Last month, I discussed making loss measurements with a light source and power meter—what we call optical insertion loss—and the causes of errors in those measurements. The conclusion was that these measurements generally had potential for errors of as much as ±10 percent, a combination of all systematic and random errors.

An error of up to 10 percent is significant. It is important when interpreting data from a display that reads to 0.01 decibels (dB). For example, a cable that measures 3.17-dB loss might actually have a loss of about ~2.85 to 3.50 dB. That meter reading has a resolution much higher than needed for the measurement—0.1 dB is plenty for most loss measurements—and can falsely indicate that the measurement is more accurate than it really is.

The confusion begins when deciding what to do with that measurement. Do you pass or fail that cable? Does it meet the limits calculated in a loss budget? Will it work with a system that has a power budget of 3 dB? For that matter, what is a loss budget or power budget?

A loss budget is a calculated estimate of a cable’s loss, usually done during a project’s design phase. You add up the estimated loss of the fiber and all the connectors and splices to get a total loss for the cable. The fiber loss is calculated by multiplying the length in kilometers (km) by the attenuation coefficient in dB/km.

The “estimate” part comes from the values chosen for the attenuation coefficient of the fiber and the loss of connectors and splices. For premises cabling, you can choose the TIA 568 values that are “worst case” and quite high—connectors are 0.75 dB for example—or you can choose something more realistic. Most connectors have a loss of 0.3–0.5 dB, except for prepolished splice connectors and multiple fiber connectors that are much higher—typically 0.5–1 dB.

To illustrate the difference, a 850-nm link 200 meters (m) long with OM3 multimode fiber having two intermediate patch connections would have a loss of 3.7 dB with TIA values and 1.8 dB with typical values.

A power budget is the maximum loss specified for a specific fiber optic network, for example, an ethernet LAN link. Theoretically, the power budget can be calculated as the difference between the output power of the transmitter and the required input power at the receiver.

However, there may be other issues to consider. When it comes to multimode fiber networks, the bandwidth of the fiber affects the power budget for high speed systems. Simply looking at the transmitter and receiver specs for a 10-gigabit (G) LAN, you might surmise that a 5-dB power budget would be required. But when you factor in the bandwidth of the fiber that creates a power penalty, you will find that a 200 m link must have less than 2-dB loss.

So let’s look at the cable we said measured a 3.17-dB loss. It’s 200 m long and has four connections, including the connections on each. What happens when we compare it to our loss budget and power budget. Does it pass or fail? Here is the data we are faced with:

• Measured loss: 3.17 dB; probable actual loss is 2.85 to 3.50 dB

• Loss budget: 3.7 dB(TIA) or 1.8 dB (typical)

• Power budget: Low speed is 5 dB; high speed is 2 dB

That’s a head scratcher, isn’t it? The 3.17-dB loss is within TIA limits but way over the typical estimate. It’s OK for low-speed networks but way too much loss for high-speed networks.

I’d call it a fail. My biggest concern is connector loss. Subtract the 0.6–0.7 dB loss of 200 m of fiber and you get about a 2.5-dB loss from connectors. If that cable has four connectors, that’s more than 0.6 dB each on the average, or if three connectors are more typical (~0.4 dB), one connector has about a 1.3-dB loss. Get started troubleshooting.

Another common decision is what do you do if the loss budget is 1.8–dB and the cable measures at a 1.95-dB loss? And it’s a cable for a 10G system with a 2-dB margin. Is that a failure? I’d call it a pass. The measured value and the loss budget are both estimates and the two values are within ±10 percent—reasonable margins of error.

Get the idea? You must consider all the possible errors and use your judgment. Now we need to troubleshoot that 3.17-dB cable—get out the optical time-domain reflectometer (OTDR). But wait, how accurate will the OTDR test be?