LCD Layout Redesign v2 — Full Reboot¶

Status: Design spike — NO code changes. Author: Design agent, 2026-04-19. Supersedes intent of: docs/plans/2026-04-19-lcd-layout-redesign.md and docs/plans/2026-04-19-lcd-layout-fixup.md (incremental patches on top of the #844 scaffold). Target file (when implementation starts): frontend/src/components-v2/panels/lcd/AmberLcdDisplay.svelte and siblings. Out of scope of this document: no .svelte, .ts, .css edits. This is a written design spike only.

1. User feedback + screenshot analysis¶

1.1 Verbatim user complaint (2026-04-19)¶

"LCD — плохо. layout самого индикатора — говно. Все еще перекрываются значения, все еще полно свободного места снизу. Полагаю, имеет смысл ПОЛНОСТЬЮ изменить layout самого LCD экрана. Возможно сделать несколько вариантов и выбрать лучший. Для этого нужно отдельный агент, с нужным дизайн скиллом, может дать ему картинку, и объяснить, что как.. Может из интернета найти индикаторы ЖКИ Elektraft, older ICOMs, etc..."

Translated: LCD is bad. Indicator layout is garbage. Values still overlap, bottom half is still empty. Redesign fully. Produce multiple variants.

1.2 Concrete defects visible in the 2026-04-19 screenshot¶

#	Defect	Root cause (code-level)
D1	VFO A digits `7.143.0` visually clip into `LSB 1` / `40m` badges	DSEG7 digit advance-width is fixed; even with the 3-column subgrid (#860), when column 2 shrinks below the digit string's natural width `overflow:hidden` crops the LSD rather than reflowing. Result: digits appear to hit the badges at narrow widths.
D2	VFO B `7.188.000` is permanently demoted (50% smaller, dim, 70% vertical scale meter)	`.lcd-meter-sub { transform: scaleY(0.7); opacity: 0.6 }` + `AmberFrequency size="small"`. #845 ("VFO-B peer promotion") was parked; nothing in the current grid reserves peer-sized real estate.
D3	Status row clogged: 20+ tokens wrap onto 2 lines on dual-RX	Single flex-wrap row with no priority/zoning. Every capability-gated chip sits in the same bucket.
D4	Filter trapezoid sits in middle; a ~40–60 px amber strip is empty below it on tall viewports	`align-content: start` on the grid + `minmax(0,120px)` cap on `filter` track. Extra vertical space is unused.
D5	Contrast picker (#877) lives under the display in a separate control-strip — fine for desktop, but on mobile the init was regressed (#877 P2 comment). Not a layout problem of the LCD itself, but affects vertical budget.
D6	No visible memory / recent-QSY slot (#862) or telemetry strip (#837) — despite the empty bottom half.

1.3 Diagnosis¶

The existing grid (#844 scaffold + #860 freq-subgrid + #871 fixup) is incrementally patched around an outgrown mental model: "VFO A is the hero, VFO B is a footnote, indicators are one long strip, everything else is a full-width row." This worked for a single-RX IC-7300-style layout but has failed to scale for IC-7610 (dual-RX first-class) and for radios with rich status token sets (FTX-1, IC-7610 PROC/DIGI-SEL/IP+/ATU).

A full redesign needs to abandon three assumptions:

"Indicator row is global." — It should be zoned per VFO, with only truly global flags (TX, SPLIT, LOCK) at the top.
"VFO B is secondary." — On dual-RX radios (IC-7610, TS-990S, FTDX101) both receivers are peers. The active receiver is signaled by ink-alpha / border-weight, never by font size.
"Filter viz is one full-width row." — It can be per-VFO in a dual-cockpit, or consolidated below meters as a dedicated widget zone.

2. Reference research¶

2.1 Elecraft K3 / K3S / KX3¶

Two-tier monochrome VFD: upper 8-digit 7-segment (VFO A), lower 13-segment dot-matrix (VFO B / text / menu).
VFO A is always the "active TX" anchor. VFO B row doubles as scrolling text for menu labels.
Right-side icon cluster (RIT/XIT/SPLIT/ATU/PRE/ATT/NB/NR) sits in a narrow column, not a top strip.
S-meter is horizontal bar above the digits.
Takeaway: "icon column on the right, meter on top, digits in the middle" gives digits the widest stable real estate and keeps the icon row from competing with digits for width.
Source: Elecraft K3 Owner's Manual, K3S Programmer's Reference.

2.2 Icom IC-756PRO / IC-756PROIII / IC-746PRO¶

TFT color (not true amber LCD — the "amber" is our skin convention), but the PRO layout is the visual vocabulary the user expects.
Upper half: spectrum scope (±12.5/25/50/100 kHz); lower half: VFO digits, dual frequency readout, mode, filter width, multi-meter.
Status icons are grouped into top-left (RX path: PRE/ATT/AGC), top-right (mode+filter), bottom (notch/NR/NB/TBW). Zoning, not a single strip.
Dual-watch/sub-receiver shows VFO B as a smaller line below VFO A but same digit style — demotion is by size, not by crop.
Source: IC-756PRO Instruction Manual, IC-746PRO LCD image.

2.3 Icom IC-7300¶

Single 4.3" TFT. Top 60% = spectrum + waterfall; bottom 40% = digits, mode, filter, S-meter bar, indicators zoned along outer edges.
Indicators clustered into "RX front-end" (top-left) and "DSP" (bottom-right); mode/filter always right of digits.
No dual-RX — single VFO is the only hero.
Takeaway: 60/40 vertical split works when a spectrum widget exists. Our analog is the AF scope.

2.4 Kenwood TS-990S¶

Dual-display hardware: 7" main + 3.5" sub. The sub is dedicated to VFO B.
Main shows band scope, waterfall, dual frequency readouts, S-meter, RIT/XIT.
Sub shows VFO B freq + AF spectrum + filter shape — effectively the widget our AmberAfScope is modeled on.
Takeaway: the AS-990S treats dual-RX as two cockpits side-by-side, each with its own digits + meter + scope. This is the reference for "Variant B — Dual Cockpit" below.
Source: TS-990S Dual Display, TS-990S brochure.

2.5 Yaesu FTDX101D / MP¶

7" TFT with user-selectable layout: MAIN only, MAIN/SUB stacked, MAIN/SUB side-by-side (two sub-modes).
Dual independent scopes (one per receiver), 3DSS + waterfall.
Takeaway: explicit layout presets let the operator choose cockpit shape. If we expose a small set of presets, we don't have to pick a single winner.
Source: FTDX101D Operation Manual.

2.6 Vintage bonus — Icom IC-781 / Yaesu FT-1000MP¶

Large segmented VFDs with generous whitespace around digits. Indicators were ghost-etched into the bezel, lighting up only when active — hence the "token-cascade alpha ghost/inactive/active" metaphor we already use.
Takeaway: our --lcd-alpha-ghost idea is correct; the variants just need to deploy it more aggressively so the bottom half isn't blank, it's ghost-etched labels waiting to light up.

3. Three variants¶

Terminology: - Cell = CSS grid area. - Zone = a named region that may contain multiple cells. - ASCII wireframes use 12 columns × rows, · = empty (ghost-etched), █ = active ink, ▒ = inactive ink.

3.1 Variant A — "Elecraft Column" (icons on the right rail)¶

Premise: push all capability-gated tokens into a narrow right-rail column; digits + scope get a wide left column. Solves D1 (digit collision — digits never compete with badges for horizontal room) and D3 (status row is vertical, scrolling-tolerant).

Single-RX wireframe¶

┌────────────────────────────────────────────────┬──────────┐
│ [TX]  [SPLIT]  [LOCK]            CLOCK 14:32  │ MODE LSB │ ← row 1 global
├────────────────────────────────────────────────┼──────────┤
│                                                │ FILT 2.4 │
│   A   7 . 1 4 3 . 0 0 0   Hz                   │ BAND 40m │
│                                                │          │
│                                                │ PRE IPO  │
│                                                │ ATT ·    │
├────────────────────────────────────────────────┤ ATU ·    │
│  S ▓▓▓▓▓▓░░░░░   +20dB                  [S]   │ NB  ·    │
├────────────────────────────────────────────────┤ NR  ·    │
│                                                │ NOTCH·   │
│  ╱‾‾‾‾‾‾‾‾╲   AF SCOPE + FILTER SHAPE          │ ANF ·    │
│ ╱          ╲  IF-shift · PBT · contour · notch │ AGC MID  │
│                                                │ RFG 255  │
│                                                │ SQL ·    │
│                                                │ RIT ·    │
│                                                │ VOX ·    │
└────────────────────────────────────────────────┴──────────┘

Dual-RX wireframe¶

┌────────────────────────────────────────────────┬──────────┐
│ [TX]  [SPLIT A→B]  [LOCK]        CLOCK 14:32  │ MODE LSB │
├────────────────────────────────────────────────┼──────────┤
│  ►A   7 . 1 4 3 . 0 0 0   Hz   40m  LSB 1     │ FILT 2.4 │
│   B   7 . 1 8 8 . 0 0 0   Hz   40m  LSB       │ BAND 40m │
├────────────────────────────────────────────────┤          │
│  Sa ▓▓▓▓▓▓░░░░  S9+20                  [S]   │ PRE / ATT│
│  Sb ▓▓░░░░░░░░  S3                            │ NB / NR  │
├────────────────────────────────────────────────┤ NOTCH/ANF│
│                                                │ AGC MID  │
│  AF SCOPE (active RX) + FILTER SHAPE            │ RFG/SQL  │
│                                                │ RIT/VOX  │
└────────────────────────────────────────────────┴──────────┘

CSS grid pseudocode¶

.lcd-screen {
  display: grid;
  grid-template-columns: minmax(0, 1fr) 140px;
  grid-template-rows:
    28px                /* global-ind */
    auto                /* vfo-stack (A, B stacked; B collapses on single-RX) */
    minmax(48px, 72px)  /* meter-stack */
    minmax(0, 1fr);     /* scope (fills remaining height — kills D4 empty strip) */
  grid-template-areas:
    "global-ind  right-rail"
    "vfo-stack   right-rail"
    "meter-stack right-rail"
    "scope       right-rail";
}

Rationale / trade-offs: - ✅ Solves D1 completely: digits own column 1, zero badges in their row. - ✅ Solves D4: scope cell is 1fr — absorbs all extra height. - ✅ Solves D3: right rail wraps indicators vertically, each gets a fixed slot. - ⚠️ Visually unfamiliar to Icom users — "Icom operators expect mode/filter next to digits, not in a rail." - ⚠️ Right rail takes 140 px that could go to the scope. - ⚠️ VFO B still stacked under VFO A (peer-by-alpha, not peer-by-position). Only half-fixes D2.

Files affected: AmberLcdDisplay.svelte (grid rewrite), new AmberRightRail.svelte (extracts all status chips), AmberVfoLine.svelte (shared single-line freq+band+mode component used by both A and B). Delete the flex .lcd-ind-row.

3.2 Variant B — "Dual Cockpit" (two-column, TS-990S-style)¶

Premise: on dual-RX, split the screen into two equal columns, each a complete cockpit: freq, meter, mode/filter/band badges, per-VFO indicators (NB, NR, NOTCH, ANF). Global indicators (TX, SPLIT, LOCK, ATU, VOX, PROC) go top. Scope spans both columns at bottom. On single-RX, column B collapses and column A stretches full-width.

Single-RX wireframe¶

┌────────────────────────────────────────────────────────────┐
│ [TX] [VOX] [PROC 5] [ATU] [SPLIT] [LOCK]        14:32      │ ← global
├────────────────────────────────────────────────────────────┤
│  ►A   7 . 1 4 3 . 0 0 0   Hz                               │
│                                         40m   LSB 2.4     │
├────────────────────────────────────────────────────────────┤
│  S ▓▓▓▓▓▓░░░░░░  S9+20                         [S|PO|SWR] │
├────────────────────────────────────────────────────────────┤
│  PRE AMP1   ATT ·   NB 3   NR 5   NOTCH ·   ANF ·          │ ← per-RX row
│  AGC MID    RFG 255  SQL ·   RIT +120 Hz                   │
├────────────────────────────────────────────────────────────┤
│                                                            │
│   AF SCOPE  + FILTER SHAPE + PBT + NOTCH + CONTOUR         │
│                                                            │
└────────────────────────────────────────────────────────────┘

Dual-RX wireframe¶

┌────────────────────────────────────────────────────────────┐
│ [TX] [VOX] [PROC 5] [ATU] [SPLIT A→B] [LOCK]    14:32      │
├──────────────────────────────┬─────────────────────────────┤
│  ►A  7 . 1 4 3 . 0 0 0       │   B  7 . 1 8 8 . 0 0 0      │
│      40m  LSB 2.4            │      40m  LSB 2.4           │
├──────────────────────────────┼─────────────────────────────┤
│  Sa ▓▓▓▓▓░░░░  +20     [S]  │   Sb ▓▓░░░░░  S3      [S]   │
├──────────────────────────────┼─────────────────────────────┤
│ PRE AMP1  ATT ·  NB 3        │  PRE ·  ATT 6  NB ·         │
│ NR 5  NOTCH·  ANF·  AGC MID  │  NR ·  NOTCH·  ANF·  AGC F  │
├──────────────────────────────┴─────────────────────────────┤
│                                                            │
│        AF SCOPE (active RX) — spans full width             │
│                                                            │
└────────────────────────────────────────────────────────────┘

CSS grid pseudocode¶

.lcd-screen {
  display: grid;
  grid-template-columns: minmax(0, 1fr);          /* single-RX */
  grid-template-rows:
    28px        /* global */
    auto        /* vfo */
    44px        /* meter */
    auto        /* per-rx indicators */
    minmax(0, 1fr); /* scope */
  grid-template-areas:
    "global"
    "vfo-a"
    "meter-a"
    "ind-a"
    "scope";
}
.lcd-screen.dual {
  grid-template-columns: minmax(0, 1fr) minmax(0, 1fr);
  grid-template-areas:
    "global   global"
    "vfo-a    vfo-b"
    "meter-a  meter-b"
    "ind-a    ind-b"
    "scope    scope";
}

Rationale / trade-offs: - ✅ Solves D2 cleanly: VFO B is a true peer. Active receiver indicated by border-weight and the ► glyph plus --lcd-alpha-active on the header; inactive by --lcd-alpha-inactive. Font size is identical. - ✅ Solves D3: per-VFO chips go into their own column; global chips get their own small row — no more 20-item flex-wrap. - ✅ Solves D4: scope gets 1fr and spans full width. - ✅ Solves D1: each column is minmax(0,1fr), digits have their own cell, mode/band go below digits (not beside), zero overlap risk. - ✅ Solves D5/D6 indirectly: extracting per-RX chips makes room for memory/telemetry above the scope (new optional aux row). - ⚠️ On narrow viewports (<720 px) two cockpits collide. Needs a container query to collapse to stacked-single at < ~640 px. - ⚠️ Per-RX indicators duplicate some chips that are actually radio-global on IC-7610 (e.g. ATU is global, NB is per-RX). Need an indicator taxonomy (see §7 Q1). - ⚠️ Scope is still single (active RX only). Dual-scope is a later enhancement.

Files affected: AmberLcdDisplay.svelte (grid), new AmberCockpit.svelte (reusable freq+meter+ind block, takes which: 'A'|'B'), new AmberGlobalInd.svelte (top strip), new AmberPerVfoInd.svelte. Retire .lcd-vfo-sub, .lcd-meter-sub. AmberAfScope unchanged.

3.3 Variant C — "Modern Icom" (scope dominant, 60/40 split)¶

Premise: IC-7300 / FTDX101 style. Top 55–60% of the LCD is a big AF scope + filter+PBT+notch visualization (the skin's whole reason for existing on radios without an RF scope). Bottom 40–45% is a data strip: VFO A+B digits, meters, zoned indicators. Mirror the IC-7300's visual hierarchy.

Single-RX wireframe¶

┌────────────────────────────────────────────────────────────┐
│                                                            │
│                                                            │
│          AF SCOPE + WATERFALL TRACE + FILTER SHAPE         │
│          (IF shift marker · notch marker · PBT)             │
│                                                            │
│                                                            │
├────────────────────────────────────────────────────────────┤
│ [TX] [VOX] [PROC] [ATU] [SPLIT] [LOCK]   [PRE][ATT][AGC]  │ ← indicator strip
├────────────────────────────────────────────────────────────┤
│ ►A  7 . 1 4 3 . 0 0 0                   40m  LSB  2.4 kHz │
├────────────────────────────────────────────────────────────┤
│  S ▓▓▓▓▓▓░░░░░░  +20         [S|PO|SWR|ALC|COMP]          │
├────────────────────────────────────────────────────────────┤
│ NB3  NR5  NOTCH·  ANF·  CONT·  RFG255  SQL·  RIT+120      │ ← DSP row
└────────────────────────────────────────────────────────────┘

Dual-RX wireframe¶

┌────────────────────────────────────────────────────────────┐
│                                                            │
│          AF SCOPE (active RX, large)                       │
│          — caller can toggle split-scope later             │
│                                                            │
├────────────────────────────────────────────────────────────┤
│ [TX] [VOX] [PROC] [ATU] [SPLIT A→B] [LOCK]  [PRE][ATT][AGC]│
├──────────────────────────────┬─────────────────────────────┤
│ ►A  7 . 1 4 3 . 0 0 0        │  B  7 . 1 8 8 . 0 0 0       │
│     40m  LSB  2.4            │     40m  LSB  2.4           │
├──────────────────────────────┼─────────────────────────────┤
│ Sa ▓▓▓▓▓░░░░  +20      [S]  │ Sb ▓▓░░░░░░  S3        [S]  │
├──────────────────────────────┴─────────────────────────────┤
│ NB3  NR5  NOTCH·  ANF·  CONT·  RFG255  SQL·  RIT+120      │
└────────────────────────────────────────────────────────────┘

CSS grid pseudocode¶

.lcd-screen {
  display: grid;
  grid-template-rows:
    minmax(0, 1.2fr)   /* scope dominates */
    28px                /* global + front-end indicator strip */
    52px                /* vfo line(s) */
    44px                /* meter */
    28px;               /* dsp strip */
  grid-template-columns: minmax(0, 1fr);
  grid-template-areas:
    "scope"
    "ind-global"
    "vfo"
    "meter"
    "ind-dsp";
}
.lcd-screen.dual {
  /* vfo + meter rows become 2-col */
  grid-template-columns: minmax(0, 1fr) minmax(0, 1fr);
  grid-template-areas:
    "scope      scope"
    "ind-global ind-global"
    "vfo-a      vfo-b"
    "meter-a    meter-b"
    "ind-dsp    ind-dsp";
}

Rationale / trade-offs: - ✅ Solves D4 hardest: scope is the largest cell; bottom-empty-space problem inverts (bottom is now packed). - ✅ Solves D1: VFO row is its own row with mode+band pushed to the right edge via justify-content: space-between and the digits in the flex-grow child — same pattern as #860 but with more horizontal breathing room because there's nothing stacked vertically alongside. - ✅ Solves D3: two indicator strips (global + DSP) with clear semantic zones. - ⚠️ Only half-fixes D2: VFO B is a peer only on dual-RX (good), but on single-RX VFO B row simply doesn't exist, so no regression. - ⚠️ On short viewports (<480 px tall, e.g. mobile landscape), the 1.2fr scope steals digit readability. Needs a scope-collapse breakpoint. - ⚠️ The user complaint was "bottom half empty" — this variant puts everything at the bottom. If the scope is not available (no hasAudioFft()), the scope cell becomes a big empty amber field. Needs a scope fallback (ghost-etched graticule + "AUDIO FFT UNAVAILABLE" label in ghost alpha).

Files affected: AmberLcdDisplay.svelte (grid rewrite), AmberAfScope.svelte (compact={false} default, add fallback slot), new AmberIndicatorStrip.svelte with zone: 'global' | 'dsp' | 'frontend' prop. AmberFrequency unchanged.

4. Recommendation¶

Recommend Variant B ("Dual Cockpit") with a Variant-C scope fallback posture.

Why B wins¶

Dual-RX is first-class for the project's reference radio (IC-7610) and for TS-990S / FTDX101 class targets. Variant A and C both leave VFO B in a degraded slot; only B treats it as a peer, directly resolving the user's D2 complaint.
Indicator zoning per-VFO is the most honest mapping of the state model: main.nbActive, sub.nbActive, main.agcMode, sub.agcMode are already separate fields in radioState. The current single strip collapses that into one, losing information; Variant B restores it visually.
Capability cascade stays clean. Each cockpit renders only chips supported by its receiver's capabilities. A single-RX radio shows exactly one cockpit — full width, no wasted columns, D4 solved by scope: 1fr. A dual-RX radio shows two — no reflow of peer cells (same win the #844 plan §9.1 P2 was after, but achieved by peer columns instead of fixed row heights).
Svelte 5 + token cascade friendly. AmberCockpit is a single reusable component that takes which: 'A' | 'B' and active: boolean; the token --lcd-alpha-{active,inactive,ghost} already drives every ink decision — B just deploys it at cockpit granularity instead of element granularity.
ESLint no-restricted-imports unaffected. AmberCockpit lives in panels/lcd/, receives all state via props (from the existing toXxxProps adapters), and imports nothing from $lib/transport or $lib/audio/audio-manager. Layering preserved.
Variant-C ideas survive as knobs, not a fork. We can give the scope 1fr on the row-track and let it absorb bottom space (picks up C's biggest win), without committing to a scope-dominated layout that breaks when FFT is unavailable.

Why not A¶

The right-rail idea is architecturally clean, but it (a) costs 140 px of horizontal budget on every viewport, including narrow mobile, and (b) looks un-Icom. The skin's purpose is "radio-console-feeling fallback"; the rail makes it feel like a debug HUD.

Why not C (as headline)¶

C solves the empty-bottom complaint by inverting it, but it penalizes radios without hasAudioFft() (a scope-shaped empty). It's also a single-hero layout — VFO B is still compact. Borrow its "scope = 1fr" row idea into B; don't adopt C wholesale.

5. Atomic implementation plan¶

Each issue is a single atomic PR, ≤3 files, ≤200 LOC. Sequence is strict; each depends on the previous one landing first.

PR1 — `refactor(#NEW1): extract AmberCockpit component (behavior-preserving)`¶

New file: frontend/src/components-v2/panels/lcd/AmberCockpit.svelte
Accepts props: which: 'A' | 'B', active: boolean, freqHz, mode, filter, band, sValue, meterSource, txActive, indicators: {nb, nr, notch, anf, pre, att, agcLabel, rfgActive, sqlActive, ritOffset|null}.
Renders existing VFO-A visuals at first; the which === 'B' && !active branch matches today's compact sub.
AmberLcdDisplay.svelte swaps its VFO A + VFO B blocks to <AmberCockpit> twice but uses the same old grid-template-areas.
Gate: zero visual regression on single-RX; dual-RX looks identical to today.
Files: 2 (new component + display). LOC: ~180.

PR2 — `feat(#NEW2): dual-cockpit grid — VFO B as peer (replaces scope of parked #845)`¶

AmberLcdDisplay.svelte: replace the single-column / dual-column grid-templates with the Variant B grid from §3.2. VFO B cockpit now uses active={activeVfo === 'B'} and renders at full size with alpha-dimmed tokens when inactive.
Delete .lcd-vfo-sub, .lcd-meter-sub CSS (the scaleY(0.7) + opacity: 0.6 demotion).
Adjust AmberCockpit to render identical sizing regardless of active; only ink alpha changes.
Gate: single-RX unchanged (cockpit = col-span-all); dual-RX both VFOs same size; screenshot in both active === 'A' and active === 'SUB' states.
Files: 2. LOC: ~150 net (mostly deletion).
Makes parked #845 obsolete. Close #845 with a pointer.

PR3 — `refactor(#NEW3): per-VFO indicator zone + global indicator strip`¶

New file: AmberIndStrip.svelte with prop zone: 'global' | 'perVfo' and tokens: IndToken[].
AmberCockpit uses <AmberIndStrip zone="perVfo"> in its own row.
AmberLcdDisplay uses <AmberIndStrip zone="global"> at the top row (TX/VOX/PROC/ATU/SPLIT/LOCK only).
Indicator taxonomy table lives at the top of AmberIndStrip.svelte as a typed const (answer to §7 Q1).
Gate: token total count unchanged; visual wrap-lines reduced from 2 → 1 per zone at 1280 px viewport.
Files: 3. LOC: ~180.
Makes the legacy .lcd-ind-row flex-wrap obsolete. Delete it in this PR.

PR4 — `feat(#NEW4): scope track fills remaining height (absorbs D4 empty amber)`¶

AmberLcdDisplay.svelte grid grid-template-rows last track minmax(0, 1fr) for the scope area (currently minmax(0, 120px)).
AmberAfScope.svelte: add compact={false} default path that draws graticule + filter trapezoid scaled to full cell.
Ghost fallback: when !hasAudioFft(), render AmberFilterGhost.svelte (graticule + filter shape only, no live data, label "AUDIO FFT UNAVAILABLE" at --lcd-alpha-ghost).
Gate: on tall viewports (>800 px) no amber strip below the scope; on short viewports (<500 px) the scope still gets ≥80 px.
Files: 3. LOC: ~160.
Addresses #862's memory-slot premise too: with the scope now 1fr, a future memory row can be inserted above the scope without shoving content down.

PR5 — `feat(#NEW5): mobile container-query collapse + telemetry reserve`¶

Container query on .lcd-screen: @container (max-width: 640px) flips dual-cockpit grid back to stacked single-cockpit with VFO B as a compact one-liner (explicit, not scaleY(0.7)).
Reserve a slim aux row (height 0 by default, auto when hasCapability('telemetry'), addressing #837) between ind-dsp/meter and scope.
Gate: mobile viewport (375 × 667) no horizontal scroll; dual-RX degrades to single-column stacked gracefully.
Files: 1–2. LOC: ~100.
Partially resolves #862 / #837 — they become trivial follow-ups (fill the reserved aux row).

Obsolete vs survivable queue items¶

Issue	Status after this plan	Reason
#845 (VFO-B peer promotion)	Obsolete — close	PR2 delivers the promotion by grid redesign.
#846 (specific indicator reorg, if same)	Likely obsolete	Subsumed by PR3 zoning. Confirm with user in Q3.
#862 (memory / recent-QSY slot)	Survivable	PR5 reserves the `aux` row; #862 fills it.
#837 (telemetry VD/TEMP/ID)	Survivable	Same `aux` row mechanism.
#835 (filter-viz wide mode)	Obsolete	Scope is always wide now. Close.
#877 (contrast in control-strip)	Unchanged	Lives below LCD in a separate strip; this redesign is inside the LCD only.

6. Migration from current state¶

Current state, in file terms: - #844 scaffolded the grid (grid-template-areas: indicators / meter-a / vfo-a / pb / filter). - #860 added a 3-col subgrid inside .lcd-vfo-main to stop digit/badge collision (partial fix; D1 remnants visible). - #871 capped the filter track at minmax(0, 120px) + align-content: start (creates D4 empty amber). - #877 moved contrast control outside the LCD into the sidebar control-strip.

Migration steps¶

PR1 — Cockpit extraction. Keep the current grid. Just reorganize VFO A + B into a shared component. No visual change. Safe, easy to revert.
PR2 — Grid swap. The indicators / meter-a / vfo-a / pb / filter template is replaced by Variant B's global / vfo-? / meter-? / ind-? / scope. The pb row is absorbed into each cockpit's own RIT offset display (cockpits own their RIT because RIT is per-receiver on IC-7610).
PR3 — Indicator split. The legacy .lcd-ind-row flex strip becomes AmberIndStrip zone="global" (6–8 chips max) + two per-cockpit AmberIndStrip zone="perVfo" (8–12 chips each, depending on capabilities).
PR4 — Scope grows. The minmax(0, 120px) filter track becomes minmax(0, 1fr) scope track. The align-content: start stays, but with a 1fr track there's no residual space to align. D4 resolved.
PR5 — Responsive + reserves. Adds container query and aux row. No change to desktop visuals unless telemetry/memory is enabled.

Rollback plan¶

Each PR is independently revertable. PR1–5 all live in AmberLcdDisplay.svelte + sibling components under panels/lcd/; no upstream wiring (state-adapter, runtime, stores) changes, so reverting is pure UI.

Risks¶

Snapshot tests on VFO rendering (if any exist under frontend/tests/) will break at PR1 if they assert on DOM structure. Audit first.
#877 mobile contrast regression is orthogonal — do not try to fix it inside this plan.
Capability flag coverage — per-VFO indicator zoning assumes radioState.main.nbActive and radioState.sub.nbActive both exist. For radios where NB is global (not per-RX), the same value would render in both cockpits — potentially confusing. Mitigate with the indicator taxonomy const (PR3).

7. Open questions¶

Indicator taxonomy — global vs per-VFO. For each of TX, VOX, PROC, ATT, PRE, DIGI-SEL, IP+, ATU, NB, NR, NOTCH, ANF, AGC, RFG, SQL, RIT, SPLIT, LOCK, DATA, CONT — is it radio-global (one value) or per-receiver (two values)? Proposal: global = {TX, VOX, PROC, ATU, SPLIT, LOCK}; per-RX = everything else. Confirm for IC-7610.
VFO B SPLIT direction. When splitActive, do we show SPLIT A→B (direction of TX) as a global chip, or mark VFO B's cockpit with a "TX" tag? Proposal: global chip SPLIT, plus a small TX marker on whichever cockpit is the TX VFO.
Is #846 the same as our PR3 indicator zoning? If yes, close #846 as duplicate. If no, describe what #846 wanted that PR3 misses.
Scope fallback text. When !hasAudioFft(), what do we show? Options: (a) ghost graticule + "AUDIO FFT UNAVAILABLE"; (b) ghost graticule + filter-shape trapezoid only, no label; (c) collapse the scope row to 0, let indicators breathe vertically. Proposal: (b) — filter shape is still useful without FFT data.
Container query breakpoint. 640 px was picked by eyeball. Is there an existing breakpoint token in the codebase (skins/ / lib/ui/tokens) we should reuse?
RIT ownership. Current code treats RIT as a single global state (ritXit.ritActive). On IC-7610 RIT is per-receiver (at least for main vs sub dial offset). Does radioState.main.rit / radioState.sub.rit exist in the state shape, or is RIT still modeled as global? If global, we keep RIT in the global strip; if per-RX, each cockpit owns it. This changes PR2/PR3 scope.
Meter B source. Today VFO B meter is hard-coded to S (S-meter). Should the B cockpit gain its own PO/SWR/ALC cycle button during split-TX? Proposal: no — during TX, the TX-VFO cockpit's meter switches to PO/SWR/ALC; the other cockpit shows S or blanks.
Filter readout placement. Variant B wireframe puts filter width as LSB 2.4 inside the badge row. The existing toFilterProps adapter has more (IF-shift, contour). Does the user want a textual filter readout in the cockpit (e.g. "SHIFT -120 Hz"), or is the scope's visual notch/shift marker enough? Proposal: scope-only.
compact prop on AmberAfScope. Currently forced compact. After PR4, compact={false} by default. Do we keep compact as a prop for any other caller? Grep shows only AmberLcdDisplay uses it — safe to drop.
Skin-wide audit. This plan changes amber-lcd only. desktop-v2 and mobile skins reuse AmberCockpit / AmberIndStrip? Proposal: yes, but that's a separate spike — flag for a follow-up plan if the user wants consistency.

LCD Layout Redesign v2 — Full Reboot¶

1. User feedback + screenshot analysis¶

1.1 Verbatim user complaint (2026-04-19)¶

1.2 Concrete defects visible in the 2026-04-19 screenshot¶

1.3 Diagnosis¶

2. Reference research¶

2.1 Elecraft K3 / K3S / KX3¶

2.2 Icom IC-756PRO / IC-756PROIII / IC-746PRO¶

2.3 Icom IC-7300¶

2.4 Kenwood TS-990S¶

2.5 Yaesu FTDX101D / MP¶

2.6 Vintage bonus — Icom IC-781 / Yaesu FT-1000MP¶

3. Three variants¶

3.1 Variant A — "Elecraft Column" (icons on the right rail)¶

Single-RX wireframe¶

Dual-RX wireframe¶

CSS grid pseudocode¶

3.2 Variant B — "Dual Cockpit" (two-column, TS-990S-style)¶

Single-RX wireframe¶

Dual-RX wireframe¶

CSS grid pseudocode¶

3.3 Variant C — "Modern Icom" (scope dominant, 60/40 split)¶

Single-RX wireframe¶

Dual-RX wireframe¶

CSS grid pseudocode¶

4. Recommendation¶

Why B wins¶

Why not A¶

Why not C (as headline)¶

5. Atomic implementation plan¶

PR1 — refactor(#NEW1): extract AmberCockpit component (behavior-preserving)¶

PR2 — feat(#NEW2): dual-cockpit grid — VFO B as peer (replaces scope of parked #845)¶

PR3 — refactor(#NEW3): per-VFO indicator zone + global indicator strip¶

PR4 — feat(#NEW4): scope track fills remaining height (absorbs D4 empty amber)¶

PR5 — feat(#NEW5): mobile container-query collapse + telemetry reserve¶

Obsolete vs survivable queue items¶

6. Migration from current state¶

Migration steps¶

Rollback plan¶

Risks¶

7. Open questions¶

Sources (references)¶

PR1 — `refactor(#NEW1): extract AmberCockpit component (behavior-preserving)`¶

PR2 — `feat(#NEW2): dual-cockpit grid — VFO B as peer (replaces scope of parked #845)`¶

PR3 — `refactor(#NEW3): per-VFO indicator zone + global indicator strip`¶

PR4 — `feat(#NEW4): scope track fills remaining height (absorbs D4 empty amber)`¶

PR5 — `feat(#NEW5): mobile container-query collapse + telemetry reserve`¶