LCD Layout Redesign — Design Spike¶

Date: 2026-04-19 Surface: frontend/src/components-v2/panels/lcd/AmberLcdDisplay.svelte Skin: amber-lcd (fallback skin for radios lacking panadapter/scope) Prerequisite for: 2026-04-18-lcd-display-enhancements.md §5.2 (per §8.7) Triggered by user feedback (2026-04-18): "Имеет смысл пересмотреть layout самого LCD дисплея, с учётом двух VFO, индикаторов и спектра/фильтра." Status: design only — no code changes in this document.

User clarification (2026-04-19): "спектр" in the triggering quote refers to AF spectrum (the combined AF FFT + passband + PBT + notch/contour widget currently in AmberAfScope). Confirmed — the LCD skin is precisely for radios without an RF panadapter, so the only spectrum on-screen is AF. Resolves open question Q1.

1. Context — current `AmberLcdDisplay` structure¶

Read of AmberLcdDisplay.svelte (138–270) gives the following vertical stack inside .lcd-screen (flex-column, gap: 6px, padding: 12px 18px):

#	Region (class)	Height (approx)	Content
1	`.lcd-meter-row` (main)	~28 px	`AmberSmeter` + source cycle button (S/PO/SWR/ALC/COMP)
2	`.lcd-meter-row.lcd-meter-sub`	~20 px (scaleY 0.7, dim)	Sub-RX S-meter, dual-RX only, hidden during TX
3	`.lcd-ind-row`	~24 px, wraps (!)	20+ status tokens, 3 hard `<span class="ind-sep">` gutters, flex-wrap
4	`.lcd-vfo-scope-row`	~96–110 px	VFO A frequency (`AmberFrequency size="large"`) + mode-box + band-box, with scope strip 30 %
5	`.lcd-vfo-row.lcd-vfo-sub`	~38 px (dual-RX only)	VFO B frequency (`AmberFrequency size="small"`) + `B` tag + mode
6	`.lcd-vfo-ctrl-row`	~24 px	A↔B, A=B, DW, SPLIT, XIT, CLR, TUNE, BK-OFF… (wraps)
7	`.lcd-rit-row`	~20 px (conditional)	RIT/XIT offset readout

Ink model (important for redesign): two-layer DOM, ghost "888.888.888" at α 0.06 + active digits at α 1.0 absolutely over it. All tokens use JetBrains Mono except the digit blocks which use DSEG7 Classic. The s-meter and AmberAfScope render to canvas with the same ink colour.

Capability gates already present: - hasDualReceiver() → rows 2 + 5. - hasAudioFft() → scope strip inside row 4. - hasCapability(...) on ~12 tokens inside row 3. - isCwMode / hasCapability('break_in') → BK-OFF button in row 6.

Current problems (grounded in the file)¶

VFO B is a second-class citizen. Row 5 is .lcd-vfo-sub — font small, tag dim, sub-mode α 0.45, no band, no meter tied to it. For a dual-RX operator running cross-band split or dual-watch, this is the wrong hierarchy.
Row 3 (indicators) is a wall of 20+ tokens with only three separators. At 1366 × 768 it wraps to two or three lines, pushing every row below downward and eating VFO space. The flex-wrap: wrap on .lcd-ind-row is a reflow grenade.
Filter-viz is anchored to VFO A's row, not the display as a whole. This makes "promote to full-width" (enhancements §3) a reflow-heavy operation: moving the strip out of row 4 requires restructuring row 4 anyway.
Single-RX and dual-RX share one linear layout with {#if} toggles. Every capability flip (enabling DW mid-session, hot-adding sub-RX in a future backend) triggers layout reflow of everything below.
VFO B has no meter, no band tag, no mode-box parity. Yet a dual-watch operator stares at VFO B's S-meter for split-frequency DX listening.
The VFO control buttons (row 6) cross the display/control boundary. A physical rig's display doesn't carry soft buttons; they belong on the bezel (sidebar). They currently eat ~24 px of the vertical budget.

16 : 7.5 surface budget¶

.lcd-slot is aspect-ratio: 16/7.5. On a 1600 px wide main column, that's 1600 × 750 CSS px. After .lcd-screen's 12 × 2 + 18 × 2 padding and the 4 px .amber-lcd outer padding, usable interior is ~1556 × 714 px.

That is enough. But it is not abundant. Every row must pay rent.

2. Design principles¶

Six principles. Each argued briefly; each has a consequence cited downstream.

P1. Dual-VFO equal weight by default, demotion on capability absence.¶

A dual-RX radio's two receivers are not primary/secondary — they are two independently-tuned DSP chains. The current "VFO A big, VFO B small" is a UI lie about the hardware. The redesign's default layout reserves equal horizontal slab area for both VFOs. Single-RX radios collapse VFO B's slab to zero; the grid's other columns absorb the space deterministically (no reflow of VFO A's dimensions).

Consequence: we need a grid (not a flex-column) so we can collapse a named cell without re-laying out its neighbours.

P2. Capability-driven cell hide without layout reflow.¶

Hiding VFO B, the scope strip, the memory strip, or the telemetry strip must never change the position or size of VFO A, the status row, the meter, or the filter-viz. Operators already rely on muscle-memory placement.

Consequence: the grid must use named grid-template-areas and fixed grid-template-rows / grid-template-columns in each variant. Hiding a cell = visibility: hidden + the area stays allocated, or a variant switch to a grid definition where the absent cell's row/column has collapsed to zero and the remaining areas have identical absolute dimensions to the other variant. We choose the second (§3).

P3. Indicators legible at 2 m — group, separate, stabilise.¶

Row 3's 20+ tokens must not wrap. Wrapping breaks scan paths. Solution: group into four fixed-width bays (TX / RF-chain / DSP / VFO-state). Bays are grid-template-columns with explicit widths; tokens inside a bay can ellipsize or hide individually, but the bay's horizontal position is immutable. Separators (ind-sep) become bay borders, not inline pseudo-spans.

Consequence: the status row becomes a 4-column sub-grid. Capability- absent tokens render as a dim placeholder (·) — the bay shape is preserved.

P4. Filter-viz is a peer of VFO, not an accessory.¶

On a scope-less radio, the filter-viz (passband shape + PBT brackets + notch/contour + AF FFT inside) is the operator's only visual representation of the receive chain. It deserves a named cell with its own grid row, not "30 % strip pinned to VFO A". Per the combined-widget note in the task brief, this widget aggregates 4 layers (AF FFT, passband, PBT, notch/contour). Its vertical budget should be sized for the worst-case (FTX-1 with all 4 layers visible), not the best-case (passband only).

Consequence: filter-viz gets a dedicated filter area, ~120 px tall, full display width minus gutters. Dual-RX and single-RX share the same allocation.

P5. Soft buttons belong on the bezel, not on the screen.¶

The amber LCD is a display. The current row 6 (VFO control buttons) puts soft buttons inside the screen. This: - Costs ~24 px of screen real estate. - Mimics a control surface the display itself never had on any real rig. - Duplicates controls already present in LeftSidebar / RightSidebar.

Redesign moves A↔B, A=B, DW, SPLIT, XIT, CLR, TUNE, BK-OFF to the right sidebar's VFO control panel (existing panel expanded). The display surfaces only state, never command entry.

Consequence: row 6 is deleted from the LCD grid. ~24 px freed, given to the filter-viz cell. The RIT-offset row (currently conditional row 7) gets folded into VFO A's right-side inline region (PB cell).

P6. Mode-specific variation is a content concern, not a layout concern.¶

CW vs SSB vs DATA can change what the filter-viz draws (CW: ±500 Hz scale; DATA: wider passband outline), what the meter shows (CW: key-down current if wired), what the VFO mode-box prints (CW vs CW-R). But the grid cells don't move. Layout is invariant under mode change.

Consequence: no {#if mode === 'CW'} at the grid level. Mode gates live inside individual components.

3. Grid proposal — unified 8 × 12 grid¶

All variants share one grid. Capability gates collapse rows or columns, never move areas.

Grid columns:          [ A=left-gutter  1fr  1fr  1fr  1fr  1fr  1fr  1fr  1fr  1fr  1fr  right-gutter ]
                          12 px         —— 12 fluid unit columns ——                                 12 px

Logical column bands:  [ gutter | VFO-A-block (6u) | VFO-B-block (6u) | gutter ]
                                       ←            split 50/50            →

3.1 Row template (worst case, dual-RX with full enhancements)¶

row-1   status      28 px   [ status-tx | status-rf | status-dsp | status-vfo ]  (4 bays)
row-2   meter-A     28 px   [ meter-A spans VFO-A-block |  meter-B spans VFO-B-block ]
row-3   vfo         72 px   [ vfo-A (freq + mode + band)  |  vfo-B (freq + mode + band) ]
row-4   pb          22 px   [ rit/xit A | offset | . . . | rit/xit B | offset     ]
row-5   filter    120 px   [ filter-viz spans full width, shared between A and B (§5)  ]
row-6   aux        80 px   [ memory strip (7u) | telemetry (5u) ]

Total: 28 + 28 + 72 + 22 + 120 + 80 = 350 px of content, plus 6 × gap: 6px = 30 px, plus top/bottom padding 24 px → ~404 px. Fits inside our ~714 px interior with ~310 px margin for growth.

That 310 px margin is deliberate. It allows: - Contrast preset MAX pushing inactive α to 0.30 (larger visual mass). - vfo row upsized to 96 px on 4K monitors (digit scaling). - One future row (e.g. CW decode text, RTTY FSK indicator) without a redesign.

3.2 Single-RX idle¶

VFO-B-block columns collapse: the grid's second 6u band shrinks to 0u, leaving a single 12u band. Row heights unchanged. meter-B, vfo-B areas disappear (no reflow — the columns underneath them simply do not exist in this variant).

┌──────────────────────────────────────────────── 1556 px ────────────────────────────────────────────────┐
│ [TX VOX PROC]  │  [ATT AMP1 DIGI IP+ ATU]  │  [NB NR CONT NOTCH ANF AGC-M]  │  [RFG SQL RIT SPLIT LOCK] │ row-1 status
├──────────────────────────────────────────────────────────────────────────────────────────────────────────┤
│ S ▁▂▃▄▅▆▇█ 9+20dB                                                                           [S|PO|SWR] │ row-2 meter
├──────────────────────────────────────────────────────────────────────────────────────────────────────────┤
│ [A]   14.195.678   USB 2.4k                                                                      [20m] │ row-3 vfo
├──────────────────────────────────────────────────────────────────────────────────────────────────────────┤
│                                                                                                         │ row-4 pb (empty)
├──────────────────────────────────────────────────────────────────────────────────────────────────────────┤
│ ┌──────────── filter 2.4k  FIL2  Shift +320  PBT −140/+60  Contour 3  Notch ■ ─── AF FFT ─────────────┐│
│ │    ‐6 ┌──[───]──┐                    ▁▃▅▇█▇▅▃▂                                                     ││
│ │ ‐20 ──┘          └──                                                                               ││ row-5 filter
│ │  ├──┼──┼──┼──┼──┼──┤  −3k −2k −1k  0  +1k +2k +3k                                                  ││
│ │                    ▲ (centre)                                                                      ││
│ └────────────────────────────────────────────────────────────────────────────────────────────────────┘│
├──────────────────────────────────────────────────────────────────────────────────────────────────────────┤
│ [M1 14.195 U DX] [M2 18.100 U FT8] [M3 21.280 SSTV] [M4 28.074 FT8]  │  VD 13.8  TEMP 42  ID 0.2A     │ row-6 aux
└──────────────────────────────────────────────────────────────────────────────────────────────────────────┘

3.3 Single-RX TX¶

Changes vs idle: - Row 1: TX token glows red (ind-tx class). ATU may switch to TUNE (blink). - Row 2 meter auto-source flips S → PO (existing behaviour). - Row 5 filter-viz dims the AF FFT layer (no useful input), passband shape unchanged. This is correct even for SSB (TX sidebands are not echo'd through the AF path). - No grid row moves. No cell resizes. The content changes.

│ [TX] [VOX] [PROC 6]  │  [ATT AMP1 · · TUNE*]  │  [NB · · · · AGC-M]  │  [· · · SPLIT ·]   │ row-1
│ PO ▁▂▃▄▅▆▇▇▇▇ 75 W                                                        [PO]            │ row-2
│ [A]   14.195.678   USB 2.4k                                             [20m]             │ row-3
│                                                                                            │ row-4
│ filter + dimmed FFT                                                                        │ row-5
│ memories unchanged  │  VD 13.2 TEMP 48 ID 18A                                              │ row-6

* = blink. · = placeholder for absent token (bay width preserved).

3.4 Single-RX, mode-specific content notes¶

Layout invariant. Content deltas only:

Mode	Filter row scale	Status row delta	VFO mode-box
SSB	±3 kHz, minor 500 Hz	NR/NB gates normal	`USB` / `LSB`
CW	±500 Hz, minor 100 Hz	add `BK` badge if break-in, hide DATA	`CW` / `CW-R`
DATA	±3 kHz, outline shows wide roofing	`DATA` token active	`USB-D` / `LSB-D`
AM	±10 kHz, minor 2 kHz	hide NR if backend strips it in AM	`AM`
FM	±15 kHz, minor 5 kHz	squelch token promoted to active bay	`FM`

3.5 Dual-RX idle (default, both receivers listening)¶

┌──────────────────────────── 1556 px ────────────────────────────────────────────────────────────────────┐
│ [TX VOX PROC]  │  [ATT AMP1 DIGI IP+ ATU]  │  [NB NR CONT NOTCH ANF AGC-M]  │  [RFG SQL RIT SPLIT LOCK] │ row-1
├──────────────────────────────────────── 6u ────────────────────┬──────────────────── 6u ────────────────┤
│ S ▁▂▃▄▅▆▇█ 9+20 [S|PO|SWR]                                      │ S ▁▂▃▄ 5 [S]                          │ row-2
├─────────────────────────────────────────────────────────────────┼───────────────────────────────────────┤
│ [A]● 14.195.678   USB 2.4k            [20m]                     │ [B]  7.074.000   USB 2.4k   [40m]     │ row-3
├─────────────────────────────────────────────────────────────────┴───────────────────────────────────────┤
│ (RIT row: empty when both offsets are zero)                                                             │ row-4
├──────────────────────────────────────────────────────────────────────────────────────────────────────────┤
│                                filter-viz (shared, active-VFO anchored)                                 │ row-5
├──────────────────────────────────────────────────────────────────────────────────────────────────────────┤
│ [M1 … M4]  (4 instead of 5 — memory cell narrower)  │  VD 13.8  TEMP 42  ID 0.2A                        │ row-6
└──────────────────────────────────────────────────────────────────────────────────────────────────────────┘

Key differences vs §3.2: - VFO-B-block has identical dimensions to VFO-A-block. Same font size for frequency, same mode-box, same band-box. - [A]● marker shows which VFO is active (receiver-controlled). Moves to [B]● on swap. This replaces the current "active VFO font-weight" hack with an explicit dot. - Meter-B uses the same AmberSmeter widget, not the scaled-0.7 degraded version. Source is locked to S (you cannot change a sub-RX meter source; it's an S-meter only). - Filter-viz is still shared (§5) — a single widget anchored to whichever VFO is active.

3.6 Dual-RX SPLIT / DUAL-WATCH¶

Visually identical to §3.5, plus: - Row 1 SPLIT token active (or DW token, see §6.2). - Row 3: the transmit VFO gets a small red TX pip next to its tag (e.g. [A] is RX, [B] TX means "listening on A, transmitting on B" — classic DX split). - Row 4 pb: if the offset between VFO A and VFO B is small (< 10 kHz, "close split"), show Δ = +2.3 kHz in row-4. Useful. Disappears for cross-band split.

│ [TX VOX PROC]  │  [ATT ...]  │  [... AGC]  │  [... SPLIT* ...]                            │ row-1
│ S 9+20    [S|PO|SWR]   │ S 5    [S]                                                        │ row-2
│ [A]● 14.195.678 USB 2.4k [20m]   │ [B]TX 14.197.300 USB 2.4k [20m]                         │ row-3
│ Δ = +1.622 kHz                                                                              │ row-4
│ filter-viz (anchored to active RX)                                                          │ row-5
│ memories │ telemetry                                                                        │ row-6

Dual-watch (same-band, shared-filter) variant: row 4 shows DW instead of Δ. Row 2's meter-B is now the same DSP chain as meter-A (diversity receiver), so both meter needles track together — a visual hint that they're coupled.

3.7 CW / SSB / DATA in dual-RX¶

No grid changes. Per-VFO mode-box content differs. The filter-viz is anchored to the active VFO's mode — this means switching A↔B can rescale the filter-viz x-axis. Acceptable; the scale ticks re-label.

4. Cell inventory¶

Every cell, with capability gate and empty-state behaviour. Cells are named after their grid-area.

Area	Purpose	Capability required	When absent
`status-tx`	TX / VOX / PROC bay	always present	never empty (TX is always shown; VOX/PROC gated inside bay with `·` holder)
`status-rf`	ATT / AMP / DIGI / IP+ / ATU bay	per token (`attenuator`, `preamp`, `digisel`, `ip_plus`, `tuner`)	bay always present; individual tokens → `·`
`status-dsp`	NB / NR / CONT / NOTCH / ANF / AGC bay	per token	same
`status-vfo`	RFG / SQL / RIT / SPLIT / DATA / LOCK bay	per token	same
`meter-A`	Main RX meter + source cycle button	always	never
`meter-B`	Sub RX S-meter, full-size	`hasDualReceiver()`	column collapses; `meter-A` widens to 12u
`vfo-A`	VFO A freq (DSEG7) + mode-box + band-box + active dot	always	never
`vfo-B`	VFO B freq (DSEG7) + mode-box + band-box + active dot	`hasDualReceiver()`	column collapses; `vfo-A` widens to 12u
`pb-A` / `pb-B`	RIT/XIT offset or close-split Δ	`rit` or `split`	row collapses to 0 px
`filter`	Full-width filter-viz (AF FFT + passband + PBT + notch/contour)	always	row remains; internal layers gate separately (no AF FFT → outline only)
`memory`	Memory / recent-QSY strip, 7u wide	(always; falls back to QSY-only if no backend memories)	row remains; if telemetry also absent, row collapses to 0 px
`telemetry`	VD / TEMP / ID tiles, 5u wide	backend reports any of VD/TEMP/ID	5u column gives width to `memory`; if both memory+telemetry empty, row collapses

4.1 Reflow rules summarised¶

Collapsing VFO-B-block does not resize VFO-A-block. Instead, VFO-A-block's column span changes from 6u → 12u via a different grid template selected by hasDualReceiver(). Both templates have identical row heights, so the vertical position of row-4, row-5, row-6 is byte-for- byte equal across variants.
Collapsing pb row (no RIT/XIT active) uses grid-template-rows with minmax(0, auto) on that row. Row-5 filter moves up exactly 22 px. Acceptable — filter-viz doesn't have absolute positioning dependencies elsewhere.
Collapsing aux row (no memories, no telemetry) frees 80 px. We do not grow the filter row to fill it; the empty space below the filter is fine. Operators reading S-meters will notice more growth than empty bottom.

5. Interaction with enhancement plan §3 and §4¶

The redesign re-sequences §3 and §4 as follows.

5.1 Filter-viz wide mode (enhancements §3) — survives, promoted¶

The redesign's row-5 is the wide filter-viz. §3's pixel proposals (Hz ruler, centre marker, -6 / -20 dB ticks, IF-shift arrow, PBT brackets, mode-aware envelope) all land unchanged. The only delta: §3's compact-variant fallback for dual-RX (30 %-wide strip next to VFO A) is deleted. In the new grid, filter-viz is always full-width, always shared between A and B. The current "compact scope strip inside row 4" disappears.

This is a simplification win. §3 previously had to specify both wide and compact layouts; the redesign forces one.

5.2 Memory / recent-QSY strip (enhancements §4.1) — survives, narrower¶

§4.1 planned 8 cells at ~12 % each. In the new grid's memory area (7u out of 12u = ~58 % of display width = ~900 px), we fit 5 cells at ~180 px each (or 4 cells in dual-RX where telemetry takes 5u and memory is 7u minus 0u). Cell content unchanged: first 5 memory slots, or memories + 2 recent QSY in single-RX.

In dual-RX, memory strip is not demoted to the sidebar (which was §4.1's fallback). It stays on the LCD, just narrower — the user asked us to treat dual-VFO as first-class, and memories remain part of the glance- scannable surface.

5.3 Telemetry strip VD/TEMP/ID (enhancements §4.2) — survives, same size¶

§4.2's three tiles fit in 5u (~680 px) comfortably — ~220 px per tile, each with big DSEG7 + sparkline + label. Unchanged.

5.4 Items deprioritised or deleted by the redesign¶

Dual-RX memory-strip demotion to sidebar (§4.1 fallback) — deleted. Memory strip stays on the LCD in both variants.
Scope-strip compact mode for dual-RX (§3 fallback) — deleted. Filter-viz is always full-width.
Sub-meter scaleY-0.7 degradation (current code) — deleted. Meter-B is full-size in the new grid.
VFO control buttons inside the LCD surface (current row 6) — moved to right sidebar. Not lost, just relocated per P5.

6. Status indicator organisation¶

Current row 3 has these 20 tokens (in source order, gated by capability):

TX VOX PROC    | ATT [AMP1|AMP2|IPO] DIGI-SEL IP+ ATU(or TUNE) | NB NR CONT NOTCH ANF AGC-<mode> | RFG SQL RIT SPLIT DATA LOCK

6.1 Four-bay grid (new)¶

┌──────────────── status-tx ────┬──────── status-rf ────────┬──────────── status-dsp ──────────┬──────── status-vfo ────────┐
│  TX   VOX   PROC 6            │  ATT   AMP1   DIGI   IP+  │  NB 3   NR 5   CONT   NOTCH      │  RFG   SQL   RIT   SPLIT   │
│                               │  TUNE*                    │  ANF    AGC-MID                  │  DATA  LOCK                │
└───────────────────────────────┴───────────────────────────┴──────────────────────────────────┴────────────────────────────┘
       3u                                3u                             3u                                3u

Each bay is a display: grid; grid-template-columns: repeat(auto-fill, minmax(...)) inside its 3u slot. Tokens flow left-to-right inside their bay and wrap within the bay only — never across bay boundaries.

Bay widths are equal (3u each) rather than content-fit. This gives consistent horizontal anchors — operators know AGC always sits in the third quarter of row 1. A real-rig analogue: the IC-7610's status line has fixed regions, not a flex-wrap.

Criteria: low frequency of change + already present in sidebar controls.

Token	Reason for demotion	New home
`RFG`	Always "on" when rf_gain < max; the slider is on the sidebar already.	Sidebar RF-gain slider
`DIGI-SEL`	Rarely toggled; a per-band setting.	Sidebar DSP panel
`IP+`	Rarely toggled; a preamp companion setting.	Sidebar RF panel
`CONT`	The contour level is visible in filter-viz as the U-curve. Redundant to show as a token.	Deleted from LCD
`ANF`	Notch-class token; keep ANF in the DSP bay only if capability present, else hide.	Same

After demotion:

status-tx (3u): TX VOX PROC → unchanged.
status-rf (3u): ATT AMP DIGI IP+ ATU → ATT AMP ATU (3 tokens, DIGI/IP+ → sidebar).
status-dsp (3u): NB NR CONT NOTCH ANF AGC → NB NR NOTCH AGC (4 tokens, CONT → filter-viz only, ANF → inside NOTCH badge).
status-vfo (3u): RFG SQL RIT SPLIT DATA LOCK DW → SQL RIT SPLIT DW DATA LOCK (6 tokens, RFG → sidebar).

Result: 20+ tokens → 16 tokens in 4 equal bays. Row 1 fits on one line at 1366 × 768 without wrapping. Below 1366 px the whole LCD skin already degrades gracefully (it's wide-screen-oriented); no special case needed.

6.3 Visual separators¶

.ind-sep (currently a 1 px dim line) becomes a proper CSS border-right on each bay except the last. Width 1 px, colour var(--lcd-ink-ghost). Height 100 % of row-1 (not 16 px as currently). This visually anchors the bays as distinct regions.

6.4 `DW` token¶

The current code does not render a DW indicator (only a DW button). Add DW to the status-vfo bay, gated by hasCapability('dual_rx') and reactive to vfoOps.dualWatch. This makes the bay a state surface, not just a label dump.

7. Typography hierarchy¶

Existing file uses font-size: 14px for indicators, 22px for the VFO-A tag, 18px for mode/band boxes, 16px for sub-VFO tag, 13px for sub-mode, 12px for RIT label, 11px for buttons, 10px for meter-source button. Inconsistent.

7.1 Proposed scale¶

Seven named tokens, matching CSS custom properties on .amber-lcd:

Token	px	Weight	Usage
`--lcd-font-freq-primary`	48	bold	VFO A / VFO B DSEG7 digits (single-RX and dual-RX equal)
`--lcd-font-mode-box`	20	700	Mode-box, band-box (both VFOs)
`--lcd-font-vfo-tag`	22	700	`[A]`, `[B]` tags — equal size both VFOs
`--lcd-font-status-ind`	14	700	Status row tokens (all four bays)
`--lcd-font-meter-label`	12	600	Meter scale marks, source button
`--lcd-font-pb-offset`	16	600	RIT/XIT offset, close-split Δ (DSEG7 mono)
`--lcd-font-aux-label`	11	700	Memory label, telemetry unit, aux-row microcopy

Key change: --lcd-font-freq-primary is the same for VFO A and VFO B. Per P1, VFO B is not demoted typographically. The active receiver gets the [A]● active-dot treatment; inactivity is signalled by the dot, not by shrinking the digits.

7.2 Active vs inactive emphasis (no size change)¶

When a VFO is inactive, its frequency digits use --lcd-ink-inactive at the active-layer α (respecting contrast preset). Ghost layer unchanged. Ratio of active-VFO : inactive-VFO ink is ~1 : 0.6 — visible, scannable, not "shrunken".

This replaces the current "small font = secondary" idiom with an "ink density = focus" idiom. The latter is closer to how real rigs work (segments all same size; only the backlit ones are "loud").

7.3 Responsive scaling¶

All tokens listed above are in px. On very small viewports (.lcd-slot narrower than 1100 px), a @container-driven multiplier drops them to 80 %. Below 900 px, we'd be outside the skin's target envelope anyway.

8. Transition plan¶

Options:

8.a Redesign-first, enhancements adapt.¶

Ship the grid refactor (§9.1) first. Then §8.1–§8.6 of the enhancements plan each merge into the new grid. Risk: enhancements land later (2–3 weeks). Reward: enhancements land correctly sized into their final cells — no rework.

8.b Enhancements-first, redesign reshuffles.¶

Land §8.1 (contrast tokens), §8.3 (wide filter-viz), §8.4 (memory), §8.5 (telemetry) into the current layout. Then redesign, breaking all four. Risk: significant rework in the redesign PR — filter-viz, memory, telemetry all need their DOM parent changed. Reward: users get the enhancements a week earlier.

8.c Coordinated sequence (recommended).¶

Land redesign scaffold first (Issue §9.1 below). This is a layout-only change: introduces the grid, moves VFO B to peer status, moves control buttons to sidebar, splits status row into 4 bays. No new widgets, no new data sources. ≤ 3 files, ≤ 200 LOC.
Land enhancements §8.1 (contrast tokens) in parallel. This is pure CSS-variable plumbing — orthogonal to the grid.
Land filter-viz wide mode (§8.3) on top of the new grid. The new filter area is already reserved; §8.3 just fills it. Because we deleted the compact variant (§5.1), §8.3 is simpler than the original plan.
Land memory strip (§8.4) and telemetry strip (§8.5) into the aux row. Both already have their cells allocated.
Dual-RX VFO-B promotion (Issue §9.2). Ship the full-size meter-B, full-size vfo-B, active-dot marker.
Status row reorg (Issue §9.3). Demote tokens to sidebar, add DW indicator, bay borders.

Net: one layout PR up-front, five content PRs in sequence. Each ≤ 3 files.

Recommendation: 8.c. It respects the existing enhancement plan's atomicity while preventing downstream rework.

Why not 8.a? 8.a stalls user-visible improvement for weeks. Why not 8.b? 8.b guarantees churn — every enhancement PR would later be reworked when the grid changes.

9. Atomic implementation issues¶

Three issues. Each ≤ 3 files, ≤ 200 LOC, independent after #9.1 lands.

Issue 9.1 — LCD grid scaffold¶

Title: refactor(lcd): unified grid with named areas for VFO A/B, status bays, filter, aux

Files: - frontend/src/components-v2/panels/lcd/AmberLcdDisplay.svelte (swap flex-column for grid-template-areas; introduce hasDualReceiver template switch; delete .lcd-vfo-ctrl-row; delete .lcd-rit-row at top level and fold into pb-A/pb-B areas). - frontend/src/components-v2/layout/LcdLayout.svelte (no change expected — grid is internal to AmberLcdDisplay; touched only if .lcd-slot min-height needs adjustment). - frontend/src/components-v2/panels/controls/VfoControlPanel.svelte (new home for A↔B, A=B, DW, SPLIT, XIT, CLR, TUNE, BK-OFF — move buttons, don't duplicate).

Out of scope: status-bay reorg (§9.3), VFO-B size parity (§9.2), filter-viz promotion (§8.3 in enhancements).

Estimated LOC: ~160 (mostly CSS grid definitions, some JSX reshuffling).

Issue 9.2 — VFO B promotion to peer¶

Title: feat(lcd): VFO B renders at peer size with meter, band, active-dot marker

Files: - frontend/src/components-v2/panels/lcd/AmberLcdDisplay.svelte (swap <AmberFrequency size="small"> for "large" for VFO B; add band tag and mode-box in B's cell; add active-dot render). - frontend/src/components-v2/panels/lcd/AmberFrequency.svelte (only if a new size="peer" variant differs from "large" — ideally not; use "large" for both). - frontend/src/components-v2/panels/lcd/AmberSmeter.svelte (delete the 0.7-scale sub-meter styling class; meter-B uses full-size).

Scope: - VFO B uses --lcd-font-freq-primary (48 px). - VFO B gets band-box + mode-box identical to VFO A. - Active VFO shown via [A]● / [B]● dot glyph (not font weight, not size). - Sub-meter rendered full-size in meter-B area; source locked to S. - Inactive VFO uses --lcd-ink-inactive at current contrast preset's α — no separate dim class.

Depends on: #9.1 merged.

Estimated LOC: ~120.

Issue 9.3 — Status row four-bay reorganisation¶

Title: refactor(lcd): status row → four fixed bays, demote low-frequency tokens to sidebar

Files: - frontend/src/components-v2/panels/lcd/AmberLcdDisplay.svelte (replace single .lcd-ind-row with four grid-area bays; remove RFG/DIGI-SEL/IP+/CONT tokens; add DW state indicator). - frontend/src/components-v2/panels/controls/RfControlPanel.svelte (surface DIGI-SEL and IP+ controls, since they're no longer in the LCD status row; if panel doesn't exist, create in this PR — still ≤ 3 files). - frontend/src/components-v2/panels/controls/DspControlPanel.svelte (same idea for RFG — if existing, just verify it's surfaced; no work likely needed).

Scope: - Four bays with equal 3u widths. - Bay separators via border-right in --lcd-ink-ghost. - Absent tokens render as · placeholder (preserve bay shape). - Add DW indicator to status-vfo bay, reactive to vfoOps.dualWatch. - Verify that DIGI-SEL / IP+ / RFG controls are reachable from sidebar (create sidebar rows if missing).