Falcon F16v5 wiring schema for high-current 12V WS2815 installation — sanity check requested

[Meuretti]

---

Falcon F16v5 wiring schema for high-current 12V WS2815 installation — sanity check requested


Building a 28-panel WS2815 LED screen (8×32 pixels per panel = 7,168 pixels total, 4 rows × 7 columns) for an art installation. Controller is Falcon F16v5, using 7 outputs (one per column). Each column = 4 panels in serial data chain via JST connectors. 7-day continuous run, exhibition hours 10:00–20:00, Swiss Art Awards Basel, 16–21 June 2026.


Power architecture:

• 4× Meanwell UHP-500 (12V, 500W each) — one PSU per row of 7 panels
• ~143A total at 100% white (verified: 0.24W/LED × 256 LEDs/panel × 28 panels = 1720W)
• ~36A per PSU, ~5.1A per panel
• Each PSU has 2× V+ and 2× V- terminals — splitting each row into two strings (4+3 panels) → ~20.5A and ~15.4A per terminal
• Power and data physically separated: V+ and GND go directly from each PSU to its panels, NOT through the Falcon

Per-row distribution (no daisy-chaining):

• Each PSU feeds two local busbars (V+ and V-)
• 7 individual dedicated wires from each busbar to each panel's V+ / GND
• All 4 V- busbars connected via a vertical GND backbone → common ground

Falcon F16v5 connections:

• External Power Input: +12V and GND from common ground (controller logic only)
• V1 and V2: GND only — no +12V (per-output E-fuse rated 5A, insufficient for ~36A/row)
• Output V-pins: unused
• Output G-pins: unused (GND runs externally via PSU)
• Output D-pins: connected to first pixel's DI and BI bridged together (full backup-data redundancy from pixel #1 onwards)

Inter-panel JST connectors carry only data (DI + BI), no V/GND.


Wire gauges (DC, runs <2m, real-world average 20–40%, 100% peaks rare and brief):

• PSU terminal → 4-panel string: 2.5 mm² (max 20.5A, real ~8A)
• PSU terminal → 3-panel string: 1.5–2.5 mm² (max 15.4A, real ~6A)
• Busbar → individual panel: 1.5 mm² (5.1A)
• PSU V- → common GND backbone: 2× 4 mm² in parallel (using both V- terminals)
• Inter-panel data (DI + BI): 0.5 mm²
• Falcon External Power: 0.75 mm²
• Falcon V1/V2 GND: 1.5 mm²

Connectors:

• Common GND backbone: copper busbar or high-current terminal block (>40A per terminal)
• Per-row local busbars (V+ / V-): WAGO 221-413 (32A / 4mm² rated) — one cluster per PSU
• Branch points to individual panels: WAGO 221-413
• BI+DI bridge at first pixel: WAGO 221-412 or solder

Specific concerns I'd like feedback on:

1. Falcon V1/V2 GND-only configuration: Is V1/V2 with GND but no +12V valid on the F16v5, or does the output driver circuit require V1/V2 +12V to function even when output V-pins are unused?
2. Common-ground topology: Per-row V- busbars chained via a vertical backbone (functioning as the common ground reference). Sufficient, or should I bring all 4 PSU V- to a single physical star point instead?
3. BI+DI bridge at first pixel: Any concerns vs. the more common BI-to-GND configuration? My reasoning: full redundancy from pixel #1 onwards.
4. Wire gauges — is 2.5 mm² sufficient for the 4-panel string (20.5A peak), or should I move to 4 mm² for safety margin? Real-world load is 20–40% on average; full white is rare and brief in artistic content.
5. WAGO 221-413 (32A / 4mm² rated) as branch connectors and per-row local busbars — appropriate for a 7-day continuous install, or should I use screw terminals for the higher-current branches?
6. Falcon power source: Separate dedicated 12V supply for the Falcon, or tap from one of the panel PSUs (with appropriate fuse)? Concern: noise on the shared rail from heavy LED switching.
7. Heat dissipation on mounting surface: Panels mounted on a wooden backing board (likely MDF or birch ply). Total dissipation ~1.7 kW peak across 65×170 cm. Do I need a thermal layer (aluminium sheet, thermal pad) between panels and wood, or can panels be glued directly to wood for a 7-day run at ~20–25°C ambient?

Better safe than sorry — appreciate input from anyone who's run similar high-current installations.

 

[merryoncherry]

Any reason not to use more data ports (to allow higher frame rate)? 512 pixels per port instead of 1024, you'd still only need 14 of the 16 ports.

Output D-pins: connected to first pixel's DI and BI bridged together (full backup-data redundancy from pixel #1 onwards)

(Also your question 3) Unless it explicitly says to bridge them that, ground BI. Bridging will cause real problems with some pixels (GS8208 for sure gets confused).

Falcon power source: Separate dedicated 12V supply for the Falcon, or tap from one of the panel PSUs (with appropriate fuse)?

Whichever way you go, be sure the ground is common, so the data has a consistent reference everywhere. I'd share a PSU so you know that it is, but if you have another PSU this is fine.

Falcon V1/V2 GND-only configuration: Is V1/V2 with GND but no +12V valid on the F16v5, or does the output driver circuit require V1/V2 +12V to function even when output V-pins are unused?

Good question, I feel like I've seen the answer but forgotten it. I'd wire them to some V+, could help if you needed to test something on the ports.

 

[Skymaster]

Common-ground topology: Per-row V- busbars chained via a vertical backbone (functioning as the common ground reference). Sufficient, or should I bring all 4 PSU V- to a single physical star point instead?

The common-ground topology as you have depicted would be perfectly adequate in this use case; except I would only use one wire. Not one to both side of the Falcon.

Common GND backbone: copper busbar or high-current terminal block (>40A per terminal)

The grounds between the rows will not be carrying any pixel current (as these are local to the rows) and will only be carrying the data return (across the columns).
Because of this, you wont get any voltage drop/rise across the conductors and therefore data would be unaffected.
These only need to be smaller gauge - 18awg/0.75mm2 would be more than sufficient.

 

[Notenoughlights]

Just looking through your power calculations here

7,168 leds running full white, I've gone and used the Quinled power sheets to find this information.

60 leds per meter 300 leds per 5 meter strip, consumes about 50W at 100% RGB.

7168 leds is close to 24 full 5m lengths, so approximately 1200w with correct power injection.

That is approx. 100A at 12V DC, and around 10-11A at 110V and 5-6A at 240V on the AC side of your power supplies (efficiency of the supplies and total loading per supply will vary this)

Each power supply if split evenly for 25A per unit will need to be around 300W so your 500W supplies will handle this just fine.

I would suggest you mount the leds onto something that is heat-dissipating and not made of wood, as that will prevent the led strips from cooling correctly leading to led failures.

I would suggest against feeding the strips directly from the PSU, if something were to go wrong and you have no fuses in line to catch a short circuit, and being a public installation, fire is not something you want. I would suggest a fused distribution board per power supply if it's within your budget i.e https://quinled.info/quinled-dig-octa-power-7/

BI to GND on ws2815 at the first led as per the ws2815 datasheet. https://www.normandled.com/upload/201808/WS2815 LED Datasheet.pdf