OSH Park 4-layer stackup: layers, εr, 50 Ω and 90 Ω widths

OSH Park 4-layer is OSH Park's 1.6 mm 4-layer build: 7.87 mil of FR408HR 2113 prepreg at εr 3.61 under layer 1, over a 39 mil core. A 50 Ω microstrip on layer 1 is 15.19 mil (0.386 mm) with the solder mask at 1 GHz.

Open OSH Park 4-layer in the Transmission Line Calculator

Layers

LayerMaterialThicknessεrCopper
MaskSolder mask0.6 mil3.9
L1Copper1.7 mil1 oz finished
L1 to L2FR408HR 2113 prepreg7.87 mil3.61
L2Copper0.0175 mm0.5 oz
L2 to L3FR408HR core39 mil3.87
L3Copper0.0175 mm0.5 oz
L3 to L4FR408HR 2113 prepreg7.87 mil3.61
L4Copper1.7 mil1 oz finished
MaskSolder mask0.6 mil3.9

The stackup table prints 7.87 mil (0.1999 mm) between layer 1 and layer 2. OSH Park states the prepreg Dk at 1 GHz. The core and mask Dk on the construction diagram carry no frequency.

The stackup table leaves the core's Dk blank. The construction diagram linked under it prints 3.87, and this preset uses that. The table prints no Dk for the solder resist. The construction diagram prints 3.90 for the purple Taiyo mask at 0.60 mil, and this preset uses that. Copper: the table prints 1.7 mil outer and 0.68 mil (0.0175 mm) inner; the diagram prints 1.80 and 0.60 mil. This preset uses the table. The diagram also prints the lower prepreg as 7.86 mil against the table's 7.87.

Trace widths on layer 1

LineWidthSolvedCalculator
50 Ω microstrip, with mask15.19 mil (0.386 mm)50.00 ΩOpen
50 Ω microstrip, mask off15.95 mil (0.405 mm)50.00 ΩOpen
90 Ω pair, 5 mil gap, with mask9.49 mil (0.241 mm)90.00 ΩOpen
100 Ω pair, 5 mil gap, with mask6.92 mil (0.176 mm)100.00 ΩOpen

Bare copper at 1 GHz, from a 2-D field solve of the cross-section with the mask coating the trace top and sidewalls. The 5 mil gap is a typed value, because the stackup leaves the gap to you. Ask the fab what space it will etch, then open the row and type that.

OSH Park's own impedance table

TargetOSH Park widthGapOSH Park predictsThis page
50 Ω single15.0 mil49.8 Ω15.19 mil
90 Ω pair9.0 mil5.0 mil91.5 Ω9.49 mil
100 Ω pair6.9 mil5.1 mil100.3 Ω6.92 mil

The impedance table on OSH Park's service page lists its own layer-1 widths with the mask. Its pairs use their own gap, which is not always 5 mil, so open a row above and type their gap to compare a pair exactly.

Substitute-material builds

OSH Park has built some orders on EM528 during FR408HR shortages. The mixed build (June and late July 2026) keeps this 7.87 mil FR408HR prepreg under layer 1 and swaps only the core for a red EM-528 core at Dk 4.11, so the layer-1 widths here still apply. The full EM528 build on OSH Park's alternate-stackup page, offered at checkout since 2026-09-21, puts 7.67 mil of EM-528 2x3313 prepreg at Dk 3.76 under layer 1 instead, so these widths do not apply to it. A note for orders on that build from 2025-12-20 to 2026-01-28 is no longer shown on the service page.

What OSH Park does not print

No typical number fills these, and none of them sets a width on this page.

The same 50 Ω line on the other 4-layer builds

The prepreg under layer 1 sets the width. A thinner prepreg gives a narrower line for the same impedance, and builds with the same prepreg under layer 1 give the same width.

Sources

How IndieRF checks its numbers: each engine against an outside answer, with the tolerance and what it does not cover.

KiCad vs JLCPCB 50 Ω width90 Ω USB pair width50 Ω microstrip width on FR-4Question about this stackup, or a correction?