NR-NTN Terminal PCB: Board-Level Evidence for Satellite Direct-to-Device
NR-NTN terminal PCB guide: S-band direct-to-device boards, LEO Doppler arithmetic, feed loss vs EIRP, thermal match drift and certification-grade evidence.
Table of Contents
5G satellite direct-to-device has spent years as a demonstration. That phase is ending: a national certification body in Korea has adopted a full NR-NTN conformance suite for satellite devices and networks — RF and radio-resource-management test cases, pre-consistency and formal acceptance — building on the reported 3GPP Release 19 S-band validations. When acceptance moves into third-party labs, the terminal’s PCB stops being a design detail and becomes an evidence chain.
This page computes that chain from the board side: the LEO link arithmetic that the modem corrects and the board must hold; the S-band geometry that looks forgiving but still prices every millimetre; the feed-loss budget that decides EIRP and battery life; the thermal drift that can walk the antenna match out of a 5G channel; and the mapping from fabrication records to the conformance items the lab will measure. Every figure is computed; the assumptions are stated where they are used.
The spectrum-neighbour topics are treated separately: terrestrial 5G millimetre-wave boards in our 5G mmWave PCB review, module-level manufacturing in the 5G communication module PCB design guide, and the space-side context in the satellite communication PCB review. This page is about the terminal: the board group behind an NR-NTN direct-to-device device.
1. What NR-NTN Certification Changes for the Terminal Board
A direct-to-device terminal is a board group with four disciplines, and certification measures all of them as built. The table maps it.
| Board in the NR-NTN terminal | What it carries | What the fabricator must confirm |
|---|---|---|
| RF front-end board | S-band PA/LNA, duplexer and filtering, antenna switch, feed to the antenna region | Feed-loss budget and match stability across temperature — both are certification inputs |
| Baseband / modem board | NTN modem, application processor, memory, high-speed digital routing | Digital return isolation from the RF carrier; power integrity on the modem rails |
| Power management layer | PA supply sequencing, battery rails, charging path | PA current draw and rail integrity set transmit phase behaviour under load steps |
| GNSS co-board | L1 receive chain sharing the antenna environment with the S-band front end | Isolation between receive chains; the passive GNSS path tolerates no intermodulation |
5G satellite direct-to-device just crossed a line: a national certification body in Korea has adopted an NR-NTN conformance suite — extending acceptance testing for satellite devices and networks, building on the reported 3GPP Release 19 NR-NTN S-band validations and covering GCF-approved RF and RRM test cases. The moment conformance goes third-party, the board group stops being a design detail and becomes an evidence chain. This page computes the board-level arithmetic: what the LEO link demands, what the radio compensates, what stays on the board, and which fabrication records map to which test items. The satellite-side context is set out in our satellite communication PCB review; the terminal-side sibling in the VSAT and LEO terminal PCB guide.
Two rows carry most of the certification risk. The RF front-end board owns the feed loss and the antenna match — both measured over the air by the conformance suite. And the power management layer quietly owns transmit phase behaviour: PA supply rails that sag under load steps show up as EVM. A fabricator who treats the front-end board as “just a 2 GHz board” has not read the test plan.
2. The LEO Link: Doppler the Radio Handles, Phase the Board Must Hold
The headline number of a LEO link is Doppler, and the first thing to know is whose problem it is.
| LEO link parameter (S-band, 550 km circular) | Value | What it means for the board |
|---|---|---|
| Doppler shift ±Δf | ±50–55 kHz | 0.8–3.3× the NR subcarrier spacing (15/30/60 kHz) — the modem compensates in frequency |
| Doppler rate | ±54–60 Hz/s | Continuous tracking through a pass of minutes, not a one-time offset |
| Link visibility | Minutes per pass | Acquisition and re-acquisition must be fast — reference stability is part of that |
| What remains on the board | Clock reference integrity and phase stability of the RF path | The modem corrects frequency; it cannot correct a noisy or drifting reference |
Δf = f·v/c and df/dt = f·a/c with v = 7.5 km/s and the radial acceleration of an overhead pass — the arithmetic of a 550 km orbit at S-band. The point of the table is the division of labour: Doppler is a radio problem with a known solution, and the PCB’s share is everything the correction assumes — a stable clock distribution, a clean reference path, and RF phase behaviour that does not wander with temperature or supply. Clock and reference discipline for multi-RF designs is treated in our GNSS anti-jamming PCB guide, where precision timing is the product.
The division of labour is the whole story: at S-band a 550 km orbit sweeps ±50–55 kHz of Doppler — several subcarrier spacings — tracked continuously through every pass. The modem’s correction is built on an assumption the board has to deliver: a clean, stable clock reference and an RF path whose phase does not wander with temperature or supply. Clock and reference discipline in a multi-RF terminal is the same craft our GNSS anti-jamming PCB guide treats at higher precision.
3. S-Band Geometry: Roomy, but Still Millimetre Work
After 60 GHz, S-band looks luxurious. The table prices it honestly.
| Geometry at 2.2 GHz (εr 3.48 class) | Value | Design consequence |
|---|---|---|
| Free-space wavelength λ0 | 136 mm | Comfortable layout scale — no mmWave crowding |
| In-medium wavelength λg | 73 mm | Feed and matching sections are tens of millimetres, not microns |
| Phase per millimetre | 4.9°/mm | λg/20 ≈ 3.5 mm is already 18° — matching structures stay dimension-controlled |
| Antenna region discipline | Keep-out, matching network and feed are one tolerance chain | Terminal antennas are certified as built — the match is a fabrication deliverable |
S-band geometry is forgiving compared with 60 GHz — and that is exactly the trap. The phase density is low enough for layout habits to drift, but the antenna match and its feed remain millimetre-scale structures whose behaviour the conformance suite measures: over-the-air matching, sensitivity, and emissions. Terminal antenna layout rules for wireless devices are set out in our antenna PCB for wireless devices guide, and the certification-adjacent checklist (SAR and pre-certification) in the 5G terminal antenna checklist.
The phase density is low enough that layout habits survive — and loose enough that they quietly degrade the antenna match, which the conformance lab then measures over the air. The keep-out, the matching network and the feed are one tolerance chain with dimensions on the drawing; terminal antenna layout method is set out in our antenna PCB for wireless devices guide and the certification-adjacent checklist in the 5G terminal antenna checklist.
4. Feed Loss: Where EIRP and Battery Life Are Won
The uplink to a LEO satellite is EIRP-starved, which turns every tenth of a dB into watts.
| Front-end loss (feed, filters, switch) | Transmit power wasted | Extra PA draw to hold EIRP (26 dBm class, η = 35% assumed) |
|---|---|---|
| 0.5 dB | 10.9% | ≈ 140 mW |
| 1.0 dB | 20.6% | ≈ 295 mW |
| 1.5 dB | 29.2% | ≈ 470 mW — on a battery device, per transmit symbol |
| 2.0 dB | 36.9% | ≈ 665 mW — the difference between a usable link budget and a hot, dying terminal |
Power wasted = 1 − 10^(−loss/10); the PA draw assumes 35% efficiency and a 26 dBm-class handheld transmit stage — a stated calculation assumption. The uplink to a LEO satellite is EIRP-starved: every tenth of a dB in the feed chain is either transmit power, battery heat, or link margin. Feed routing, filter placement and switch insertion loss are therefore stackup-review decisions with the link budget attached — not layout conveniences discovered after the first over-the-air test.
The 1.5 dB row is the sobering one: nearly a third of the transmit power gone before the antenna, recovered only by burning another 470 mW at the PA — per transmit symbol, on a battery. Feed routing, filter and switch placement, and connector transitions are therefore stackup-review decisions with the link budget attached. The module-manufacturing side of this discipline is treated in our RF module PCB manufacturing guide.
5. Thermal Drift: the Match That Walks Out of the Channel
The quietest certification failure is a match that moves with temperature.
| Laminate drift class (Δεr over temperature) | Resonance shift @ 2.2 GHz | Versus a 10–20 MHz NR channel |
|---|---|---|
| Conventional dielectric, Δεr +2% | −21.7 MHz | Matching walks out of the channel across the temperature range |
| Low-drift dielectric class, Δεr +0.5% | −5.5 MHz | Inside the channel budget, with margin for assembly drift |
A patch or matching structure resonates at f ∝ 1/√εeff: a 2% dielectric shift moves resonance by about 1% — 21.7 MHz at 2.2 GHz, which is wider than a full 5G NR channel. The full automotive/industrial temperature range, plus the post-reflow shift, stacks on top. This is why the antenna environment on a certified NTN terminal is specified by drift class and verified with coupons at temperature, not assumed from a room-temperature datasheet. Wide-temperature laminate behaviour is treated in our wide-temperature PCB guide, material classes in the PCB material comparison chart.
The mechanism is simple — resonance scales with 1/√εeff — and the consequence is not: a conventional dielectric shifting 2% moves the match 21.7 MHz at 2.2 GHz, wider than an entire NR channel. The antenna environment is therefore specified by drift class, verified with temperature coupons, and checked again after reflow. Wide-temperature material behaviour is treated in our wide-temperature PCB guide, material classes in the PCB material comparison chart.
6. Certification-Grade Evidence: Mapping Board Data to Test Items
Third-party conformance changes the commercial geometry: the terminal maker’s iteration loop is now measured in certification slots, and the fabricator’s records are part of the loop. The table maps them.
| Fabrication record | Maps to conformance item | Why the link is direct |
|---|---|---|
| Dk/Df batch data + impedance coupons per panel | EVM under frequency offset, reference sensitivity | The modem’s correction budget assumes the chain the boards actually delivered |
| Copper roughness class + etch compensation records | EVM / ACLR across the band | Loss flatness, not just loss at one frequency, sets the modulation quality |
| Swept feed-loss and antenna-match data | Antenna matching consistency, TRP-type metrics | The certified device is measured as built — the match is in the shipping unit |
| Reflow warpage and assembly records | Post-assembly spurious and match drift | Assembly shifts the antenna environment; the record proves control |
| Grounding and shielding plan | Spurious emissions, blocking | Isolation between TX, RX and GNSS chains is a layout deliverable |
The certification moment changes the commercial conversation: pre-consistency and formal acceptance now run through third-party labs, so the terminal maker’s iteration loop is measured in certification slots, not bench hours. A fabricator who ships the evidence chain — batch data, coupons, swept feed measurements, assembly records — shortens that loop measurably. The module-level version of this discipline is treated in our RF module PCB manufacturing guide; the mmWave-side measurement method in the controlled impedance and 50 ohm trace guides.
Read the table bottom-up and it is the DFM list; read it top-down and it is the certification plan. Either way the conclusion is the same: a fabricator who ships batch data, coupons and swept feed measurements with the boards is not selling paper — they are shortening the terminal maker’s path through the lab.
7. The DFM Checklist Before the NR-NTN Terminal RFQ
Nothing here is exotic; each line traces to a number computed above, and each is cheap on a drawing and expensive to reconstruct after a certification slot is missed.
| # | Drawing or PO line item | Why it is on the list |
|---|---|---|
| 1 | Antenna keep-out and matching network as one tolerance chain, dimensions on the drawing | The certified match is a fabrication deliverable (section 3) |
| 2 | Feed-loss budget attached to the link budget at stackup review | 1.5 dB of feed = 29% of transmit power wasted (section 4) |
| 3 | Laminate drift class specified for the antenna environment | Δεr +2% walks the match out of a 10–20 MHz channel (section 5) |
| 4 | Clock reference routing and return integrity as a specified structure | The modem corrects Doppler; it cannot correct a noisy reference (section 2) |
| 5 | GNSS co-chain isolation from the S-band front end | The passive GNSS path tolerates no intermodulation (section 1) |
| 6 | Dk/Df batch records and impedance coupons shipped with the lot | Conformance labs test the built device; the records are the traceability (section 6) |
| 7 | Post-reflow match/warpage verification on first articles | Assembly is part of the antenna environment (section 5) |
| 8 | Grounding/shielding plan between TX, RX and GNSS chains | Spurious and blocking are layout outcomes, not afterthoughts (section 6) |
Eight lines, each tracing to a number computed above. Items 1–5 decide whether the terminal meets the link; 6–8 decide whether the evidence exists when the certification slot arrives. An NR-NTN terminal RFQ that carries these eight answers is quotable on engineering.
The pattern across the whole page: direct-to-device terminals are certified as built, and the build is the PCB. For the adjacent decisions — the space-side stackups that carry the satellite link, the terrestrial mmWave boards sharing the 5G family, or the material classes behind the drift budget — the LEO satellite PCB design guide, the 5G mmWave PCB review, the VSAT and LEO terminal PCB guide and our RF PCB manufacturer overview are the right starting points.
NR-NTN Terminal PCB — Frequently Asked Questions
What NR-NTN certification changes, what the modem handles vs the board, thermal match drift, the evidence chain, and what the RFQ should contain.
What is NR-NTN, and why is its certification a PCB topic?
NR-NTN is the 3GPP 5G New Radio profile for non-terrestrial networks — direct-to-device service over satellite links, with S-band connectivity validated in Release 19. The milestone is institutional: a national certification body has adopted an NR-NTN conformance suite covering GCF-approved RF and RRM test cases, so acceptance moves from vendor demonstrations to third-party labs. The moment that happens, the terminal PCB is judged by measured, repeatable conformance items — EVM, sensitivity, matching consistency, spurious — which are set by fabrication quality and material stability.
The modem compensates Doppler — what is left for the PCB?
Frequency correction assumes a clean reference. At S-band, a 550 km orbit produces ±50–55 kHz of Doppler — several subcarrier spacings — tracked continuously through each pass. The radio handles that arithmetic; the board's job is everything the correction presumes: a stable clock distribution, a reference path immune to supply and thermal noise, and RF phase behaviour that does not drift. A noisy or drifting reference shows up exactly where conformance measures it — acquisition time, EVM under frequency offset, sensitivity.
Why does antenna matching drift over temperature, and how much is tolerable?
A patch or matching network resonates at f ∝ 1/√εeff, so dielectric drift moves the match. A conventional dielectric shifting 2% moves resonance about 21.7 MHz at 2.2 GHz — wider than an entire 10–20 MHz 5G NR channel — while a low-drift class at 0.5% stays inside the budget with margin. The full temperature range plus post-reflow shift stack on top, which is why the antenna environment is specified by drift class and verified with temperature coupons.
What evidence chain should come with an NTN terminal board?
The records that map to conformance items: Dk/Df batch data and impedance coupons per panel (EVM, sensitivity), copper roughness class and etch compensation records (loss flatness, EVM/ACLR), swept feed-loss and antenna-match measurements (matching consistency), reflow and assembly records (post-assembly drift and spurious), and the grounding/shielding plan (blocking, emissions). A fabricator shipping this chain shortens the terminal maker's certification iteration loop measurably.
What should an NR-NTN terminal RFQ include?
The conditions that change the board: the S-band band plan and channel bandwidth, the antenna architecture and its keep-out, the temperature range and drift class for the antenna environment, the GNSS co-existence requirement, the transmit power class and battery constraints, and the acceptance evidence you expect — swept feed data, impedance coupons, batch traceability, post-reflow verification. Suppliers who ask for these before quoting are pricing engineering, not guessing.
Building an NR-NTN or direct-to-device terminal board?
Send us the S-band band plan, the antenna keep-out and match architecture, the temperature range and drift class, and the GNSS co-existence requirement. We will review the feed-loss budget, the clock reference structure and the antenna-environment laminate at stackup review, and ship the evidence chain — batch data, coupons, swept feed measurements — with the boards to IPC Class 3.
Contact Us → WhatsApp: +86 137 6047 3650Request a PCB Quote
Upload your Gerber ZIP file and project requirements. Our engineering team will review your PCB material, stackup, impedance needs, surface finish, and production quantity before quoting.
Please prepare:
- Gerber files in ZIP format
- PCB material or stackup requirements
- Controlled impedance notes if available
- Prototype or batch production quantity