Cable Crosstalk Estimator
Near-end (NEXT) crosstalk estimate using a simple capacitive coupling model.
How to Use
- Enter the aggressor's voltage swing and its rise or fall time.
- Enter the length the two conductors run side by side and their mutual capacitance per metre.
- Enter the victim line's impedance and the velocity factor of the cable or board (about 0.66 for many cables, 0.5 for FR-4).
- Read the coupled voltage and NEXT in dB; Show Work tells you whether the run is long enough for the crosstalk to saturate.
Worked Example
A 3.3 V logic signal with 1 ns edges runs 10 m alongside a 50 Ω victim wire, coupled by 10 pF/m, in cable with a velocity factor of 0.66. The edge is 3.3 V/ns. Signals travel 0.66 × 3×10⁸ m/s, so the saturation length is 1.98×10⁸ × 1 ns ÷ 2 = 0.099 m: only the first 9.9 cm counts.
Coupled voltage: 10 pF/m × 0.099 m × 3.3 V/ns × 50 Ω = 163 mV, which is 20·log(0.163 ÷ 3.3) = −26 dB. Slowing the edge to 10 ns stretches the saturation length to 0.99 m but leaves the 163 mV unchanged on this long cable.
The common mistake: using the whole cable length. 10 pF/m × 10 m × 3.3 V/ns × 50 Ω gives 16.5 V, five times the signal itself. The simple formula only holds while the line is shorter than v·tr/2.
Show Work
Formulas
History of Crosstalk Engineering
Crosstalk was a problem from the early days of the telephone, when conversations on wires strung side by side could be heard faintly on each other. Twisting the two wires of a circuit, which Alexander Graham Bell patented in 1881, makes the coupling from a neighbour cancel along the length, the same principle used in every Ethernet cable today.
Near-end crosstalk (NEXT) became the critical metric for local-area networking in the 1980s. Each Ethernet cable category, from Category 3 for 10BASE-T to Category 6A for 10 Gigabit, is defined largely by the NEXT it must meet at its rated frequency, achieved with tighter and more carefully varied twists and better insulation.
On PCBs, crosstalk arises from mutual capacitance and mutual inductance between parallel traces. DDR memory layout rules limit parallel run length, mandate minimum spacing (typically 3W trace-to-trace), and balance routing via interleaved ground traces on busy signal layers. HyperLynx and Ansys SIwave simulate these effects accurately; this tool gives a first-order lumped estimate for cables and short PCB runs.
About This Calculator
Reference: Ethernet Cable Categories, Pinouts & PoE.
Enter aggressor amplitude, rise/fall time, run length, mutual capacitance per meter, and victim impedance. The tool estimates coupled voltage V_c ≈ C_m · ℓ · (dV/dt) · Z, converts to NEXT in dB, and reports dV/dt and total lumped mutual capacitance. Use the mutual capacitance from the cable datasheet or a field solver if you have it; shielded and twisted constructions have far less than untwisted wires side by side.
Near-end crosstalk grows with length only up to the saturation length v·tr/2; beyond it the tool caps the coupled length, as transmission-line theory predicts. For detailed work, use a field solver or signal-integrity simulator such as HyperLynx, Ansys SIwave or ADS. To reduce crosstalk: slow edges with series R, twist pairs, add shielding, use differential signaling, or increase spacing. Everything runs client-side; no values leave your browser.
Frequently Asked Questions
What are NEXT and FEXT?
Near-end crosstalk (NEXT) appears on the victim at the same end as the aggressor's driver; far-end crosstalk (FEXT) appears at the opposite end. On cables and inner PCB layers, where the dielectric surrounds the conductors evenly, FEXT largely cancels and NEXT dominates.
Why does crosstalk stop growing with length?
Near-end noise builds up only while the aggressor's edge is still travelling along the coupled section. Once the round trip, 2 × length ÷ speed, is longer than the edge, extra length adds duration, not height. With a 1 ns edge in a 0.66c cable that happens at about 10 cm.
Do slower edges reduce crosstalk?
On short runs, yes, in proportion. On long runs, no: the saturated level depends only on the coupling, not the edge rate. A 10 m cable gives the same 163 mV with a 1 ns or a 10 ns edge; a 0.5 m run drops to 82.5 mV with 10 ns edges.
How can I reduce crosstalk?
Increase spacing (on boards, at least three trace widths apart), keep the return path close with a solid ground plane or shield, use twisted pairs or differential signalling, shorten parallel runs, and slow edges where the run is short.
How accurate is this estimate?
It is a first-order capacitive model that assumes the inductive coupling matches, as it does in a uniform dielectric. Treat it as a quick check; for tight designs, use measured cable data or a field solver.
How do I use the Cable Crosstalk Estimator?
Just type your numbers. The answer shows up right away — there is no button to press. Change anything and it updates by itself.
Is it free? Does it work without internet?
Yes to both. It is free with no sign-up, and once the page has loaded it keeps working even with no internet.
Where does my data go?
Nowhere — every calculation runs on your own device. Nothing you enter is uploaded, logged, or stored.
Common Use Cases
Long Signal Cables
A 3.3 V logic signal with 1 ns edges on a 10 m cable couples about 163 mV into a neighbouring wire, enough to upset some inputs.
PCB Traces
Two FR-4 traces running 5 cm side by side with 10 pF/m coupling pick up about 82 mV from a 3.3 V, 1 ns edge.
Ribbon Cables
Adjacent wires in a 2 m ribbon with 20 pF/m coupling and 5 V edges can couple over 1 V, which is why ribbon cables alternate signals with grounds.
Ethernet and Telecom Cable
Cable categories are largely defined by how much NEXT they allow at their rated frequency, achieved with tightly twisted pairs.
Last updated: