The insertion loss calculator above converts a before-and-after measurement into decibels. Insertion loss is defined by exactly that procedure: measure the power delivered to a load, insert the component under test into the path, measure again, and take the ratio. It is the standard figure of merit for connectors, cables, filters in their passband, splitters, switches and every other component that is supposed to pass a signal through rather than change it.
Arb Digital publishes free engineering calculators that separate quantities people routinely conflate. The useful thing this page adds is the split between reflected and dissipated loss. A component can register two decibels of insertion loss while dissipating almost nothing, simply because it is badly matched and bounces power back towards the source, and the fix for that is completely different from the fix for genuine absorption.
What This Insertion Loss Calculator Does
The hero figure is the insertion loss in decibels. Because decibels are a logarithmic ratio, the same number applies whether your readings are in watts, milliwatts or microwatts, provided you use the same unit for both. The tool also accepts levels already expressed in dBm, in which case the loss is simply the difference between them, and voltage readings taken at a constant impedance.
The supporting grid gives the fraction of power that survives, expressed as a percentage, which is often far more intuitive than the decibel figure. It gives the voltage ratio, which is what you would see on an oscilloscope rather than a power meter. It gives the share of the loss attributable to reflection, computed from the return loss you supply. And it gives the total for a cascade of identical sections, because decibels add along a chain.
How to Use It
- Take the reference measurement first. Connect the path with the component removed and the two cable ends joined directly, and record the level. This is the reading everything else is compared against.
- Insert the component and measure again. Change nothing else — not the source power, not the cables, not the analyser settings. Any difference between the two setups other than the component contaminates the result.
- Choose the matching input mode. Power readings and dBm levels behave differently in the arithmetic, and voltage uses a factor of twenty rather than ten.
- Add the return loss if you have it. Most vector network analysers give it on the same sweep as the through measurement, and it is what separates reflection from absorption.
- Set a cascade count for budgeting. If a link contains six of the same connector, entering six projects the total straight away.
The Formula: How Insertion Loss Is Calculated
From a pair of power readings, insertion loss in decibels is IL = 10 log10(Pref ÷ Ptest). From a pair of voltage readings taken in the same impedance it is IL = 20 log10(Vref ÷ Vtest). The factor of twenty appears because power scales with the square of voltage, not because voltage decibels are a different unit. From two levels already in dBm the loss is simply the difference.
Mismatch loss comes from the reflection coefficient. A return loss of RL decibels corresponds to |Γ| = 10−RL/20, and the power that fails to enter the component because it reflects is −10 log10(1 − |Γ|²) decibels. Subtracting that from the measured insertion loss leaves the dissipative share, which is the part that turns into heat inside the component.
Cascading is straightforward because decibels are logarithmic: N identical sections cost N times the loss of one. That is the whole reason the industry works in decibels rather than ratios, and the ARRL transmission line reference sets out the conventions used here.
Work the defaults through by hand. A reference reading of 100 mW falling to 63 mW gives a ratio of 1.5873, and 10 log10(1.5873) = 2.007 dB. That means 63 per cent of the power survives and the voltage ratio is √0.63 = 0.7937. With a measured return loss of 15 dB, |Γ| = 10−0.75 = 0.1778, so |Γ|² = 0.0316 and the mismatch loss is −10 log10(0.9684) = 0.140 dB. The remaining 1.867 dB is genuinely dissipated. Four such components in cascade would cost 8.03 dB in total, leaving under 16 per cent of the original power.
Reflected Loss and Dissipated Loss Have Different Cures
This is the distinction the page exists to make. Insertion loss as measured is a single number that hides two mechanisms. Some power never enters the component at all because the impedance discontinuity at its input reflects it. The rest enters and is converted to heat by conductor resistance, dielectric loss or deliberate attenuation.
Reflected loss responds to matching. Improve the return loss and that share shrinks towards zero without changing the component. Dissipative loss does not care about matching at all; a better connector, shorter cable, lower-loss dielectric or different filter topology is the only route. Spending a week on a matching network to fix a loss that was 90 per cent dissipative is a familiar and avoidable waste.
There is also a thermal consequence. In a high-power transmitter the dissipated share is real heat inside the part, and a component rated for its through power can still fail if it is absorbing more than expected. The reflected share goes back towards the source instead, where it stresses the amplifier rather than the component.
Why the Reference Measurement Decides Everything
Insertion loss is defined relative to the path without the component, so the quality of the answer is limited entirely by the quality of that reference. A reference taken with a different cable, a different connector torque, or a different analyser power level does not measure the component; it measures the difference between two setups.
The usual discipline is calibration. On a vector network analyser you calibrate to the plane where the component will sit, using a known through connection, so the reference is stored rather than remembered. Working with a signal generator and a power meter, take the reference immediately before and immediately after the test measurement and check that the two agree, because source drift over a long session is easily as large as the loss you are trying to measure.
Watch the connectors as well. Adding a component usually means adding two connector pairs that were not previously there, and each pair contributes its own loss. Strictly, what you measured is the component plus its interfaces. For a low-loss part that overhead can be a large fraction of the total, which is why precision measurements use adapters that are calibrated out rather than added in.
Where This Sits Next to the Other Loss Tools
These pages measure four different things and they are easy to confuse. This one covers a discrete component in the path, characterised by a before-and-after measurement. The attenuation calculator covers loss accumulating along a length of cable or medium from a per-metre coefficient, so its input is a distance rather than a pair of readings. The free space path loss calculator covers the spreading of a radio wave between two antennas, which is geometric dilution rather than absorption. And the decibel calculator handles the underlying ratio arithmetic for any quantity at all, with no assumption about what caused the change.
If the mismatch share of your loss is large, the impedance matching calculator designs the network that removes it, and the RLC impedance calculator and reactance calculator help characterise what the component looks like electrically. The transmission-line theory underneath all of it is covered in MIT OpenCourseWare's 6.013 Electromagnetics and Applications.
Arb Digital builds free tools like this one because useful pages earn attention. If you want tools, calculators or content built for your own audience, we can help.
Browse All Free Tools Talk to Arb DigitalCommon Mistakes to Avoid
- Using 20 log for power readings — the factor of twenty belongs to voltage ratios only, and applying it to watts doubles the answer.
- Mixing units between the two readings — a reference in watts against a test in milliwatts adds a spurious 30 decibels that looks entirely plausible.
- Changing the setup between reference and test — a different cable or a different source level means you measured the setup change, not the component.
- Treating all insertion loss as heat — a badly matched part reflects power rather than absorbing it, and its thermal load is far lower than the decibel figure suggests.
- Forgetting the connectors you added — inserting a component usually adds two connector pairs, and for a low-loss part those interfaces can dominate the measurement.
Related Free Tools From Arb Digital
Use the decibel calculator for general ratio arithmetic, the attenuation calculator for loss along a length of cable, and the free space path loss calculator for the over-the-air part of a link budget. When mismatch turns out to be the problem, the impedance matching calculator designs the network that fixes it and the cable impedance calculator covers the line itself. The EIRP calculator takes the surviving power through the antenna, and the reactance calculator converts component values into ohms at your frequency. Everything Arb Digital publishes is on the free online tools hub.
Frequently Asked Questions
It is the reduction in power delivered to a load caused by inserting a component into the signal path, expressed in decibels. It is defined by the measurement procedure: record the level with the component removed and the path joined directly, insert the component, record again, and take the ratio of the two.
Ten times the logarithm for power ratios, twenty times for voltage or current ratios. The difference exists because power is proportional to the square of voltage, so a factor of two in voltage is a factor of four in power. Both give the same decibel figure for the same physical change, provided the impedance is the same at both measurement points.
Insertion loss describes a discrete component placed in a path and is found from a before-and-after measurement. Attenuation usually describes loss accumulating along a length of cable or through a medium, expressed as a coefficient per unit distance and multiplied by the length. The units end up the same but the inputs and the physical picture are different.
Because the power never entered it. An impedance discontinuity at the input reflects part of the incident wave back towards the source, so it registers as missing at the output while dissipating nothing inside the component. That share is recoverable by improving the match, whereas genuinely dissipated loss is not.
Yes, in decibels, because decibels are logarithmic. Six connectors each costing 0.2 decibels cost 1.2 decibels in total. In linear terms the transmitted fractions multiply instead, which is exactly the same statement written differently and is why the industry works in decibels.
As accurate as the loss you are trying to resolve. Measuring a 0.1 decibel connector needs a reference stable to well under that, which usually means a calibrated vector network analyser rather than a generator and a power meter. Source drift across a long session can easily exceed the quantity being measured.
Only if the component provides gain, which makes it an amplifier rather than a passive part. If a passive component appears to show negative insertion loss, the reference measurement is wrong: usually the source level drifted upwards, the readings were taken in different units, or a connector in the reference path was faulty and improved when it was remade.
This tool is provided for educational and preliminary engineering use. It assumes both readings were taken in the same impedance with the same source conditions, and the mismatch split assumes a single measured return loss applies across the band of interest. Verify component specifications against a calibrated vector network analyser measurement at your operating frequency.
