Type-C Orientation & Signal MUX (TX/RX Flip + DP Alt-Mode)
← Back to: USB / PCIe / HDMI / MIPI — High-Speed I/O Index
This page defines the routing and control rules that make USB/DP work identically in both Type-C plug orientations—by enforcing correct lane mapping, stable MUX timing, and a clean AUX/HPD/SBU sideband path.
The goal is measurable: no orientation-dependent failures, no hot-plug glitches, and a verified margin budget around the MUX with clear pass criteria.
H2-1. Definition & Scope: What “Orientation & Signal MUX” Actually Covers
This page focuses on the Type-C routing layer: how a design flips and routes SuperSpeed TX/RX pairs, DP lanes, AUX/HPD, and SBU paths so both plug orientations behave identically. The output is a set of routing rules, control rules, and pass/fail criteria that make bring-up predictable.
- SS TX/RX differential pairs: lane mapping, polarity correctness, and flip equivalence (A-side vs B-side).
- DP Alt-Mode lane reuse: lane assignment patterns, muxing constraints, and “same internal mapping” goals.
- AUX/HPD routing: correct path selection and stable hot-plug behavior (routing integrity only).
- SBU paths: low-leakage / low-capacitance constraints and symmetry (to avoid breaking AUX-like signaling).
- Control interface: CC-based orientation decision → MUX control (GPIO/I²C), fail-safe defaults, glitch-free switching.
The routing boundary is normally between an internal PHY/SerDes (or a retimer) and the Type-C receptacle: SoC / PHY / Retimer ↔ (Orientation MUX) ↔ Type-C receptacle. This boundary is where orientation equivalence is enforced and where a small routing mistake can look like “random link flaps.”
- Routing rules
- Explicit lane mapping / polarity / path constraints so A-side and B-side are electrically equivalent.
- Control rules
- Orientation decision → MUX switching timing, default state, and glitch-free requirements.
- Pass / fail criteria
- Measurable acceptance checks (training stability, AUX/HPD behavior, margin placeholders) that isolate routing-layer faults early.
- USB-C PD / VBUS power path: negotiation, load switches, power budgeting, back-drive blocking.
- Retimer / redriver tuning: CTLE/DFE settings, link training visibility, margining sequences.
- ESD/TVS part selection & IEC details: device series selection, gun setup, waveform-level compliance procedures.
- Protocol stacks: USB enumeration/class details, DP protocol/EDID/HDCP deep dives, bridge feature sets.
Cross-protocol mentions (USB/DP) are used only to define routing objects (lanes, AUX/HPD, SBU) and acceptance checks. Any protocol-specific mechanisms remain on their dedicated pages.
H2-2. Why It’s Hard: Failure Modes That Look Like “Random Link Flaps”
- Lane mapping / polarity mismatch: the two orientations are not electrically equivalent (pair swap is incomplete or inconsistent).
- Control race / wrong default state: MUX selection changes during a sensitive window, or the power-off default is unsafe.
- AUX/HPD routed incorrectly: DP sideband signals follow the wrong path or are loaded by a switch/protector in a way that breaks detection.
- SBU clamp / leakage / asymmetry: a “small” protection or switch choice distorts low-speed sideband behavior (especially AUX-like signaling).
- Margin pushed over the edge: MUX insertion loss + crosstalk increases sensitivity; long channels reveal the deficit.
When a failure correlates with orientation, mode, or hot-plug timing, the fastest path is to verify the routing equivalence and the MUX control timeline before blaming software or higher-level protocol logic.
These checks classify the problem into one of four buckets: mapping/polarity, sideband routing, switching timeline, or margin. Each bucket maps cleanly to later chapters (control plane, DP routing, SBU constraints, SI budgeting) without expanding into unrelated protocol/power content.
H2-3. Type-C Signal Anatomy: Pins, Pairs, and What Must Flip
This chapter defines only the signals that change routing when a Type-C plug flips: SuperSpeed differential pairs, DP lane reuse, and the sideband group (CC, SBU, AUX/HPD as routing objects). USB2 D+/D− is acknowledged only as a parallel path and is not expanded.
Only routing equivalence is described here. Protocol negotiation details and power policy remain out-of-scope by design.
- SS TX/RX pairs: the A-side and B-side pin groups map to different external pair positions, so the routing layer must enforce a consistent internal port naming across both orientations.
- DP lane reuse: DP Alt-Mode reuses SuperSpeed pair resources as DP lanes, so the MUX must support lane assignment without breaking the orientation mapping.
- AUX/HPD as routing objects: these sideband paths must land on the correct receiver pins for the active orientation/mode.
For both orientations, the design must expose an identical internal view: the same “lane names” connect to the same functional blocks, and the system achieves the same success rate for X attach cycles under Y operating conditions (placeholders).
DP Alt-Mode typically reuses Type-C high-speed resources by remapping SuperSpeed pairs into DP lanes. The key routing-layer requirement is that lane reuse must remain orientation-consistent and keep AUX/HPD landing on the correct sideband path. Details such as EDID/HDCP handling remain on their dedicated pages.
H2-4. Control Plane: How Orientation Is Decided (CC) and How MUX Is Driven
- GPIO-based select (SEL/EN): ensure deterministic power-on defaults, avoid glitches during reset, and keep selection stable across sensitive windows.
- I²C configuration: treat writes as part of a timed sequence (ack/retry and readiness gates) so the MUX never briefly enters an unintended state.
- Event boundaries: attach/detach and mode-entry transitions must be explicit in the control timeline, even if the platform hides the details.
- No SEL/EN switching inside a X ms window around attach/mode-entry (placeholder).
- Exactly one orientation decision per attach; orientation remains locked until detach.
- Sideband behavior (AUX/HPD) shows no unintended pulses beyond X events per Y cycles (placeholders).
- Defined reset state: the MUX should have a known default that does not create unintended cross-connections or sideband disturbances.
- No accidental toggling: during rail ramp and reset release, SEL/EN must not glitch into intermediate states.
- Recoverable sequencing: if a control write fails (I²C), the system must converge to a stable routing state rather than oscillate.
This chapter defines control timing and default-state behavior only for reliable routing. PD policy and power-path details remain on separate pages.
H2-5. TX/RX Flip for USB 3.x/USB4: Lane Mapping Rules That Never Lie
Type-C orientation changes which receptacle pin group is active. A robust design enforces an orientation-equivalent mapping: regardless of A-side or B-side insertion, the same internal ports connect to the same functional blocks (TX/RX role preserved at the routing layer).
This section defines routing invariants and control timing constraints only. Retimer/PHY tuning (EQ/DFE/CTLE) remains on dedicated pages.
- A-side vs B-side attach success delta ≤ X% over Y cycles.
- Mapping changes per attach = 0 inside the stable window (see next card).
- No unexpected retrain events beyond X per Y minutes (platform-defined).
- Allowed switching: only before link entry or after detach, when the platform defines the route as idle.
- No-switch window: once orientation is locked, SEL/EN must remain stable for X ms (placeholder) around link entry.
- Fail-safe default: on reset/power-off, the default route must not create unintended cross-connections that confuse the platform.
H2-6. DisplayPort Alt-Mode Routing: Lanes, AUX/HPD, and the “Gotchas”
DP Alt-Mode reuses high-speed resources by assigning Type-C lanes to DP lanes through a crossbar/MUX. The routing-layer requirement is not the DP protocol itself, but the ability to maintain consistent lane assignment across orientation changes and mode entry.
EDID/HDCP and DP policy are intentionally excluded. This section focuses only on lane muxing plus AUX/HPD routing consistency.
- Orientation-consistent assignment: A/B insertion must map into the same internal DP lane naming.
- Single control decision per attach: choose lane assignment once per attach/mode-entry, then lock until detach.
- No-switch window applies: do not change MUX state inside the stable window around mode entry (X ms placeholder).
- DP bring-up success rate is orientation-symmetric: delta ≤ X% over Y cycles.
- HPD unintended pulses ≤ X per Y cycles.
- AUX transaction retry budget ≤ X per Y seconds (platform-defined).
- MUX state changes inside mode-entry stable window = 0.
H2-7. SBU Path & Protection: What to Protect, What Not to Break
The SBU path often carries low-speed or analog-sensitive sideband functions (e.g., AUX / accessory / vendor-defined use). For design review, the dominant risks are not protocol features, but unintended loading (capacitance/leakage), asymmetry between SBU1/SBU2, and placement/return-path mistakes that turn protection into a failure source.
This section defines constraints and acceptance criteria only. TVS part numbers and IEC gun test procedures belong to the USB Port ESD/TVS page.
Why: extra C reshapes edges and analog-sensitive signaling margins.
Acceptance: orientation-symmetric behavior across worst-case cable/adapter set; no “one-side-only” fragility.
Why: orientation and mode transitions amplify asymmetry into intermittent, orientation-dependent failures.
Acceptance: A/B plug orientation pass rate delta ≤ X% over Y cycles (placeholder).
Why: a long or ambiguous return path turns protection events into coupling/noise injection.
Acceptance: layout gate: verify protector-to-connector proximity and a single, low-inductance return path.
- Default state must be safe: on reset or power-off, avoid unintended shorting or pulling SBU into a wrong domain.
- No-switch window applies: do not change SBU selection inside mode-entry and orientation-lock windows (X ms placeholder).
- One path per attach: select SBU routing once per attach/mode entry and hold until detach.
Device part numbers and IEC details are intentionally moved to the USB Port ESD/TVS page to avoid scope overlap.
- SBU effective capacitance ≤ X pF (placeholder).
- SBU1/SBU2 symmetry: ΔC / Δleak ≤ X (placeholder).
- Protector placement: connector-side, with a single explicit return path (layout gate).
- Orientation symmetry: A/B pass-rate delta ≤ X% over Y cycles.
- No SBU route changes inside mode-entry stable window; changes per attach = 0.
H2-8. MUX Device Taxonomy: Passive vs Active, Crosspoint vs 2:1, What Each Buys You
Selection becomes deterministic when device classes are separated by what they do at the routing layer: passive pair switching, active switching with conditioning, and retimer+MUX combos that add re-timing behavior. This section defines trade-offs, interfaces, and fail-safe rules only.
Retimer/Redriver tuning (EQ/DFE/CTLE presets, margining) is intentionally excluded and belongs to the Retimer/Redriver page.
- What it buys: improved channel tolerance and more robust operation across longer routes or adapter/cable variance (platform-defined).
- What it does not fix: incorrect lane mapping, inconsistent AUX/HPD routing, or control-plane switching instability.
- Interface reality: SEL/EN and/or I²C control, with default-state behavior that must match system reset sequencing.
Any mapping device (passive or active) must obey a stable no-switch window around orientation lock and mode entry (X ms placeholder).
- On power-off: confirm whether the device opens all paths or forces a default connection; avoid accidental cross-connection.
- During reset: avoid transient wrong mapping that could be interpreted as a valid mode entry.
- Hi-Z windows: ensure the platform does not see undefined behavior that triggers unstable state transitions.
- Control pin biasing: define pull-ups/pull-downs so the default state is deterministic.
- If routing is short and mode set is limited, then a passive 2:1 class is often sufficient (validate symmetry and loss margin).
- If the channel is longer or variability is high, then active switching can improve tolerance (interfaces and reset discipline become critical).
- If many mode/lane combinations are required, then crosspoint approaches offer extensibility (requires stricter invariants and validation).
- A/B orientation pass-rate delta ≤ X% over Y cycles.
- Mode-entry MUX state changes inside stable window = 0.
- Lane/switch symmetry within platform-defined thresholds (ΔIL/ΔC ≤ X).
H2-9. Signal Integrity Around the MUX: Budgets, Placement, and Layout Rules
A Type-C orientation MUX turns the channel into three measurable segments: pre-segment (SoC/retimer-side routing), the MUX segment (device + pads), and the post-segment (port-side escape toward the connector). Each segment contributes insertion loss (IL), return loss (RL), and crosstalk (XTALK). Orientation robustness requires the A/B paths to remain symmetry-closed.
This section focuses on MUX-adjacent budgets and layout invariants. It does not repeat a full SI textbook or protocol compliance workflow.
- Post-segment becomes short and controlled; connector-side discontinuities are easier to bound.
- Pre-segment becomes longer; IL/XTALK sensitivity increases on the SoC-side route.
- Control lines may traverse noisier regions; require stronger glitch-free control discipline.
- Pre-segment is short and lower loss; training is typically more tolerant to internal routing variance.
- Post-segment grows and approaches the connector/EMC zone; port-side layout must be highly disciplined.
- Orientation symmetry must still be guaranteed by construction (avoid flip-dependent geometry).
No placement is universally correct. Choose the placement that closes the segmented budget (IL/RL/XTALK) while preserving a stable control window and consistent A/B symmetry.
- Eye margin ≥ X (placeholder) in both orientations.
- BER ≤ X (placeholder), or error counters within Y-minute window = Z (placeholders).
- Training stability: success rate ≥ X% across A/B and worst-case cable/adapter set.
- Segment symmetry: ΔIL / ΔRL / ΔXTALK bounded by X thresholds (placeholders).
H2-10. Bring-up & Debug Playbook: Make Orientation Bugs Impossible to Hide
- Define a minimum cable/adapter set (short / medium / worst-case).
- Freeze power and thermal conditions for baseline runs (placeholders).
- Enable logs for orientation detect, MUX select, mode entry, and settle timestamps.
- Expose a force-control method (GPIO/I²C) to lock MUX state for A/B testing.
The smallest “cannot-hide” set is 2 (A/B) × 2 (USB/DP) × 2 (cold/hot) = 8 runs per build.
- Control plane: orientation detect state, mux select, mode-entry state, and settle-done timestamps are consistent and monotonic.
- Routing/mapping: A and B orientations map to the same internal port view (no hidden polarity swap, no missing sideband path).
- Signal integrity: segmented budgets around the MUX remain within placeholders; failures correlate to a segment overflow rather than “randomness”.
- A/B orientation: training success rate ≥ X% across the defined worst-case set.
- Forced-state runs: each forced selection yields a stable and explainable outcome (no “random”).
- Logs: every failure aligns with a recorded state tuple (orientation/mux/mode) and a bounded time window.
- Mode entry: mux_select changes inside stable window = 0.
H2-11. Applications & IC Selection Logic (Stay Within Scope)
Turn real use-cases into a MUX class decision (without drifting into PD/ESD/retimer tuning)
This chapter compresses the page into a repeatable decision path: scenario → requirements → device class → pass criteria. Only the routing/orientation layer is covered: SuperSpeed lane flipping, DP Alt-Mode lane muxing, and consistent AUX/HPD/SBU paths.
- In scope: high-speed MUX/crossbar selection, sideband (AUX/HPD/SBU) switching requirements, fail-safe/default-state rules, and orientation bring-up criteria.
- Out of scope: USB-PD policy/state details, IEC/TVS component sizing, and retimer parameter tuning (linked to sibling pages when needed).
Laptop/Tablet Type-C (USB + DisplayPort Alt-Mode, dual-stack)
- Use case: one receptacle must support USB SuperSpeed and DP Alt-Mode, in both plug orientations.
- MUX pain: orientation mapping + lane muxing must remain equivalent; AUX/HPD must always reach the correct endpoint.
- Hard requirements: crossbar support for USB+DP lane combos; deterministic sideband path (AUX/HPD over SBU or dedicated path); fail-safe default that does not create “ghost” links during attach/detach windows.
- SI pressure: moderate; passive is viable when the channel is short and insertion loss margin is healthy (thresholds handled by the SI chapter).
- Minimum validation: A/B insertions × (USB only / DP Alt-Mode) × (auto control / forced MUX states) must produce identical link outcomes.
- Passive crossbar / Alt-Mode MUX: TI HD3SS460, TI TMUXHS4446, Diodes PI3USB31532, NXP CBTL08GP053
- Active redriving switch (when margin is tighter): TI TUSB546A-DCI (or TUSB546-DCI), Parade PS8740
- Low-speed SBU switching (if not integrated): TI TS3USBCA4
Dock/Hub Upstream Port (longer channel, more connectors, higher interoperability risk)
- Use case: upstream Type-C must survive worst-case cables, dock internal routing, and multiple connector discontinuities.
- MUX pain: passive switching can consume the eye margin; orientation mapping bugs look like “random flaps” under load.
- Hard requirements: crossbar + stable sideband path; attach/detach must be glitch-free and fail-safe under power sequencing.
- SI pressure: high; favor integrated redriver/retimer solutions when passive loss + reflections exceed the budget.
- Minimum validation: A/B insertions + worst cable + temperature sweep must keep link training stable (no orientation-dependent regression).
- Active redriving switch: TI TUSB546A-DCI, Parade PS8740, Diodes PI2DPX1217XUAEX / PI2DPX1263XUAEX
- Retimer option (when redriving is not enough): Parade PS8828A, Parade PS8830 / PS8833, Diodes PI2DPT1021Q
- If a discrete crossbar is still used: TI TMUXHS4446, Diodes PI3USB31532
Industrial Type-C (EMC/ESD stress, strict fail-safe behavior, serviceability)
- Use case: a field port must stay stable across noise events and power transients, and recover deterministically.
- MUX pain: default state mistakes create intermittent attach behavior; sideband leakage/capacitance can break AUX.
- Hard requirements: explicit powered-down behavior (open/Hi-Z or defined connect), predictable switching (no glitch), and sideband symmetry (AUX/HPD path integrity).
- SI pressure: depends on enclosure/cable; prioritize devices with clear fail-safe modes and robust control behavior.
- Minimum validation: forced-state debug must isolate “mapping vs control vs SI” within one session (no hidden orientation bugs).
- Crossbar MUX with sideband support: TI TMUXHS4446, NXP CBTL08GP053, Diodes PI3USB31532
- Discrete SBU/AUX switching (when required): TI TS3USBCA4
- Active redriver switch (if the channel is harsh): TI TUSB546A-DCI, Parade PS8740
Four knobs that fully determine the MUX class (no extra theory needed)
Reference IC list by device class (for BOM kickoff)
The list below stays within this page’s scope: orientation switching, lane muxing, and sideband path integrity.
- Texas Instruments HD3SS460 (USB-C Alt-Mode passive crosspoint)
- Texas Instruments TMUXHS4446 (USB-C crossbar; includes low-speed SBU switching support)
- Diodes Incorporated PI3USB30532 (USB3/DP crossbar switch family)
- Diodes Incorporated PI3USB31532 (USB3.2 Gen2 / DP1.4/DP2.1 UHBR10 class crossbar)
- NXP CBTL08GP053 (USB-C high-performance crossbar with sideband switching)
- NXP CBTL06GP213 (general-purpose high-speed multiplexer family, usable for USB3/DP)
- Texas Instruments TUSB546-DCI / TUSB546A-DCI (USB-C DP Alt-Mode linear redriver crosspoint switch)
- Parade Technologies PS8740 (USB-C redriving switch for USB/DP Alt-Mode)
- Diodes Incorporated PI2DPX1217XUAEX / PI2DPX1263XUAEX (USB-C linear redriver families with AUX-SBU switching variants)
- Texas Instruments TUSB1002 (dual-channel USB3.1 10Gbps linear redriver; used when a separate switch is already selected)
- Texas Instruments TUSB1104 (USB-C 10Gbps USB3.2 x2 adaptive linear redriver)
- Texas Instruments TS3USBCA4 (USB-C SBU multiplexer for AUX / analog audio / debug reuse)
- Texas Instruments TMUXHS4446 (also provides low-speed SBU switching capability for DP AUX paths)
- Parade Technologies PS8828A (USB 3.2 Gen2x2 + DP Alt-Mode retimer class)
- Parade Technologies PS8830 / PS8833 (USB4 retimer family with DP Alt-Mode support)
- Diodes Incorporated PI2DPT1021Q (USB-C DP Alt bi-directional retimer with adaptive equalizer)
- Intel JHL9040R (Thunderbolt 4 / USB4 retimer class; used in host platforms)
Selection flow: requirements → device class (orientation/MUX layer)
Follow the four requirement nodes from left to right. Each path ends in a device class that matches the routing scope of this page.
Recommended topics you might also need
Request a Quote
H2-12. FAQs (Field Troubleshooting, Within Scope)
Each answer is fixed to four lines: Likely cause / Quick check / Fix / Pass criteria. Thresholds use placeholders X/Y.
Works one orientation, fails flipped — what is the first mapping sanity check?
Likely cause: Orientation map (lane swap/polarity) is wrong, or MUX SEL polarity is inverted for A/B insertion.
Quick check: Read/log orientation_state and mux_sel; force MUX=A then MUX=B with the same cable and compare outcomes.
Fix: Correct the lane mapping table and GPIO polarity; keep MUX selection stable from attach through training completion.
Pass criteria: A/B insertions (Y cycles each) succeed ≥ (100−X)%; training retries ≤ X per 100 cycles.
USB SuperSpeed is OK, but DP Alt-Mode is dead — is AUX/HPD routed via the wrong path?
Likely cause: DP lane muxing is correct, but AUX/HPD/SBU routing is missing, mis-switched, clamped, or leaking.
Quick check: Verify HPD toggles; run continuous AUX transactions (e.g., repeated reads) and count errors; force/bypass the SBU path if possible.
Fix: Ensure AUX/HPD uses the intended switch path in DP mode; enforce low-leakage/low-C sideband routing symmetry; avoid unintended clamps.
Pass criteria: AUX error ≤ X per 10k ops; HPD has no extra pulses over Y minutes; DP remains stable for Y minutes.
DP flickers only after hot-plug — HPD glitch or a MUX default-state window?
Likely cause: MUX default/connect window briefly creates a wrong mapping, or HPD glitches during switching.
Quick check: Timestamp HPD events vs mux_en/mux_sel edges; compare auto mode vs forced fixed MUX state.
Fix: Make switching glitch-free; add guard time X ms; keep MUX Hi-Z/disabled until mapping settles; assert HPD only after stable mapping.
Pass criteria: Hot-plug Y times with zero flicker; HPD pulse count = 1±0; no retrain within Y minutes.
Link trains, then flaps — is there a MUX control race during the training window?
Likely cause: MUX selection changes during training, or orientation state re-evaluates mid-window due to noisy CC decisions.
Quick check: Log mux_sel transitions and correlate with the training phase; force MUX static and see if flaps stop.
Fix: Freeze MUX state from attach through training completion; add CC debounce X ms; enforce single-writer state machine rules.
Pass criteria: MUX transitions during training = 0; flaps ≤ X per hour; stable throughput for Y minutes.
Only long cables fail — is the MUX insertion-loss budget exceeded or crosstalk amplified?
Likely cause: Passive MUX adds insertion/return loss; cable + connector discontinuities push the channel beyond margin; local crosstalk near MUX worsens.
Quick check: Short vs long cable A/B; reduce rate/mode as a margin probe; compare errors to a test point before/after MUX if available.
Fix: Move to an active redriving switch or retimer class; adjust placement (near receptacle vs SoC) by budget; tighten spacing/return paths around the MUX.
Pass criteria: At target mode, errors ≤ X per hour over Y hours; margin ≥ X dB / X mV / X UI (placeholder).
AUX reads are unstable — SBU switch leakage/C too high, or asymmetry?
Likely cause: SBU path capacitance/leakage or asymmetry degrades AUX; sideband return path is noisy or discontinuous.
Quick check: Force/bypass the SBU switch path if supported; compare AUX error rate vs temperature and vs orientation A/B.
Fix: Use a lower-C, lower-leakage sideband switch; keep SBU routing symmetric; isolate SBU from noisy nets and provide a clean reference return.
Pass criteria: AUX error ≤ X per 10k ops across temp range; A vs B delta error ≤ X%; AUX margin ≥ X% (placeholder).
EMI worsens after adding a MUX — is the return path broken near the MUX?
Likely cause: Reference plane split or return-path detour near the MUX; insufficient ground stitching and containment around high-speed switching.
Quick check: Inspect plane continuity under the pairs; verify via fencing and stitch density; correlate the EMI peak to the MUX region (near-field scan if available).
Fix: Restore continuous reference under pairs; add stitching vias; reduce stubs; tighten containment around the MUX and connector transitions.
Pass criteria: EMI peak reduced ≥ X dBµV (failing band); link errors remain ≤ X per hour in both orientations.
Random failures on sleep/resume — fail-safe state vs power sequencing mismatch?
Likely cause: Powered-down MUX state creates a partial/ambiguous connection, or MUX enables before rails and orientation state are stable.
Quick check: Capture timestamps for power_good, mux_en, mux_sel, cc_attach; run sleep/resume loop Y cycles.
Fix: Define explicit fail-safe (Hi-Z vs default connect) consistent with system power strategy; delay MUX enable by X ms after power_good; re-latch orientation before enabling.
Pass criteria: Y cycles with failures ≤ X; link up within X s after resume; no extra attach/detach events logged.
DP works but USB SuperSpeed drops — lane muxing mode conflict or wrong profile strap?
Likely cause: Crossbar mode/profile selects a DP mapping that disables or misroutes USB SS lanes; mode transitions overwrite orientation mapping.
Quick check: Read mode/profile strap/register state; force USB-only mapping and verify SS; verify DP entry does not change USB mapping unexpectedly.
Fix: Correct profile selection logic; serialize mode transitions; apply orientation mapping after profile; keep distinct profiles for USB-only vs USB+DP.
Pass criteria: Mode switching Y times: USB SS errors ≤ X per hour; DP remains stable; mapping state matches expected table in 100% of transitions.
One board revision fails while the previous passes — did length-match break around the MUX?
Likely cause: Pair skew increased, via/stub count changed, or a reference plane discontinuity was introduced in the MUX region.
Quick check: Compare deltas around MUX (pair lengths, skew, via transitions, spacing, plane splits); re-run A/B orientation bring-up on both revisions.
Fix: Restore length-match to within X (mm/mil); remove stubs where possible; route over continuous reference; re-place MUX to reduce discontinuities.
Pass criteria: Bring-up success ≥ (100−X)% across Y units; A/B training time differs ≤ X% (or ≤ X ms).
Works in the lab, fails during ESD events — protector placement or ground path issue? (scope-only)
Likely cause: Discharge return path injects noise into the MUX region; protector is too far from the connector; ground/chassis bonding is incomplete.
Quick check: Review protector-to-connector distance and the low-inductance return path; compare failure rate pre/post ESD; check orientation/DP-only correlation.
Fix: Place protection at the port with a short, controlled return; keep MUX reference/return clean. Detailed TVS part numbers and IEC workflows belong to the “USB Port ESD/TVS” sibling page.
Pass criteria: Post-ESD recovery time ≤ X s; subsequent Y hot-plugs succeed 100%; errors do not exceed X per hour.
Field reports “only some docks fail” — routing constraints or orientation detection threshold?
Likely cause: Orientation/Alt-Mode entry thresholds are marginal; certain docks/cables expose edge cases in mapping or sideband constraints.
Quick check: Build a dock×cable matrix; log CC/orientation decision + alt-mode entry reason; force orientation/MUX states and compare outcomes.
Fix: Tighten debounce/windowing for orientation decisions; ensure mapping tables cover all supported DP lane combos; keep conservative default states during attach/detach.
Pass criteria: Top-N docks pass rate ≥ (100−X)% across Y sessions; no orientation-dependent failures; time-to-enumerate ≤ X s.