Chinese 7-segment RGB LED clock problem

[musicant]

After watching Big Clive's build of a clock a month or three ago, I ordered the fancier version from AliXP (SMT, more matrix display modules), as well as a 7-seg version using discrete RGB 5050 LEDs to light the segments.

https://www.aliexpress.com/item/Large-Inch-Rainbow-Color-Digital-Tube-DS3231-Clock-DIY-kit-with-customizable-colors-Electronic-kit-Gift/32916987930.html

The matrix kit arrived first, and I had no problem getting it going. The RGB kit I just completed powered up, but no blue LEDs were turning on.

After redoing a few suspect µC joints, no joy, so I dug deeper. I discovered the 3 LEDs in the 6-pin package are all facing the same way (anodess on one side, cathodes, the other), but the PCB is laid out such that two of the LEDs are in parallel (anodes connected together, ditto cathodes)!!!

This would allow some sort of Charlieplex I assume...IF those two LEDs were facing opposite directions, and my assumption was that the vendor had supplied the wrong LEDs. However, I can't find such "alternating" LEDs anywhere, Ali, Digikey, ...ok...that's all I checked.

Any opinions? Do such LEDs exist (I thought they did)? Is this just another unscrupulous vendor disposing of defective kit?

[tooki]

Did you figure it out? That kit looks neat, I wouldn’t mind getting one, assuming it’s not crap.

How did you determine that the LEDs are wired in parallel? I find it hard to believe that this was deliberate, since it couldn’t work properly. Is it possible that they got bridged during soldering?

Do you have the schematic? Could you share it, if so?

[Psi]

If i had to make up a story to fit this, it would go something like.

The kit was designed to use up a large quantity of reject LEDs that someone got for free. They were rejects because the blue die was bonded around the wrong way.

The kit manufacturer has now run out of these reject LEDs and so now must use LEDs that don't have this fault.

They saw this coming and designed a new PCB to support the non-reject LEDs but somehow your kit got mixed up and the wrong combination of PCB version vs LED type got put together into a kit.

Or, it could be someone else in china copied the kit design and wasn't aware it was designed to use reject LEDs.

[Psi]

China has a massive manufacturing industry and this means lots of rejects parts. They're created when people make mistakes or when production machinery is being setup/tuned and test-runs are being done.
There's a big industry in china making things out of reject parts because they're so cheap and therefor the profit is high.

[tooki]

The PCB connections were easy to check of course, and I pulled up one LED to verify its pinout. Schematic? Ha. No. Damn kit didn't even come with instructions. The vendor finally sent a doc after much cajoling.

You mean this one? http://files.banggood.com/2018/07/1323917.doc (It took 2 mins of googling to find it.)

Good story theories. My best theory is that the PCB simply used a wrong footprint, but after making 100 boards, they figured they'd sell'em anyways.

I think it's a lot more likely that there are solder bridges somewhere, connecting some pins together that shouldn't be, making the PCB appear to be wrong.

Hadn't even considered the "reject LED" theory...but then, I thought that the alternate pinout was a thing. I'm still looking for such a beast.

I don't think there'd be any compelling reason to manufacture SMD LEDs with a strange pinout, since one can achieve the same effect in the PCB layout, and making an alternate pinout would cost more (since it would forgo economies of scale). And I doubt one would end up with such a pinout by mistake.

[Psi]

i've seen many LEDs sold cheap that have the wrong pinout to what they should.
5mm leds with the long leg negative or SMT leds where the datasheet says green dot is + but it's actually -
Also seen COB leds where the LED was soldered onto the Aluminium backing pcb around the wrong way, so silkscreen says + but its actually -

China makes so many led so fast that on occasions someone hits the 'go' button on 10000 leds without double checking the polarity on the work order.