Loss testing for MPO and MTP systems

Large scale deployment of fiber optic systems has recently stated using multimode ribbon fiber and typically MPO/MTP connectors, most commonly in large data centres.

Here we give an overview of how to tackle various issues related to testing the cabling for these systems. We mainly discuss multimode systems, however single mode systems share very many common practices.

Typically, these data centre applications are for 40G / 100G Ethernet. Cabling may be installed for 100G Ethernet even if the immediate requirement is 40G, to allow for later incremental bandwidth upgrades. The 24 fibers used for 100G Ethernet can alternatively be used to carry three 40G systems. So this allows a choice of upgrade paths, eg to incrementally add extra 40G channels, or replace with a single 100G channel. Standards are also being developed for 400G Ethernet.

The incremental cost of installing 24 fiber 100G cabling, instead of 12 fiber 40G cabling, is actually quite modest, particularly when compared to later re-wiring and down time. The major additional complexity may be a patch panel to give adequate flexibility for future deployment options.

Standards compliance

Internationally recognised standards are now in place, which define just about everything to do with data centre design.

Just in the area of fiber optic cabling, there are standards for fiber types (Typically multimode OM3. OM4 etc), cable types (fire retardance, bend resistance, riser cables etc), cable installation (fire stops, ducting etc) connectors (LC, MPO etc), labelling conventions, Ethernet, optical wavelengths (CWDM etc), optical safety (mandatory), and finally: cable acceptance testing.

There is one aspect of fiber test standards that everyone can agree on: they are currently a work in progress, and this is particularly true of MPO testing.

So at the project definition stage, it is a good idea to review and refine the fiber cable acceptance test requirements, which sensibly end up as a combination of standards based requirements and the mandatory requirements of the particular transmission equipment in use.

A very small acceptable optical loss!

The majority of LAN transmission systems using MPO/MTP connectors have extremely tight end-to-end optical loss requirements, perhaps around 1.2 dB. This tiny loss figure creates unique challenges for precision loss measuring for cable certification. Much of the currently available test equipment and procedures are incapable of properly achieving the required repeatability (accuracy) to give confidence that the specification has been achieved.

The two reasons for this tight specification are to minimise per port electronics cost, and because these systems are dispersion limited, not power limited. These systems use 850 nm and 50u core fiber, and end to end length is usually limited (by dispersion) to around 300 meters.

Inspection & cleaning

Connector condition and cleanliness is a very major consideration, and is the most common cause of problems. Absolutely essential are a good quality microscope with an MPO/MPT specific adaptor, a supply of suitable specific cleaning materials (more than one type is good), and adequate time and know-how to use them. The default available MPO cleaners tend to do a reasonable job of cleaning the fiber ends only, but extra cleaning materials should be available to clean alignment pins, and sometimes the whole connector end face.

Inspection and cleaning should be performed every time before a connector is mated.

Cabling disturbance

Because of the combination of very low loss requirements, and fiber complexity, it is essential that testing and commissioning are accomplished with the lowest possible level of disturbance to connections. Also, once loss testing is done, any disturbance may require the loss testing to be repeated. This means that an unusually strict test regime is appropriate, so that each acceptance level is completed with exceptional confidence. You don’t want to go back in and disturb things.

Continuity testing

This is the most basic test: Does light get from end to end? As a minimum, this test is easily accomplished with a VFL source and a one low quality breakout cable. If a numbered breakout cable is used each end, this could also double as a polarity test phase. Continuity testing usually doesn’t need documenting.

The continuity test phase is usefully accompanied by thorough cleaning and inspection of the connectors with a microscope,. This achieves a few objectives. Bad connectors are the main cause of installation failures. Doing this here enables the installer to identify & re-work bad connectors at the earliest (lowest cost) point. It also ensures that a bad connector does not contaminate the test leads and degrade other connectors.

If the connector quality is inadequate, there is little point proceeding to the next stages until this issue is fixed.

Simple continuity testing has one big problem: it doesn’t identify if fibers are swapped back-to-front. So for this we talk of:

Polarity testing

Typically, the cable installation requires polarity testing, to determine that the correct 1 – n arrays of fiber at one end of a system, are mapped to the correct 1 – n arrays of fiber the other end. If the installation has a patching cable section so that connections can be split off to particular transmission equipment, niggling polarity mistakes can occur here. The polarity test needs a very high level of confidence, since if not 100% correct, there will be no chance of a working outcome, and a great deal of wasted work to find the swapped fibers. This will probably want a basic level of documenting, sufficient to record which fibers where tested, and the general array direction.

Labelling

You now have your system physically in place and light going to and fro in the right combinations.

This is the perfect time to check and rectify so that all associated labelling is 100% correct; otherwise the next phase of loss testing could result in incorrect re-assembly, and non-working systems.

Up to this point, the work has been quite straightforward, however a very methodical approach is needed.

Loss testing accuracy & confidence issues

People use the terms repeatability, accuracy, reproducibility or uncertainty somewhat interchangeably. For loss testing in particular, there is little practical difference between them all, so we will use the term “uncertainty”. The real issue is what uncertainty happens when repeatedly measuring one item with different items of test equipment or test cords.

Quoted uncertainty values always imply a ± value, whether specifically mentioned or not. For example, a quoted uncertainty of 0.1dB is the same as ±0.1dB, usually with a 95% confidence level.

As mentioned, the acceptable maximum loss figure may only be 1.2 dB. If the test uncertainty is say 0.3 dB, then all measured losses above 0.9 dB would have to be considered marginal. So every small improvement in the uncertainty figure, yields a usefully larger maximum loss level, which will reduce installation cost. This easily explains the previous care on planning test leads, doing cleaning etc. In practice, accuracy is optimised by:

  1. Use of low loss Elite test leads, and ensuring they are in top condition.
  2. Use of an Encircled Flux compliant test source, followed by a mandrel wrap.
  3. Use of very accurate loss test equipment, remembering that you are working on the limits of achievable loss test accuracy.
  4. Rigorous inspection and cleaning every time a connector is mated
  5. Performing test lead verification at the start of the job
  6. Use of appropriate data recording systems, since there will be plenty of test data.

Test cords

A major loss test consideration is the test leads. “Elite” MPO / MTP leads are available with per connector / fiber loss of below 0.35 dB. Otherwise with “standard” leads are clearly inadequate for testing. Here are the USConnec specifications for these connectors.

  MM MT Elite®
Multimode MT
Ferrule
Standard
Multimode MT
Ferrule
SM MT Elite®
Single-mode MT
Ferrule
Standard
Single-mode MT
Ferrule
Insertion Loss 0.1 dB Typical (All Fibers)
0.35 dB Maximum
(Single Fiber)
0.20 dB Typical (All Fibers)
0.60 dB Maximum
(Single Fiber)
0.10 dB Typical (All Fibers)
0.35 dB Maximum
(Single Fiber)
0.25 dB Typical (All Fibers)
0.75 dB Maximum
(Single Fiber)
Optical Return Loss > 20 dB > 20 dB > 60 dB (8° Angle Polish) > 60 dB (8° Angle Polish)

Testing will probably need a combination of straight patch leads, and breakout leads. Also ensure that connector pin polarities are correct, since trying to swap these pins in the field is not viable. A few tested through connectors is also helpful.

Spare leads will be required, in case one gets degraded or damaged.

The typical application environment uses a mix of LC and multi fiber connectors. So naturally one might assume that LC – MPO breakout test leads are an obvious choice. However SC – MPO breakout leads are typically more robust, easier to handle, and off slightly improved optical performance.

Number the leads and ends A/B

Test cords require Performance Verification before each use. This performance verification test is quite specific, and failure to do this preliminary test may invalidate all the following test results. The basic objective of the performance test is to ensure that the loss of the cords meets the basic accuracy requirements. However, there is a sting in this tail as follows:

  • The test standards require a loss of eg 0.1 dB or better for multimode fiber
  • The “elite” average loss meets this specification
  • However the actual “per fiber maximum” loss does not.

We hope the standards folk will resolve this in due course. One potential way around this is that if the actual loss of each connection is recorded for each fiber during this test, then this loss can be used to compensate the following test results and improve accuracy. However in practice, this is excessively hard to implement, so the practical policy is to only use “good” cords.

The Test Cord Verification Test is quite simple: test each test cord connector using two reference cords, and the loss of the connection must be within the allowed limit. In the case of MPO connectors, this will take a bit longer due to the number of fibers.

What test uncertainty can I expect?

This is a much discussed issue. Without going into any great detail, there are a few obvious pointers:

The test cord performance test using “Elite” connectors could have losses of from 0 – 0.35 dB per connector, giving a per mated connector uncertainty of ±0.18 dB, This means that any overall test uncertainty must be at least ±0.18 dB, and is likely to be larger once other factors as included.

However once the test cords are plugged into presumably lower grade cords, the loss there could possibly jump to 0 – 0.75 dB, giving a minimum per mated connector uncertainty of ±0.38 dB. These are simple deductions based on connector specifications, and calculated uncertainty including all factors could only get worse.

Alternatively, if suitable experiments are performed with the actual equipment and test cords, it may be possible to demonstrate a lower level of uncertainty, however this is specific to a particular item of test equipment, and particular test cords, and so has limited viability.

General loss test equipment choices

There are three broad choices to make, which will depend on the scale of expected work:

  1. Use a basic source and meter. This has minimal equipment cost, but is fiddly to perform (and so prone to error), since both instruments need moving for each fiber to be tested, so it’s only sensible for a very small project.
  2. Use a basic source and special MPO compatible meter, so that only the source needs moving for each fiber. For a modest equipment cost, this greatly speeds up and simplifies testing. This also works for any fiber type, or wavelength, or number of fibers per connector, so it has excellent flexibility, and the same equipment can be used for other work. A good choice for many mid-size projects, or cabling contractors.
  3. A special MPO test set, However these usually only work at 850 nm and 12 fibers, so they are both expensive and inflexible, and so are useful for a specific large project.

Whatever test equipment is selected will need some good reporting software. It may be a project requirement that it’s impossible to modify the test results, to prevent tampering.

Procedure

Loss testing using solution 2 will generally proceed as follows (remembering to inspect and clean on each insertion):

  1. Begin by verifying the test cords. Once these are proven to be adequate:
  2. Reference the source and meter using a breakout cable and the desired number of reference cords.
  3. Set up the MPO power meter at the far end of the test system
  4. Start with the light source at “fiber 1” on the breakout cable , take the loss reading
  5. Move the light source only to the next fiber on the breakout cord, and take the loss reading
  6. Etc
  7. Go back to fiber 1 and note the reading (is within limits compared to the 1st time)

We found that using a Kingfisher light source KI2803 with power meter KI2600XL-Ge5, and this technique, the repeatability was determined by the test cords. Ease of testing and speed were both good, since only the source needed moving for each fiber.

Loss testing using solution 1 will differ in that the power meter will need to be moved for each fiber. This coordinated shuffle greatly slows down the test process, and is error prone.

 


You may print, photocopy and redistribute this document for educational use only, provided that this page is printed as is, without modification. This content may not be used commercially.

Kingfisher International are specialist designers and manufacturers of handheld fiber optic test equipment, with a mission to simplify optical communications.

Our equipment is used in all phases of fiber optic manufacture, installation and maintenance. Our comprehensive documentation and resources help you get started easily.

Kingfisher products are Australian made, with a global distribution & support network spanning over 70 countries.