Insertion loss is measured by comparing signal power (or sound level) before and after it passes through a component or system, then expressing the difference in decibels (dB). The core process is the same across fiber optics, RF electronics, and acoustics: establish a baseline reference without the device in the signal path, insert the device, and record how much signal is lost. The specific equipment and procedures vary by field, but the math stays consistent.
The Basic Formula
Insertion loss uses a simple ratio converted to decibels. In its most common electrical form:
IL (dB) = −20 × log₁₀(V_out / V_in)
Where V_out is the signal voltage after passing through the device and V_in is the voltage before. You can also express this using power instead of voltage, which changes the multiplier from 20 to 10. A higher dB number means more signal was lost. A perfect, lossless component would have 0 dB insertion loss. A component reading 3 dB has lost roughly half the signal power.
In RF and microwave engineering, insertion loss corresponds to the S21 parameter in a scattering matrix, which describes how much signal travels from the input port to the output port of a device. You don’t need to calculate this by hand; test instruments do it automatically.
Measuring Insertion Loss in Fiber Optics
Fiber optic insertion loss testing requires two pieces of equipment: a stabilized light source at one end and an optical power meter at the other. The procedure has two phases: setting a reference, then measuring the cable plant.
To set the reference, you connect a test reference cord (a short, known-good patch cable) directly between the light source and the power meter, then set the meter to read 0 dB. This baseline accounts for the cord itself so it doesn’t skew your results. Once referenced, you disconnect the far end of the reference cord from the power meter, connect it to the cable plant you’re testing, and attach a second cord from the far end of the cable plant back to the power meter. The meter now displays the insertion loss of everything between your reference points.
Which connectors get included in that measurement depends on your referencing method, and getting this wrong is one of the most common mistakes in fiber testing.
Reference Methods: 1-Jumper, 2-Jumper, and 3-Jumper
The 1-jumper reference is the industry default and the most accurate. You connect a single test reference cord between the source and meter, set zero, then test the link. This method includes the connectors on both ends of the cable plant in your measurement, giving you the most complete picture of total loss with the least uncertainty.
The 2-jumper reference connects two cords together during the reference phase, which zeroes out the connection between them. Your final measurement then captures only one end connector instead of both. This method has the highest uncertainty of the three and should only be used when one or both ends of the link terminate with a plug that can’t connect directly to the meter.
The 3-jumper reference uses a substitution cord between two reference cords during calibration, zeroing out two connector pairs. It’s more complex and less accurate than the 1-jumper method. You’d use it when the connector type on your link doesn’t match the ports on your test equipment, which comes up with newer very small form factor connectors like CS, SN, or MDC types that require hybrid breakout cords.
Whatever method you use, document it. Your loss budget calculations need to match the test method, because each approach includes different connectors in the final number.
Acceptable Loss Values for Fiber
The TIA-568 standard sets limits for what’s considered a passing result. For a mated pair of connectors (one reference-grade, one standard-grade), the maximum allowable loss is 0.50 dB for both multimode and singlemode fiber. When calculating a loss budget for an entire cable run, you add up every contributor: each connector pair, each splice, and the fiber length itself. If your measured insertion loss comes in at or below that total, the link passes.
Measuring Insertion Loss in RF and Electronics
For RF components like filters, attenuators, cables, and connectors, a vector network analyzer (VNA) is the standard instrument. A VNA sends a known signal into one port of the device and measures what comes out the other port across a range of frequencies, giving you insertion loss as a function of frequency rather than a single number.
Before testing, you calibrate the VNA to remove errors introduced by the test cables and connectors themselves. This typically involves connecting known standards (a short circuit, an open circuit, and a matched load) to each port, then a through connection between ports. The VNA uses these references to mathematically correct every subsequent measurement. Once calibrated, you connect the device under test and the VNA directly reports the S21 magnitude in dB.
Input power level matters. VNAs have a default power setting (often around −17 dBm), but for devices with very high insertion loss, you may need to increase the input power to keep the signal above the instrument’s noise floor. NIST research has shown that comparing measurements at different power levels (for example, −17 dBm versus 0 dBm) helps verify accuracy for high-loss devices.
The Department of Defense test method standard for insertion loss (MIL-STD-220) requires that test equipment maintain a repeatable measurement within ±1.0 dB and hold a stable output signal within ±1% over a 2-minute period. For high-isolation measurements, the setup’s own dynamic range needs to exceed the expected insertion loss by at least 80 dB.
Measuring Insertion Loss for Sound Barriers
In acoustics, insertion loss quantifies how effectively a noise barrier reduces sound at a specific location. The ANSI/ASA S12.8 standard defines three methods for outdoor noise barriers. The most straightforward is the “direct” method: measure sound levels at a receiver location before the barrier is installed, install the barrier, then measure again at the same spot. The difference in decibels is the insertion loss.
When before-and-after measurements at the same site aren’t practical (the barrier already exists, or conditions have changed), the standard allows two indirect alternatives. One uses measurements at an equivalent site with similar terrain, ground conditions, and atmospheric properties. The other uses predicted sound levels based on modeling to stand in for the “before” data.
Sound sources for these tests can be naturally occurring (traffic, industrial noise), controlled natural sources, or artificial sources like loudspeakers. Microphone placement at the receiver location and atmospheric conditions during testing both affect results, so the standard prescribes limits on acceptable conditions and requires documentation of the setup.
Common Sources of Measurement Error
Dirty connectors are the single biggest source of false readings in fiber optic testing. A speck of dust on a fiber endface can add tenths of a decibel to your result or cause wildly inconsistent numbers. Always inspect and clean connectors with a fiber cleaning tool before every test, including reference cords.
In RF measurements, impedance mismatch between the test equipment and the device under test creates reflections that distort the insertion loss reading. Proper calibration reduces this, but worn or damaged connectors on your test cables reintroduce error. Replace test port cables regularly if you do frequent measurements.
Temperature drift affects both optical and electronic measurements. Components change their loss characteristics as they warm up, and instruments themselves can drift. Allow test equipment to stabilize at room temperature before calibrating, and avoid testing in environments with large temperature swings.
For devices with very low insertion loss (close to 0 dB), measurement sensitivity becomes a challenge. Small errors in the reference can represent a large percentage of the actual loss. For devices with very high insertion loss, the signal can drop below the noise floor of your instrument, producing unreliable numbers. Matching your equipment’s dynamic range to the expected loss range of the device is essential for trustworthy results.