How Accurate Is That Measurement?

And why should you care?

When making a measurement on a newly installed fiber optic cable, do you ever wonder when you make a measurement, just how accurate that measurement is? Just what does accurate mean, anyway?

A measurement is accurate if the value we measure is close to the real or actual value. The other important issue for measurements is their “precision,” or how repeatedly we can take measurements of the same quantity. In order to have accuracy, you must first have precision. Precision comes from standardizing the measurement process to reduce the variation among tests.

Now that you know the difference between precision and accuracy, we can analyze just how accurate the fiber optic measurements you make are. Since most measurements are of cable plant loss, let’s consider them.

Measuring insertion loss with a light source and power meter or optical loss test set requires you also use reference test cables on each instrument and mating adapters to connect up to the cables under test. There are several factors associated with these test tools that affect the precision and accuracy of the measurement, some of which are controllable by the user. The user has little control over the wavelength of the source or the linearity of the power meter, but the instruments’ battery charge status and condition of the reference cables and mating adapters, which are critical to reduce measurement errors, are under the control of the user.

Reference cables should be of high quality, determined by testing then against each other for insertion loss. The mating adapters should have metal or ceramic alignment bushings and be rated for use with single-mode fiber, even if multimode is being tested. Some standards, like TIA-568-B, require the addition of a mode conditioner to the launch cable attached to the source, usually done by wrapping the reference cable around a mandrel of a certain size for each fiber type. And needless to say, everything must be kept spotlessly clean.

Careful control of all these conditions should allow making measurements with a precision of about +/-0.1 dB. You can test this yourself. Set up your instruments, set your references and test a single cable, preferably a short one with no connectors, according to OFSTP-14. Record the loss to 0.01 dB resolution if possible. Then disconnect the cable you are testing and retest ten or more times. Calculate the average of the readings and see how much each individual reading differs from the average. It should be within 0.1 dB if you control your process well.

Once we know how to make a repeatable measurement—a precise measurement—we want to know how accurate it is. In order to know how accurate a measurement is, we must know what the “real” value is for comparison. And here, I’m afraid, we are encounter a problem.

The issue is not just the source and power meter. We can send them to calibration labs that will check conformance to their specifications and actual performance—the source wavelength, meter linearity and power calibration to NIST standards—but when we measure the loss of a cable with these instruments, the measurement depends on the test conditions controlled by our reference cables and mating adapters.

Testing a single fiber optic connector requires mating it to a reference connector, as we define connector loss as the loss of a mated pair of connectors—
since a single connector has no loss, per se, if it is designed to connect two fibers. The loss we measure with different reference connectors may vary considerably. Prove it to yourself. Test the same cable we tested above with several different sets of reference cables. I’ll bet the individual measurements vary by as much as +/–0.25 dB. On a typical cable plant with several connections, it can be +/–0.5 dB. Any connector you mate to a higher loss connector will most likely have higher loss, so it is to your advantage to make sure that your reference cables are all in good condition with low loss connectors.

Why is this such a big deal? When you contract to install a fiber optic cable plant, you agree to meet certain specifications for component losses and the total cable plant loss. Getting your payment will probably depend on meeting those specifications. The more accurate your measurements are, the less likely you are to reject good cables or accept bad ones. Either case can be expensive because you may fix a cable that was really acceptable or be called back later to fix one that tested acceptable earlier. Learning how to use and maintain your test equipment properly can be financially rewarding.                 EC

HAYES is a VDV writer and trainer and the president of The Fiber Optic Association. Find him at



Stay Informed Join our Newsletter

Having trouble finding time to sit down with the latest issue of
ELECTRICAL CONTRACTOR? Don't worry, we'll come to you.