Besides test equipment is not cost effective or readily
available for field use. Manufacturers of fiber and cables have the expensive lab equipment to
reliably test bandwidth (or actually dispersion), but there are no good field testers. If you need
bandwidth data for an unusual application, ask the manufacturer of the fiber or cable to assist you or
use a simulation program, which are available from some manufacturers of fiber.
One exception is long distance networks which need testing for chromatic dispersion (CD) and
polarization mode dispersion (PMD). Here is more on CD and PMD.

Reflectance and Optical Return Loss
Standard: FOTP-107
More on Reflectance Testing
What is Reflectance and Optical Return Loss?
Reflectance is the light reflected back from a connector or splice. Optical Return Loss (ORL) is
generally used to combine the reflectance from connectors or splices with the backscatter from the
pdfcrowd.com open in browser PRO version Are you a developer? Try out the HTML to PDF API
fiber, so the term is primarily used for longer cable runs. Reflectance was once called back
reflection, but that term, which is really redundant, has lost favor.

When do I need to test optical return loss?
Reflectance or optical return loss mostly affects very high bitrate digital or analog singlemode
systems. None of today's multimode systems are very sensitive to reflectance or ORL although high
amounts can create background noise in short links adversely affecting BER or data transfer. For
affecting laser sources, ORL is only important in the first few connectors in the cable nearest the
laser transmitter. But some short SM cable plants used in premises systems are too short to
attenuate the reflected power from connectors so undergoes multiple reflections until it builds up
background noise that cab affect receivers.

How do I test optical return loss?
Use an OTDR on cable plants, OCWR on patchcords. ORL testing with what people call a ORL
tester ( or what Telcordia/Bellcore calls an OCWR or optical continuous wave reflectomenter) is only
applicable to short patchcords. If you try using them on an installed cable plant, the integrated
backscatter from the length of the fiber will overwhelm the connector back reflection. Twenty km of
fiber gives about the same backscatter as a connector with 20 dB ORL and you cannot tell where
the backscatter comes from! It's better to use an OTDR to find ORL problems on installed cable
plant.

How accurate are ORL measurements?
The measurement uncertainty of ORL is very high, about +/-1 dB for singlemode and +/-5dB for
multimode, according to round robin results from standards committees. This is a function of the
difficulty in creating a reference for the measurement, the reflection is very small compared to the
test signal causing noise problems and the dependence of the measurement on the connector on
pdfcrowd.com open in browser PRO version Are you a developer? Try out the HTML to PDF API
the test apparatus is very high. To minimize uncertainty, keep the connections extremely clean and
inspect the connectors continuously with a microscope and repolish when needed. Making
measurements to a 0.01 dB resolution is ridiculous; remember this is a +/- 1 dB measurement.

Do I need special instruments to test ORL?
No, the special ORL tester is unnecessary. A good laser source and power meter, along with a
coupler that costs a few hundred dollars will make a very good tester.
Note: Most instructions for using OCWRs suggest using a mandrel wrap to reduce the
reflectance from the connector on the end of the cable. If the cable has bend-insensitive
fiber, as do many patchcords, this method does not work. Instead dip the end connector in
index matching gel or fluid (vaseline, alcohol or mineral oil works in a pinch.)
Fiber Optic Cleaning Procedures
More on Cleaning

How do you clean connectors for testing?
With fiber optics, our tolerance to dirt is near zero. Airborne particles are about the size of the core of
SM fiber and are ususlly silica based- they may scratch PC connectors if not removed! Test
equipment that has fiber-bulkhead outputs need periodic cleaning, since they may have hundreds of
insertions of test cables in short time frames. Here's a summary of what we have learned.
1. Always keep protective "dust caps" on connectors, bulkhead splices, patch panels or anything
else that is going to have a connection made with it. Dust caps themselves may contain dust so
pdfcrowd.com open in browser PRO version Are you a developer? Try out the HTML to PDF API
whenever a connector is to be used, clean it.
2. Use any of the commercial cleaning kits to clean connectors and mating adapters. Alternatively,
use lint free pads and isoproply alcohol to clean the connectors. Some solvents MIGHT attack
epoxy, so only pure alcohol should be used. Cotton swabs and cloth leave threads behind. Some
optical cleaners leave residues. Residues usually attract dirt and make it stick.
3. All "canned air" has a liquid propellant. Years ago, you could buy a can of plain dry nitrogen to
blow things out with, but it's long gone. Today's aerosol cleaners use non-CFC propellant and will
leave a residue unless you 1. hold them perfectly level when spraying and 2. spray for 3-5 seconds
before using to insure that any liquid propellant is expelled from the nozzle. These cans can be used
to blow dust out of bulkheads with a connector in the other side or an active device mount
(xmit/rcvr). NEVER use compressed air from a hose (they emit a fine spray of oil from the
compressor!) or blow on them (you breath is full of moisture , not to mention all those yukky germs!)
4. A better way to clean these bulkheads is to remove both connectors and clean with Alco Pads,
then use a swab made of the same material with alcohol on it to clean out the bulkhead.
5. Detectors on FO power meters should also be cleaned with the AlcoPads occasionally to remove
dirt. Take the connector adapter off and wipe the surface, then air dry.
6. Ferrules on the connectors/cables used for testing will get dirty by scraping off the material of the
alignment sleeve in the splice bushing. Some of these sleeves are molded glass-filled thermoplastic
and sold for multimode applications. These will give you a dirty connector ferrule in 10 insertions!
You can see the front edge of the connector ferrule getting black! The alignment sleeve will build up
an internal ledge and create a gap between the mating ferrules - viola: a 1-2 dB attenuator! Use the
metal or ceramic alignment sleeve bulkheads only if you are expecting repeated insertions. Cleaning
the above reguires agressive scrubbing on the ferrules with the AlcoPad and tossing the bulkhead
away.
7. You can buy a cleaning kit for fiber optics. They are good solutions but perhaps not as cost
effective as making your own to meet your needs.
pdfcrowd.com open in browser PRO version Are you a developer? Try out the HTML to PDF API


(C) 2002-14, The Fiber Optic Association, Inc.
Return To The FOA Home Page
Return To FOA Tech Topics