Loading…
Convert a trace length to propagation time, and a time budget back to length, for microstrip, stripline, asymmetric stripline, embedded microstrip and coplanar waveguide. The panel also reports effective Dk, wavelength, electrical length, critical length and the mismatch that produces one picosecond of skew.
Direct answer
A microstrip on 0.2 mm of FR-4 has an effective Dk near 3.3, so its delay is about 6 ps/mm. A stripline in the same laminate is fully embedded, has an effective Dk near 4.3 and runs about 6.9 ps/mm. A 100 mm route is therefore about 600 ps on the outer layer and 690 ps on the inner layer. That 90 ps difference is why a layer change has to be compensated, and why a length match on two different layers is not a length match.
Calculator
Choose the cross-section, set the laminate and the frequency, then enter a length or a time. The panel converts both ways, reports the effective Dk and the guided wavelength, and calculates the length at which the route stops behaving as a lumped element. The exact button runs the 2-D Laplace solve for a covered or two-Dk section.
Model and validity range. Hammerstad-Jensen and Kirschning-Jansen for microstrip dispersion, Cohn for stripline, Ghione-Naldi for coplanar and a 2-D Laplace solve for asymmetric or covered sections, with a causal wideband Dk. The conversion is the TEM relation; the model holds where the cross-section is uniform and the return path is continuous, and the exact solve is a 2-D result that does not see a via or a launch.
Direction
Geometry
Material and frequency
Timing
Critical length criterion
Propagation delay and length
Delay, length, wavelength, electrical length, critical length and layer-change skew.
100.00 mm is 604.9 ps
100.00 mm is 604.9 ps
Closed-form model. Run the exact 2-D solve to include the real permittivity map.
153.65 ps/in.
0.605 ns.
3.94 in.
0.02% from the static value.
Closed-form model.
At 1.000 GHz.
At the reference frequency.
t_r / 2 criterion.
0.35 / t_r.
1 mm of mismatch; 0.165 mm per ps.
What this does not cover
Free, no account. An engineer reviews these inputs before the fixed fee is issued.
Results are engineering estimates, not design verification. Confirm them against the fabrication drawing, a TDR coupon or a 3-D extraction before release. Read the tool terms.
A wrong number, a rejected in-range value or a missing model? Send it to the engineer.
Every result above traces to an equation below. Read the equations and their validity limits.
Equations
Each block gives the equation the calculator evaluates, every symbol with its unit, and the range over which the model holds, with what it cannot see stated alongside. The button in the corner copies the LaTeX source.
The delay is the time light takes to cross the same length in the effective dielectric, which is the only material dependence.
The relation is exact for a uniform TEM line and the standard quasi-TEM approximation for a microstrip. The length is the routed length; the effective permittivity is the fraction of the field in the laminate seen by the wave.
A stripline has one dielectric, so its effective permittivity is the laminate Dk. A microstrip has air above the trace, so its effective permittivity is lower and its delay per millimetre is shorter.
Model limits
The relation assumes a uniform cross-section and a continuous return path. A via, a layer change, a connector launch or a plane split adds delay and reflection that the one-way formula does not contain.
The inverse of the delay relation, used to turn a timing allowance into a routed length.
The inversion is algebraic because the relation is linear in length. What makes it a calculation rather than a division is the effective permittivity: a 5% error in effective Dk moves the length by 2.5%.
Use the delay of the layer the route is actually on. A length solved from a microstrip delay and routed as a stripline will miss the budget by the difference between the two effective Dk values.
Model limits
The result is a straight-line allowance, not a routing instruction. The router needs the length window plus the layer mix and the transition delays, and a serpentine adds coupling that changes the delay.
The wavelength on the line, which is shorter than the free-space wavelength by the square root of the effective permittivity.
A quarter-wave stub, a resonant via or a periodic structure is sized from the guided wavelength, not from the free-space one. At 1 GHz on FR-4 the guided wavelength is about 146 mm for an effective Dk of 4.3, not 300 mm.
The effective permittivity at the frequency of interest is the one that matters. Using the 1 GHz value at 28 GHz overstates the wavelength by the dispersion of the laminate.
Model limits
The formula describes the guided mode on a uniform line. A stub with an open end, a discontinuity or a nearby conductor has a different resonance, and a 3-D solve is the number to use.
The length at which the one-way delay becomes a chosen fraction of the rise time, the boundary between lumped and distributed behaviour.
The divisor n is a convention. A value of 2 is the common first check; 4 and 6 are more conservative and produce a shorter threshold.
Below the critical length a lumped model is adequate for the edge. Above it the route has to be treated as a transmission line, with termination and return-path control.
Model limits
The boundary is a design convention, not a sharp physical transition. A route just below the threshold still has reflections, and a route well above it is not automatically a problem if the termination is right.
The same delay expressed as a phase angle at a frequency, which is how a matching network or a delay line is specified.
A 1 GHz wave has a 1000 ps period, so 1 ps of delay is 0.36 degrees. The relation is linear in frequency until the line's dispersion moves the effective permittivity.
The phase angle is useful for a delay line, a phase shifter or a clock tree, but it is not a substitute for the time-domain delay when the edge contains many frequencies.
Model limits
The angle is defined at one frequency. A wideband signal sees a different phase at every component, and the group delay, not the phase at the clock, is what preserves the edge.
Definition
The propagation delay of a uniform line is its length divided by the speed of light in the effective dielectric. Nothing else in the equation depends on the signal, so a timing budget can be built from geometry alone once the effective Dk is known.
The effective Dk is the part that varies. A stripline is surrounded by one dielectric and its effective Dk is the laminate Dk. A microstrip keeps part of its field in air, so its effective Dk is lower, it runs faster and its value moves with frequency. A coplanar or asymmetric line is somewhere between the two.
Layer Changes
A microstrip and a stripline on the same board do not have the same delay per millimetre. The outer layer has air above the trace, so its effective Dk is lower and it is faster. A route that changes layer therefore changes its delay at the transition, and a length match that ignores the layer mix will miss the timing budget.
Use the delay per millimetre of each layer, multiply by the length on that layer, add the two, and compare with the budget. If the route has to hit a specific time, solve for the length on one layer after the other is fixed. The panel's time-to-length direction does that for a single layer; do it twice for a layer change.
For a DDR or SerDes interface, the timing budget and the layer mix belong together. Our high-speed PCB layout service work writes the constraint set per layer, and the DDR Length Matching tool turns a skew allowance into a length window.
Consequence
A route shorter than roughly half the electrical length of its edge behaves as a lumped element; a longer one behaves as a transmission line and needs termination. The exact divisor is a convention, not physics, which is why the panel exposes t_r / 2, t_r / 4 and t_r / 6 instead of hiding one.
Rise time also sets the bandwidth the route has to carry. A 100 ps edge has about 3.5 GHz of bandwidth by the 0.35 / t_r rule, and the delay at that frequency is not always the delay at the clock frequency. The dispersion term in the microstrip model is what connects the two.
A delay that looks wrong usually means the cross-section, not the arithmetic. Check the stackup in the Stackup Planner and the impedance in the Impedance Calculator before adjusting a length constraint. Our SI/PI verification service and DFM/DFA layout service work carry the result into the constraint set and the fabrication review, and PCB layout pricing quotes the layout once the stack is fixed.
FAQ
Send the stackup, the interface list and the length and delay budget inputs. An engineer re-solves them against the real layout before the fixed fee is issued.