Last month, I wrote about measurement uncertainty and metrology, the science of measurements. This month, I get more specific and cover the uncertainty of some basic fiber optic measurements, starting with optical power.


Optical power is the “voltage” of fiber optic measurements. Just like voltage is the basic measurement of electricity and the basis of the measurements of practically all electrical parameters, optical power is the equivalent for fiber optics.


Optical power is used when measuring the output of a transmitter or input of a receiver to test a fiber optic transmission system. That is a measurement of absolute power, generally expressed in decibels referenced to a milliwatt of optical power (dBm). When measuring loss, we measure power before and after the component being tested and that is relative power measured in decibels (dB).


Decibels are a logarithmic conversion of power in watts. Log scales such as decibels are used when very large ranges are involved, such as in fiber optics where power levels can go from nearly 100 milliwatts down to 1 microwatt. It’s easier to use +20 dBm to –30 dBm than trying to use watts, milliwatts and microwatts without making mistakes.


What are the errors when measuring power? For absolute power, calibration is the biggest source of errors. The sensors used in optical power meters are very sensitive to light wavelengths. Power meters are usually calibrated at 850 nanometers (nm), 1,300 nm and 1,550 nm, the three most common light wavelengths used in fiber optic systems. The difference between the sensitivity of a power meter detector at 850 nm and 1,300 nm can be more than 3 dB, so it is important to check the wavelength set on the power meter to ensure it is the same as the light being measured.


Even if you have chosen the correct calibration wavelength, the calibration uncertainty can be as much as ±5 percent or ±0.2 dB because of errors transferring calibration from a standards lab to the manufacturer of the power meter and then to individual meters. Calibration in the United States is traceable from a National Institute of Standards and Technology transfer standard to the lab of the power meter manufacturer and then to each meter they manufacture. Each step adds some uncertainty to the calibration.


Meters also should be recalibrated frequently, generally every one or two years, to ensure the meter has not drifted off calibration. Calibration is also done at several power levels to check the power meter’s linearity. Since meters are also used for measuring loss that requires measuring and comparing power at two different power levels, it is important to calibrate power at different levels to ensure loss measurements will be more accurate.


Remember last month when I wrote about random and systematic errors? In power measurements, calibration causes systematic errors. A meter that is not properly calibrated will measure all powers in error. When making power measurements, random errors must be considered, and many are a result of operator problems.


If you are using a patchcord to measure optical power from a source, the coupling of light from a source to the detector of a power meter will depend on the quality of the connectors on the patchcord. If the connectors are bad—scratched, scuffed or dirty—the coupling will be bad and power levels will generally be measured lower because of the poor connector quality. The scratching and scuffing of the connectors can create systematic errors. The dirt can create random errors if improper cleaning procedures are used. To ensure proper power measurements, use only patchcords in good condition, and regularly inspect and clean the connectors.


In some cases, the connection to the source and power meter may be a variable. If the adapters on the source and meter are not precise, you may get a different measurement every time you connect up to make a measurement. Try that yourself on your instruments to see how repeatable they are.


Remember how putting stress on fiber optic cables causes loss? That’s another source of power measurement errors. If the cable attached to the source or meter is bent too tightly, especially near the connectors, the loss caused by the stress will reduce the power measured, adding to the random errors.


Knowing how to make accurate optical power measurements is important even if you never test an actual communications system, because every fiber optic technician tests loss. Testing the loss of a cable or cable plant involves making two optical power measurements—one at each end of the cable being tested—doubling the importance of reducing errors.


Next month’s column will cover measuring loss.