Variance in a test setup can introduce uncertainty into the return loss (RL) measurement process, reducing its overall accuracy. In such cases, devices under test (DUTs) that should have been rejected may erroneously pass the quality assurance (QA) process, causing unforeseen problems when put into use. Production managers and technicians must be aware of potential false passes so they can take steps to eliminate their causes, ensuring the accuracy and validity of the test process.
This article focuses on RL measurement setups that are prone to introducing testing errors. Ways to mitigate these errors are discussed. Some errors are unavoidable; therefore, steps for correcting them out or reducing them are also described here. Note that RL and reflectance are sometimes confused. This article discusses reflectance of single connectors or events, which in the fiber-optic industry is sometimes categorized under “return loss.” All values in this article will use a positive sign convention, which is the standard for return loss measurements, but not for reflectance.
Accounting for measurement offsets
For a simple single-channel test setup, it may not be necessary to make corrections to the RL reference. However, test setups incorporating switches or other devices increase the number of events in the optical path, resulting in losses that were not accounted for during the factory calibration.
When an optical switch is placed in the measurement path, it introduces insertion loss (IL). Such loss comes from the cable connection from the meter to the switch input, and from the switch itself. This IL affects the RL measurement accuracy because it changes the perceived magnitude of reflections as seen by the detector in the instrument. This reduction in signal, if not accounted for, will potentially yield false passes.
An optical switch can introduce 1 to 2 dB of IL. The effect of this on RL is significant — roughly double the IL. For instance, a switch with an IL between 1 dB and 2 dB will add 2 to 4 dB of RL. The system will therefore see a smaller reflection than is the actual case.
In Figure 1 and Figure 2, a 100% mirror is used for simplicity. Without a lossy component, i.e., a switch, the pulse is reflected back completely; therefore, the measured RL is 0 dB. Adding a lossy component, the pulse experiences degradation in both the forward (Rf2) and reverse (Rb2) directions, resulting in a RL measurement offset. The loss causes the system to see a smaller pulse and yield an incorrect RL measurement. In the case of a system with an offset of 2 dB of RL, if a correction is not applied, a DUT that measures 46 dB should actually measure 44 dB if the proper correction factor is applied. If the pass/fail criterium is 45 dB, then a faulty connector would have passed through the quality control (QC) inspection.
Switches also have varying losses across the channel set. For large channel count switches, this variation can be as much as 2 dB from the lowest to the highest loss channel. Smaller switches may vary by 1 dB from the lowest to highest loss channel.
Since each switch channel has a different loss, it is advisable to find a correction factor for each one, thus correcting this variance. If not corrected, a reflection of a given magnitude will appear different to the RL meter depending on the channel and its loss. The RL meter would therefore measure different values on different channels for the same reflection.
External reflectance standards
An established method to correct for this system loss is to use an external reflectance standard. This is a device with a known reflectance value connected to each channel and then measured as a reference. Each measured value is compared to the accepted value of the reflectance standard. The difference between the accepted value and the measured value yields the measurement error. The error value can be applied to subsequent measurements as a correction factor.
A reflectance standard can be of any value, but should be within the dynamic range of the measurement system. If a RL meter only measures from 10 dB to 65 dB, a reflectance standard in the 75-dB range should not be used. Conversely, a 14-dB reflectance standard should not be selected if the RL meter only measures from 30 dB to 75 dB.
For CW based systems, a common method for correcting RL is using a polished flat connector open to the air, which is accepted as ~14.7 dB for standard single-mode wavelengths and standard silica glass. This reflectance standard is easily obtained and inexpensive. Most fiber-optic cable assembly lines have ready access to flat polished connectors and a simple reference cable can be used in many cases.
However, for most optical time domain reflectometer (OTDR) and pulse-based systems, use of a polished flat connector is not possible because the meter’s dynamic range begins at ~30 dB and a 14-dB reflection would saturate the system. A reflectance standard within the system’s limited dynamic range could be used; however, such devices are not always available, are difficult to implement, and can degrade in accuracy over time and use. Fortunately, there are pulse based RL measurement systems that employ measurement methods to accommodate a wide dynamic range that includes a large 14.7-dB reflection.
Some OTDR and pulse-based systems incorporate a reflectance standard within the unit itself, which simplifies referencing, but does not eliminate the effects of lossy components on the RL measurement.
Applying correction factors
During calibration of an RL meter, the instrument is usually calibrated over its full dynamic range and is verified to be linear over this range.
If the user introduces a switch into the setup, each channel has an associated forward and reverse loss. The forward and reverse loss are typically different, which means that each channel now has a different RL calibration plot. If an open PC connection (14.7 dB) is referenced for a 12-channel switch with varying loss of 0.7 dB to 1.6 dB in each direction, one might get RL values from 16.1 dB to 17.9 dB.
The calibration plots for each channel will be offset by the difference of the expected RL value and the measured RL value. This error propagates through all measurements for all channels. If one were to measure a true 50-dB reflection on a connection, with the RL through the switch varying from 15.9 dB to 17.9 dB, the result would be off by about 1.2 dB to 3.2 dB, adding an unacceptable level of error.
To apply a correction factor, one simply needs to subtract the offset measured when referencing an open flat connector (14.7 dB) from all measurements performed on that specific channel.
An indirect method that assumes the RL is double the system IL could also be used. However, since the forward and reverse IL may not be identical, unknown error is introduced into each measurement, making this method unreliable. This method also can add unnecessary complexity, because it requires that the reference cable be connected directly to the optical power meter (OPM) when RL is referenced.
When using an open flat connector as a reflectance reference, it is critical to ensure that its end face is perfectly clean and free of defects. If the end face has any contamination, scratches, etc., the resulting reflection may be affected, invalidating the correction factor found during the process.
Referencing APC connectors
Unlike flat polished connectors, open APC connectors do not have a standard RL value when measured. The RL can vary from -50 dB to -75 dB depending on the polish quality and the true angle of the polish. This can cause issues when referencing to obtain a correction factor.
In this case, it is possible to use a reference stub. A basic reference stub is a short cable with an APC connector on one end and an open PC connector on the opposite end. The open PC connector has a reflectance of 14.7 dB.
For maximum accuracy, the APC connectors should be of the highest quality, with low loss through the APC-APC connection. Before using a reference stub, first measure the IL of the APC connection using an OPM to verify it has low loss, since additional IL adds uncertainty to the measurement.
To establish the reference reflection, the APC side of the stub is connected to the APC connector of the cable and the PC side is left open to the air (Figure 3). The setup is then referenced as usual. Once the correction factor has been found, the short stub is disconnected and the user can begin testing devices applying the correction factor.
Accounting for other effects of lossy components
Test setups containing lossy components decrease the dynamic range at the low end of an RL meter. For every 1 dB of loss, the RL meter loses approximately 2 dB at the low end of the range.
If an RL meter has a dynamic range from 10 dB to 80 dB and the setup has a loss of 1 dB, the RL signal is doubled, degrading by 2 dB. This reduces the detection capability of the receiver and brings the dynamic range on the low end to 78 dB. The dynamic range at the high end is affected as well, degrading to 8 dB. This is because the signal has been reduced overall and no longer saturates the detector in the meter.
Now that multifiber connectors (e.g., duplex LC, MPO/MTP, etc.) have become the norm, an optical switch is almost always found in a fiber-optic cable production line. As such, some equipment manufacturers have integrated the switch directly into their equipment. The loss of the switch is accounted for during calibration, which eliminates its effect on the dynamic range. However, there may still be channel-to-channel differences due to varying losses on the reference cable setup for each channel.
The wide range of fiber-optic cables that need to be qualified (i.e., hybrid, multifiber, etc.) has led to test setups becoming more complex. Added complexity often leads to IL in the test setup, resulting in uncertainty and erroneous RL measurements. When lossy components such as optical switches are in the optical path, it is important to reference and establish a correction factor for each switch channel using an external reflectance standard to ensure the RL measurements are reliable.
Chris Heisler is CTO at OptoTest Corp. He graduated from CalPoly San Luis Obispo with a Bachelor of Science in Electrical Engineering with a focus on Communications and Fiber Optics. He continued with his Master’s Degree at CalPoly in Digital Signal Processing and Photonics. Chris has been working in the fiber optics industry for 13 years. At OptoTest, he has helped to refine the return loss measurement process and push development of test equipment that satisfies customer’s needs. Chris is also a regular contributor to the TIA TR42 Fiber Optic Engineering Committees.