Tinyland · maker log

Glue you can see in UV

A 3D-printer bed adhesive that is stronger than Frank’s “Instagoo,” loaded with strontium-aluminate phosphor so a 365 nm flash shows exactly where you laid it down — and a Klipper macro that refuses to print until the bed is covered.

The starting point: Frank’s Instagoo

Frank’s recipe is 21 g PVP-K30 + 9 g PVA in 200 ml of ~50% isopropyl alcohol — a 70:30 PVP:PVA solution at roughly 15% solids. It’s the same family as the commercial brush-ons (Vision Miner Nano, whose SDS discloses only an IPA carrier and a trade-secret polymer; and Luke’s Laboratory, sold as Standard/Double/Triple “strength,” i.e. the same liquid at different solids). None publishes a quantified bond strength. We keep Frank’s structure and turn three honest knobs.

The recipe

Everything by weight, for a 0.01 g scale. The default below is the 4 oz batch.

Why it’s stronger than Frank’s

• PVP-K90 instead of K30. K90 has ~30× the molecular weight (~1.3M vs ~40k), which raises viscosity, film cohesion, and tack — the single biggest lever, and a drop-in swap for Frank’s K30.
• PEG-400 tackifier. Pressure-sensitive tack in PVP/PEG blends is real and peaks near 36 wt% PEG. We start conservative (20% of the PVP+PEG fraction) so the film still releases; nudge it up for more grab.
• A trace boric-acid crosslink. ~0.7% of the PVA mass lightly bonds the PVA for cohesive strength while keeping the film water/alcohol-redispersible — so prints still pop off on cooling. (More than ~1% and it stops releasing; this is the one number the research had to correct downward.)

The phosphor (and the water trap)

Strontium aluminate SrAl₂O₄:Eu,Dy is the long-afterglow green glow pigment: excited across UV-A (peak ~365 nm), it emits a broad green band at ~520 nm. That is the whole sensing trick — a glued patch lights up green under UV; bare bed stays dark. It is non-toxic with no GHS hazard class, but it’s a hard mineral dust: mask and gloves when weighing powder.

The trap: bare strontium aluminate hydrolyzes in water to non-luminescent Al(OH)₃/Sr(OH)₂ and slowly stops glowing. You can’t just go anhydrous — PVA needs water to dissolve. The fix is a silica/fluoride-coated (“waterproof”/encapsulated) grade (Techno Glow / GLO Effex sell 35–50 µm encapsulated powder) plus a modest water fraction. Coated phosphor is mandatory here, not optional.

Mixing it

1. Pre-make a 1% boric stock: 1 g boric acid in 99 g warm distilled water.
2. Dissolve the PVA in the warm-water portion (~45–50 °C) until clear.
3. Separately dissolve PVP-K90 + PEG-400 in the ethanol (K90 is slow — give it time).
4. Combine the alcohol phase into the cooled PVA/water phase; add remaining free water.
5. Stir in the boric stock last, slowly, to avoid local gel lumps.
6. Disperse the coated phosphor just before use — it’s dense and settles. Apply thin, even coats.

In a pinch: a simpler PVP-40 derivation

The recipe above wants PVP-K90, an encapsulated phosphor, and a boric-acid crosslink — specialized stock. While that’s in the mail, here’s a derivation that uses only common, on-hand chemicals (PVP-40, PVA lab powder, 91% IPA, distilled water, a heavy PEG/PEO powder, and plain strontium aluminate). It’s meant for an automated applicator on less mission-critical printers, and it scales to 1/2/4 oz mini batches.

It still beats Frank’s: same PVP/PVA backbone, but with a touch more adhesive solids and a PEG/PEO tackifier — the documented bond-strength lever — instead of the K90 + crosslink combo. Trade-offs: the plain (uncoated) phosphor hydrolyzes in the aqueous carrier over time, so mix small and use fresh; and heavy PEO is stringy, so keep it low or your applicator will cobweb. Both glues read identically under UV.

Seeing coverage: the sensor

Flood the bed with 365 nm UV and read the ~520 nm green that comes back. The AS7341’s F4 (515 nm) and F5 (555 nm) channels straddle the phosphor peak, and its on-chip interference filters already reject 365 nm — so for a single-point read you often don’t need a separate glass filter. Take a bare-bed baseline once, then express coverage as the green rise above it. Pulse the UV only during the read (eye safety + repeatable charge). For a one-shot whole-bed heatmap, image it instead with a mono global-shutter camera behind a ~500 nm longpass.

Core BOM

Optional upgrades

The Klipper gate

PRINT_START calls a host script (via the gcode_shell_command extension) that reads the AS7341 and writes the result back through Moonraker’s SAVE_VARIABLE. A second macro then reads that value and aborts with action_raise_error if coverage is below threshold — before any heating or motion. Two-point calibration (bare → 0%, glued → 100%) is required.

[gcode_macro _COVERAGE_GATE]
gcode:
    {% set min_cov = params.MIN_COVERAGE|default(70)|float %}
    {% set cov = printer.save_variables.variables.coverage_pct|default(-1)|float %}
    {% if cov < min_cov %}
        { action_raise_error("COVERAGE GATE: %.1f%% < %.1f%%. Re-glue and restart." % (cov, min_cov)) }
    {% endif %}

Full, usable files: coverage_gate.cfg · coverage_gate.py

Safety

• Flammable carrier (~55% ethanol/IPA). Mix and store away from flame/sparks; let the alcohol flash off before the bed gets hot.
• Phosphor dust: N95/P2 (P100 better) + eye protection when weighing powder; no GHS hazard, but it’s an inhalation/eye irritant.
• Boric acid is a reproductive hazard if ingested — gloves, keep away from food/children/pets.
• 365 nm UV-A is an eye/skin hazard at multi-watt levels — enclose it and pulse only during the read.

Sources

Frank’s Instagoo · DrGhetto/Hackaday Super Goop · PVP K30↔K90 (ULI Polymer) · PVP/PEG tack (Feldstein et al.) · boric-acid/PVA crosslink (IOP) · strontium-aluminate hydrolysis + encapsulation (ScienceDirect; Wikipedia) · AS7341 (AMS/Adafruit) · Klipper Command Templates + gcode_shell_command (KIAUH). Full citation list in the repo README.