Tyler · research · 31 Oct 2026 (HKT)

Seams

What the line between two butted tiles really is, which panels and LED modules make it thinnest, and what a printed enclosure will realistically achieve. Includes heat, safety and tolerances. est. = my estimate or engineering assumption, not a datasheet value.

Short answer

No buyable 10″ square panel has a border under 3.0 mm per side (Youritech ET100SQ01-GX: 185.82 − 179.82 = 6.0 mm → 3.0 mm each, L/R). Today's enclosure (wall 2 + clearance 0.3 + lip 3) gives a spec seam of 10.6 mm, about 11.7 mm realistic. The thinner-panel gain is real but small; the enclosure is the bigger lever.

"Barely noticeable" (≤ 1 mm) is not reachable with a printed enclosure and any buyable small-quantity LCD/OLED. Best realistic: ~3 mm with a flush-edge (glass-to-edge) design on a ~0.9 mm-border OLED, ~1.1 mm extra on LED modules (which butt on the pixel pitch). The ≤1 mm club is video-wall LCDs (0.88 mm spec, ~1.4 mm installed est.) and fine-pitch LED cabinets, none of which are Tyler-sized or buyable singly.

Tail edge: the flex-tail side is far worse than L/R. Current model: 198.7 mm tile pitch − 179.6 mm active ≈ 19 mm vertical seam est. (assumes centred active area). Fix the tail before chasing 0.5 mm elsewhere.

Threshold: what counts as "barely noticeable"

My working rule (a judgement, not a standard): ≤ 1.0 mm combined seam = barely noticeable at normal viewing distance (≥ ~1 m, 1 mm ≈ 3.4 arcmin, reads as a hairline, same class as the 0.88 mm commercial video-wall spec). 1.0–1.5 borderline · ≤ 3 thin line · ≤ 6 clear gap · > 6 a frame. A dark line on a bright image is still visible well below 1 mm up close (< 0.5 m), so treat ≤ 1 mm as "stops being distracting", not "invisible".

Ranked shortlist (by realistic combined seam)

Seam = visible dead band between the active areas of two butted tiles, L/R edges. Spec = datasheet numbers + enclosure nominal. Realistic = adds panel outline ±0.2, print ±0.3, active-area offset ±0.2/side, butt gap 0.3 (see Tolerances). Rank uses the flush-edge realistic seam (same order in every enclosure variant). Border per side = (outline − active) ÷ 2 unless marked datasheet drawing. Ranked rows are all buyable in small quantity.

Panel only = 2 × border (bare panels butted, no enclosure) Current = wall 2.0 + clearance 0.3 + lip 3.0 over the border (lip > border eats active area) Thin wall = wall 1.2 + clearance 0.2, lip = border Flush edge = no front wall or lip; glass/PCB edge is the tile edge

Benchmarks (not buyable at Tyler size, not ranked)

Item	Class	Border/side	Seam spec	Seam installed	Note

Seams drawn to scale

Same scale on every row (1 mm = a fixed length; the thin dashed lines mark the 1 mm "barely noticeable" band). Blue = active picture, black = panel border + lip, grey = enclosure wall + clearance, orange = realistic tolerance allowance.

activeborder / lipwall + clearancetolerance allowance

Recommendation

10″ Tyler

Keep ET100SQ01-GX for now (only buyable 10″ square; 3.0 mm L/R). Gain comes from the enclosure: moving from current to thin-wall cuts realistic seam 11.7 → 10.1 mm; flush-edge → 7.1 mm.

To go below ~6 mm at 10″ you need a custom panel (ask Youritech / Tailor Pixels / All Vision for a ≤1.5 mm-border 10″ square; All Vision "<2.5 mm" is a marketing claim unverified). Add this to the outreach tracker.

4″ Tyler

Hot Display TFT-H040A20DKINV6N40: 1.451 mm on L/R/top (datasheet drawing) vs 2.54 mm for ET040HD20-K, saves 2.2 mm per joint panel-only. Trade-off: parallel RGB + 3-wire SPI (NV3052C), 600 nit, backlight 16.5 V × 60 mA ≈ 1 W; no MIPI.

If the tile may grow to 5″: ET050SQ01-GX (1.15 mm, MIPI, 400 nit, ~$24).

Pixie (LED)

HUB75 modules are pitch-continuous: the board edge is N × pitch, so the seam is only what the enclosure adds. P3 64×64 = 192 × 192 mm matches the 10″ tile footprint; P2.5 64×64 = 160 mm. With a flush-edge shell the extra is ~1.1 mm (≈ 0.4 px at P2.5); with a thin wall ~4.1 mm (≈ 1.6 px). Edge-LED offset is not published unverified, so measure a sample.

What each change implies

New calc() numbers only: swap panel dimensions (use datasheet 185.82×192.32 / 179.82 for the 10″, currently 186×192 / 179.6); set lip ≤ border (1.45 at the Hot Display) and wall toward 1.2 where the sliders allow. Same layer stack, same magnet pads.

Enclosure redesign needed for: thin wall with magnet/pogo pockets (local inner pads), a re-routed tail edge (the ~19 mm vertical seam), flush-edge / glass-to-edge (back-tray shell, bonded cover, edge protection), a different panel's connector and flex fold, any LED module (new tray, metal back, vents).

Heat & Safety

Class	Operating / storage	Heat source	Sealed-tile behaviour
IPS LCD (ET100SQ01-GX, ET050SQ01-GX, Hot Display 4″)	−20…70 °C / −30…80 °C [Youritech pages, Hot Display datasheet]	Backlight. Hot Display 4″: 16.5 V × 60 mA ≈ 1.0 W at 600 nit, LED life 20–30 kh to 50 % at 25 °C (datasheet). Scaled to a 10″ at 1000 nit (area ×6.3, brightness ×1.7) ≈ ~10 W est. (6–15 W); ET100 backlight spec not published.	~10 W in a ~0.09 m² sealed shell → surface +11–14 K, interior hot spots +25–35 K est. → 55–70 °C in a 30–35 °C room: at the 70 °C limit and above PLA/FR-PETG. High temperature lowers contrast and ages polarizer/backlight films (general knowledge, not sourced). Cap to ~400–500 nit or add a heat path.
On-cell AMOLED (2.16″, 3.92″, flexible 6.5″ class)	2.16″ (Youritech ET022AM01-HT): −20…70 °C / −30…80 °C. 3.92″ and flexible rows: not published in catalogue	Emissive; heat scales with content. 2.16″ well under 1 W est. Blue pixels run hottest.	Low panel heat, but burn-in/aging is temperature-sensitive: in one large-area white OLED study +10 °C cut lifetime 1.65× (not a general rule); a 55″ WRGB display showed visible burn-in after 1000 h of static patches. Static art tiles: add pixel shifting and dimming.
HUB75 LED tiles (P2.5 / P3 64×64)	Temperature range not stated on Waveshare page unverified	≤ 20 W per module at 5 V / 4 A (Waveshare), i.e. ~780 W/m² on a 160 mm board at full white. Typical content 3–8 W est.	Sealed with no airflow, 20 W gives ~+25 K surface est., so needs a metal back, rear vents and a brightness cap. 4 A through edge pogo pins is also a contact-rating question (check the pin rating).

Venting and heat spreading

Vent the back face only (edges butt, front stays clean): bottom inlet slots + top outlet slots on a wall-mounted tile give a chimney path with no fan.
1–1.5 mm aluminium back plate or 0.1 mm graphite/copper sheet from behind the backlight to the back cover; 1 mm thermal pad from SoC/driver to that plate.
Put the SoC at the top, the battery at the bottom/coolest corner, and keep ≥ 5 mm (or a thermal barrier) between battery and backlight/SoC est. practice.
NTC on the board: derate brightness above ~55 °C interior; hard-cut the backlight at 65 °C. Budget: interior rise ≤ 25 K over a 35 °C room.

Safety

Material	Heat (Tg / HDT)	Flammability	Use for Tyler
PLA	Tg 59.7–60.5 °C measured; HDT 57 °C (Bambu), service 50–60 °C (Prusa)	Not rated (typically burns) general knowledge	No for a sealed 10″ tile (will creep at 55–60 °C)
PETG	Tg 78.7–81.2 °C; service 70–90 °C (Prusa)	Not rated; FR variants: Prusament PETG V0 UL94 V-0 (≥ 0.75 mm); Spectrum FR PETG "designed to meet" V-0, but HDT only 58–63 °C	Minimum; non-FR OK for prototypes
ASA / ABS	ASA Tg 103.7 °C (service 90–110 °C); ABS Tg 107.5 °C (95–110 °C)	Not rated; Spectrum FlameGuard ASA "designed to meet" V-0 (1.6 mm), HDT 86 °C	Good default shell
PC / PC-ABS	PC Tg 147.5 °C (service 100–115 °C)	PC inherently better; Spectrum PC/ABS FR V0: V-0 at 1.5 and 3.0 mm, HDT 90 °C (claimed, not UL-listed)	Best if you need V-0 near the battery

Li-ion/LiPo: charge only 0–45 °C; discharge −10…60 °C (some −20…60 °C); above 60 °C risk of damage. Use a protection circuit (PCM) with temperature sensing, plus a fuse/PTC, and lock out charging outside 0–45 °C.
Swelling: thickness grows 3–5 % across a charge/discharge cycle and up to 7 % after ~300 cycles (Renata). Don't use a fixed % as final: cavity ≥ max cell thickness + life growth + foam + housing tolerance + assembly clearance. Early concept: ≥ 10 % + 0.5 mm compressible foam est. No sharp points on the cell, no rigid clamping.
Spacing: battery away from the backlight and SoC, not under pogo/magnet pockets, in its own pocket with a barrier (PI film / ceramic paper) est. practice.
Ambient: design for 0–35 °C indoor (HK) est.; a tile left in a parked car or in sun will exceed the 70 °C LCD limit and the battery 60 °C limit, so state "indoor only".

Tolerances

Why realistic ≠ spec

Panel outline: LCD modules usually ±0.2 mm on unmarked dimensions (QD-TFT2803 spec, a 10.1″ module spec); cover-glass drawings ±0.1 mm.
Glass edge chipping: visible edge chips are rejected, non-visible ones under a bezel can be accepted (one inspection standard). A flush-edge design exposes the glass edge, so the reject rate and breakage risk go up. Specify "no visible chips" and add an edge guard.
FDM printing: plan ±0.5 % with a ±0.5 mm floor; ±0.2 mm in XY when calibrated. Over 190 mm, ±0.5 % ≈ ±0.95 mm, so print the pocket and wall in one job and verify with a coupon. Clearances per side: snug 0.1–0.2, sliding 0.2–0.3, loose 0.4–0.6 mm.
Magnets / butt gap: magnets pull tiles into contact, so the seam is wall-to-wall plus warp; allow ~0.3 mm for flatness and pogo spring push est.
Real video walls: LG's 0.44 mm bezel is 0.88 mm combined, but manufacturers want ≥ 0.5 mm mechanical clearance between panels, so ~1.4 mm installed est., and they use micro-adjust mounts.
My model: thin wall realistic = 2 × (1.2 wall + 0.5 clearance [outline ±0.2 + print ±0.3] + border) + 0.4 active-area offset + 0.3 butt gap = 2b + 4.1. Flush edge = 2b + 1.1. Current = 2 × (2.0 + 0.5 + max(border, 3.0)) + 0.7.

Sources and what is unverified

Unverified / estimates: border per side is derived from outline − active assuming a centred active area (only the Hot Display's L/R/top value is read from the drawing); T+B total is outline − active; ET100 backlight power; 3.92″/flexible AMOLED and LED-module temperature limits; LED edge offset; All Vision "<2.5 mm", made-in-china "<1 mm" and LED-cabinet "≤ 0.1 mm" claims are vendor marketing; the ≤1 mm threshold and all "realistic" allowances are my own model. Existing Tyler pages assume ~2 mm for ET050SQ01-GX and "near-borderless" for ET022AM01-HT; the datasheets give 1.15 mm and 0.95 mm.