--- title: "Should you cut ground under connector pads?" topic: "Ground cut" answer: "Yes, on high-speed pairs whose pads are wider than their traces. A hole the size of the pads takes USB-C’s 6% impedance dip to 0.5%, and one 0.2 mm bigger overshoots by 6%." url: "https://nearfieldnotes.com/notes/ground-cut/" site: "Near Field Notes" author: "Michael Walker" note: 1 family: "Signal integrity" published: "2026-11-02" revised: "2026-11-02" revision: 1 video: "https://nearfieldnotes.com/videos/ground-cut/" --- Note 1 · [Signal integrity](https://nearfieldnotes.com/families/si/) · Rev 1 · [Watch the video · 0:12](https://nearfieldnotes.com/videos/ground-cut/) Ground cut # Should you cut ground under connector pads? **Short answer:** Yes, on high-speed pairs whose pads are wider than their traces. A hole the size of the pads takes USB-C’s 6% impedance dip to 0.5%, and one 0.2 mm bigger overshoots by 6%. **Note 1’s opening figure: The hole, growing.** 0.1 mm a step ![Note 1’s opening figure: The hole, growing](https://nearfieldnotes.com/notes/ground-cut/fig-opening.png) *What the figure shows:* USB-C SuperSpeed pads with a hole in the ground under them, stepping from no hole to 0.3 mm bigger than the pads. No hole: −5.6%. A hole 0.2 mm smaller than the pads: −5.1%. A hole 0.1 mm smaller than the pads: −3.6%. A hole the size of the pads: −0.5%. A hole 0.1 mm bigger than the pads: +2.7%. A hole 0.2 mm bigger than the pads: +5.8%. A hole 0.3 mm bigger than the pads: +7.5%. Peak impedance error a 40 ps edge reads at USB-C SuperSpeed pads, 0.1 mm above the ground. ## What a ground cut is *USB-C pads on a JLCPCB 4-layer board* A hole in the ground layer, right under a connector’s pads. It’s a signal integrity fix for high-speed pairs, like USB-C’s 10 Gbit/s [SuperSpeed lines](https://nearfieldnotes.com/glossary/#g-superspeed-pair) and HDMI. The pads are wider than their traces, so over solid ground they add capacitance and the impedance dips. At these speeds a dip bounces part of the signal back. Through the hole the pads see ground farther down, and the impedance holds. **Fig. 1: The layers, pulled apart.** ![Fig. 1: The layers, pulled apart](https://nearfieldnotes.com/notes/ground-cut/fig-1.png) *What the figure shows:* Figure 1: what a ground cut is. USB-C SuperSpeed pads with the board's copper layers pulled apart. Layer 1 has the two signal pads, their traces and the ground pads beside them. Layer 2, the ground 0.1 mm down, has a hole the size of the two pads right under them. Layer 3, ground 1.4 mm down, is solid, and shows through the hole. USB-C SuperSpeed pads with a hole the size of the pads. On the board the layers are stacked tight. They’re pulled apart here to show the hole. **Fig. 1b: Solid ground and a pad-sized hole.** ![Fig. 1b: Solid ground and a pad-sized hole](https://nearfieldnotes.com/notes/ground-cut/fig-1b.png) *What the figure shows:* Figure 1b: USB-C SuperSpeed pads over solid ground and over a hole the size of the pads. The impedance at the pads reads −5.6% and −0.5%. Ground cut ### Specimen 2 holds the impedance within 0.5% **Field marks.** Specimen 2 has a hole in the layer-2 ground under both pads, 0.8 by 1.0 mm. The impedance at the pads stays within 0.5% of the line’s. **Look-alike.** Specimen 1 has the same pads over solid ground. Each pad adds 78 fF to ground, and more capacitance means less impedance: `Z = √(L/C)`. It dips 5.6%. **Range.** Any pad wider than its trace. With the ground 0.1 mm down, a [90 Ω](https://nearfieldnotes.com/glossary/#g-characteristic-impedance) pair’s traces are 0.18 mm wide, narrower than USB-C, HDMI and choke pads and the pads of 0402 and 0201 capacitors. **Habitat.** The SuperSpeed pair of a USB-C receptacle: 0.30 by 1.00 mm pads on a 0.5 mm pitch. A JLCPCB 4-layer board. Key to ground cut ## Do these pads need a hole in the ground? 1. 1a Pads less than 0.05 mm wider than their traces No hole 2. 1b Pads 0.05 mm or more wider Go to 2 3. 2a Connector or choke pads A hole the size of the pads 4. 2b A series 0402 or 0201 capacitor A hole under the part, 0.1 mm bigger With the ground 0.1 mm down, the traces are 0.17 to 0.21 mm wide. With it 0.2 mm down they are 0.32 to 0.41 mm, as wide as most pads. ## Size the hole yourself *5 footprints, 2 ground depths* **Note 1’s explorer: Ground cut.** ![Note 1’s explorer: Ground cut](https://nearfieldnotes.com/notes/ground-cut/fig-explore.png) *What the figure shows:* The footprint seen from above, 3.6 mm across: copper pads on the top layer, the ground plane below, and the hole in it. *What the figure shows:* Impedance error at the pads against the hole's size, for each footprint and board. The line crosses zero near the size of the pads and keeps climbing as the hole grows. Peak impedance error a 40 ps edge reads at the pads. JLCPCB 4-layer boards. **Fig. 1c: What a fast edge reads.** ![Fig. 1c: What a fast edge reads](https://nearfieldnotes.com/notes/ground-cut/fig-1c.png) *What the figure shows:* Figure 1c: the impedance a 40 ps edge reads as it passes USB-C pads. With no hole it dips to 86.5 Ω, with a pad-sized hole it stays near 91.7 Ω, and with a hole 0.2 mm too big it climbs to 97.0 Ω. Impedance a 40 ps edge reads as it passes USB-C pads, 0.1 mm above the ground. 0 ps is the pads. [Note 16](https://nearfieldnotes.com/notes/tdr/) shows how to read a trace like this. Hole size ### A hole 0.2 mm too big is as far off as no hole **Field marks.** The hole runs 0.2 mm past the pads. The impedance climbs 5.8% above the line’s. **Look-alike.** No hole dips 5.6% below it. A hole the size of the pads stays within 0.5%. **Range.** Each 0.1 mm past the pads adds about 3% under USB-C pads, 7% under HDMI pads and 11% under choke pads. Over the hole the pads lose capacitance and the line gains inductance, so *Z* climbs past the line’s. Making the hole bigger to be safe makes it worse. #### Conditions - **Footprints:** USB-C SuperSpeed pads 0.30 by 1.00 mm on a 0.5 mm pitch. HDMI and DisplayPort pads 0.30 by 1.90 mm. 0402 and 0201 capacitors on KiCad’s pads. 0805 choke pads. - **Board:** JLCPCB 4-layer, 1.6 mm: JLC04161H-3313 with the ground 0.10 mm down, JLC04161H-7628 with it 0.21 mm down - **Route:** 90 Ω differential, 100 Ω for HDMI, widths from a 2D field solver - **Hole:** In the layer-2 ground, with layer 3 solid below it - **Measure:** Peak impedance a 40 ps edge reads at the footprint, against the line’s own - **Solver:** openEMS, 327 runs in all. Copper is a perfect conductor and the laminate has no loss. - **Check:** A finer mesh moves the USB-C results by 0.2% or less Simulated only. Nothing on this page has been measured on a board yet. #### Limits - Pads only. The connector’s pins and shell, and the vias around them, aren’t in the model. - ESD diodes add 3 to 7 times a pad’s capacitance. A hole 3 mm long takes a 0.2 pF diode’s dip from 17% to 6%, and a 0.5 pF diode’s only from 34% to 29%. - Noise on both lines together leaks into the gap between layers 2 and 3, 0.1 to 0.9% of it below 10 GHz with a hole 0.1 mm past the pads. Under 0.1% radiates. Keep the hole centered under the pair. #### Files - padsim/geom.py Every footprint and hole, as the solver saw them - runs/results.json All 327 runs #### Sources - [JLCPCB’s stack-ups, read from its own stack-up data](https://github.com/gsuberland/jlcpcb_autogenerated_stackups) - [openEMS, the field solver](https://github.com/thliebig/openEMS) What should the next note be? [Ask for one](https://nearfieldnotes.com/#ask-for-a-note) --- Source: https://nearfieldnotes.com/notes/ground-cut/ Cite as: Michael Walker, “Ground cut”, Near Field Notes, Note 1, Rev 1, 2 Nov 2026. https://nearfieldnotes.com/notes/ground-cut/ Figures: free to reuse under CC BY 4.0 (https://nearfieldnotes.com/license/). Each one's image is linked under its name. Credit Michael Walker, Near Field Notes, and link the figure: this address plus #fig- and its number, like https://nearfieldnotes.com/notes/ground-cut/#fig-1.