One thing to be aware of is that some white LEDs are based on a UV LED with a white phosphor mix, and some phosphors are subject to early burnout. I have some early white LEDs that are exceptionally dim now due to the phosphor burnout.

As for the construction of the sign, US$300 for a sign of that size/complexity/number of LEDs isn’t all that bad. Blue (and white) LEDs are a bit more expensive than red LEDs, and there are about 250 blue LEDs. So, the LED parts alone may have eaten up a significant portion of that cost, with another chunk taken the by the material and machining for the box, and a bit more for the labour of wiring it up.

As for the wire, stripped network cable is a surprisingly good source for inexpensive hook up wire. I use it quite often. Plus, significant lengths of it are often surplus.

The power supply is definitely a bit wonky, though. That little voltage margin is not a good design, especially given that the mains voltage can vary +/- 10 percent or so under extreme conditions. If it were me designing it, I think I would have went with a bit more head room, even if that had decreased the efficiency a bit. Or, went with a voltage regulator chip. Or, an LED driver chip.

Nice fix on the power connector. I just hope the wall wart never goes bad and needs replacing (since the now non-standard connector will be impossible to duplicate, unless someone knows the trick). As for the switch, I do think I would have plugged the hole, just to keep dust/debris out.

As for intermittent failures, some LEDs seem to have a plating on the leads that is prone to flaking off and/or making bad contact. This was especially true with ancient LEDs. I thought that problem had mostly been fixed with recent production devices. But, without knowing the history/heritage of the particular LEDs, it’s impossible to know.

Also, I’ve thought about building segments, using a similar technique, for a really large 7 segment display. Might yet get to that.

Dave