What is the typical FR-4 propagation delay?
About 177.7 ps per inch (6.997 ps/mm) on an inner layer with εr 4.4, and about 149.8 ps per inch (5.899 ps/mm) for a 0.3 mm wide microstrip over 0.2 mm.
Start with your stackup, or an effective Dk from your fab, and this PCB propagation delay calculator returns the signal delay per inch and per millimeter on microstrip and stripline, the velocity factor, and the delay of a trace of any length.
Use your laminate’s Dk at the signal frequency.
Optional: fill in for a differential pair to use its odd-mode delay.
Delay per mm
5.899 ps/mm
Hammerstad-Jensen microstrip, quasi-static with copper thickness. Solder mask adds dielectric above the trace, so a masked outer-layer trace is a little slower than this.
tpd = √εeff ÷ c. On FR-4 with εr 4.4, a stripline has 177.7 ps per inch (6.997 ps/mm). A 0.3 mm microstrip over 0.2 mm has εeff 3.13, or 149.8 ps per inch. Multiply by the length for the delay of the trace.
A signal on a transmission line travels at the speed of light divided by the square root of the effective dielectric constant. Stripline is simple. Microstrip needs a model for how much of the field is in air.
For microstrip, the Hammerstad-Jensen fit gives εeff from u = w ÷ h and εr:
The calculator also applies Hammerstad-Jensen’s correction for copper thickness, and for pairs the Kirschning-Jansen odd-mode εeff, which reads up to about 2 % slow against a 2-D field solve with ½ oz or 1 oz copper. A common simpler estimate is 1 + q (εr − 1) with q = ½ (1 + 1 ÷ √(1 + 12 h/w)). For a 0.3 mm trace over 0.2 mm it gives 3.27 against 3.13 from the full model with 1 oz copper. Nearly all of that difference is the copper thickness, which the simple formula ignores.
tpd = √4.4 ÷ 299.79 mm/ns = 6.997 ps/mm. Over 100 mm the delay is 700 ps, and the velocity factor is 0.477.
With w/h = 1.5, εr = 4.4 and 1 oz copper, εeff = 3.13. So tpd = √3.13 ÷ c = 149.8 ps/in (5.899 ps/mm), faster than the stripline above by about 16 %.
On the same stripline, length = 1 ns ÷ 6.997 ps/mm = 143 mm (5.63 in).
tpd = √εr ÷ c. Stripline delay depends only on Dk, so this table works for any trace width. For a laminate not listed, enter its Dk in the stripline propagation delay calculator above.
| εr | Material | ps/in | ps/mm | Velocity factor |
|---|---|---|---|---|
| 4.00 | FR-4, low end | 169.5 | 6.671 | 0.5 |
| 4.20 | FR-4 | 173.6 | 6.836 | 0.488 |
| 4.30 | FR-4 | 175.7 | 6.917 | 0.482 |
| 4.40 | FR-4 | 177.7 | 6.997 | 0.477 |
| 4.50 | FR-4 | 179.7 | 7.076 | 0.471 |
| 4.60 | FR-4, high end | 181.7 | 7.154 | 0.466 |
| 3.66 | RO4350B design Dk | 162.1 | 6.381 | 0.523 |
| 3.55 | RO4003C design Dk | 159.6 | 6.285 | 0.531 |
Hammerstad-Jensen εeff for a 50 Ω trace on each stackup, 1 oz copper, no solder mask. For your own width and height, use the microstrip propagation delay calculator above.
| Stackup | 50 Ω width | εeff | ps/in | ps/mm |
|---|---|---|---|---|
| JLCPCB 4-layer 1.6 mm, 7628 prepreg (h 0.2104 mm, εr 4.4) | 0.372 mm | 3.19 | 151.3 | 5.96 |
| JLCPCB 4-layer, 2116 prepreg (h 0.1164 mm, εr 4.16) | 0.206 mm | 2.98 | 146.3 | 5.76 |
| JLCPCB 4-layer, 3313 prepreg (h 0.0994 mm, εr 4.1) | 0.175 mm | 2.93 | 144.9 | 5.71 |
| JLCPCB 4-layer, 1080 prepreg (h 0.0764 mm, εr 3.91) | 0.135 mm | 2.78 | 141.4 | 5.57 |
| 2-layer 1.6 mm FR-4 (h ≈ 1.53 mm, εr 4.6) | 2.79 mm | 3.42 | 156.8 | 6.17 |
| Rogers RO4350B 0.254 mm / 10 mil (design Dk 3.66) | 0.523 mm | 2.77 | 141 | 5.55 |
Solder mask adds dielectric above a microstrip and slows it a little. FR-4 Dk depends on resin content and frequency, so use your laminate’s number at the signal frequency, and if the core and prepreg differ on a stripline use a thickness-weighted average. The model is quasi-static, so it does not include dispersion. Each trace of a pair can also see a different mix of glass and resin: TI notes the glass has εr of about 6 and the epoxy about 3, and that the difference can cause skew within a pair. The delay in vias and packages is separate, so add it to the trace delay when the total matters.
About 177.7 ps per inch (6.997 ps/mm) on an inner layer with εr 4.4, and about 149.8 ps per inch (5.899 ps/mm) for a 0.3 mm wide microstrip over 0.2 mm.
PCB signal speed is the speed of light divided by √εeff. On an inner FR-4 layer with εr 4.4 that is 0.477 of c, or 143 mm per ns (5.63 in per ns). The 0.3 mm microstrip over 0.2 mm runs faster, at 0.565 of c.
Part of a microstrip’s field runs through air, which lowers the effective dielectric constant. A stripline is fully embedded, so εeff equals εr. In the example above that is εeff 3.13 against 4.4.
Multiply the delay per length by the length: tpd = √εeff ÷ c, and delay = tpd × length. A 100 mm stripline in εr 4.4 has 700 ps of delay.
The signal speed as a fraction of the speed of light, 1 ÷ √εeff. For εr 4.4 stripline it is 0.477, and for the example microstrip 0.565.
Not on stripline, where the delay depends only on Dk. On microstrip it changes slightly: a wider trace keeps more of the field in the dielectric, so it is a little slower. The chart in the calculator shows the effect for your stackup.
About 143 mm (5.63 in) on an inner FR-4 layer with εr 4.4. Signals travel at roughly half the speed of light in FR-4.
On stripline, yes. On microstrip, the odd-mode signal of a pair is faster because more of its field is in air: a 0.2 mm pair with a 0.2 mm gap over 0.2 mm runs at about 140 ps/in, against about 147 ps/in for one trace alone.
See all free electronics tools →
ProtoFlow is free AI PCB design software: describe a circuit and get a schematic with real parts, PCB layout with an autorouter, SPICE simulation and Gerber files. Native KiCad, Altium and EAGLE import and export. Free on macOS, Windows and Linux.
Free to start. No card required.