WS2812B LED Clones: Work Better Than Originals!

Commodity electronics manufacturing is a tough game. If you come out with a world-beating product, like WorldSemi did with the WS2812B addressable RGB LED “pixel”, you can be pretty sure that you’re going to be cloned in fairly short order. And we’re all used to horror stories of being sold clones instead of what was ordered. But what if the clones were actually an improvement?

[Gonazar] bought some strips of “WS2812” LEDs and prototyped a project. When stepping up to larger production, he thought he’d go directly to WorldSemi. Long story short, the cheaper LED modules that he’d previously bought weren’t from WorldSemi, but were actually SK6812 clones labelled as WS2812Bs. When he switched to the real thing, he discovered that they had some temperature and pressure sensitivities that the clones didn’t. The clones were better!

They weren’t even straight clones. It turns out that they have a much higher PWM frequency, resulting in less flicker at low brightnesses. The distributor came clean, saying that they swapped them out without note because they spoke the same protocol, but were a strict improvement.

And this isn’t the first time this has happened. [cpldcpu] documented the differences between the SK9822 and its original: the APA102. Again, the SK version clone is superior. In this case, it implements a data-read-and-latch behavior that’s in the APA102’s datasheet, but that somehow didn’t make it into the APA102’s silicon.

We usually assume that clones are of worse quality than the originals, but of course, this doesn’t have to be the case. The clones had the virtue of hindsight — they could see what was wrong with the original and improve on it. It’s an irony of brand loyalty (or lazy parts sourcing) that we continue to insist on a known part number to the point that a distributor would lie to us in order to substitute in an improvement.

So let this serve as a reminder: all that matters is the blink. Keep your eyes open.

Thanks [Jarrett] for the tip!

56 thoughts on “WS2812B LED Clones: Work Better Than Originals!

  1. I had a design that could have used these LEDs recently. I just couldn’t bring myself to put a part that had Adafruit listed as a supplier on a piece of serious industrial equipment. I wasn’t sure how I was going to explain that to my manager if/when the supply ran dry in a few years. Even if it did come from digikey or arrow.

    Sad that no one like Cree or Osram has come up with anything similar.

    1. Plus a million to that . I wouldn’t risk designing these into a permanent installation, but I’d happily pay 50p a pop for something similar from a reputable manufacturer. TI have a control chip with similar 1-wire interface but much better functionality (TLC5973) – combine this with die from Cree,Osram,Nichia etc, and they;d have a winner.

      1. I see your “plus a million” and raise you “plus a bilion”. This exact statement is ridiculously true! For a professional/industrial product, having a reputable company offer such a solution would open up many doors for reputable products with more state of the art features without the state of the art cost. Although I am very happy with every purchase I have ever made at Adafruit, I can’t list them as a supplier for a part on a volume industrial product.

          1. Because it’s still a product from an obscure company, designed for toy applications, with a not-very-complete Chinglish datasheet, with no guarantee of future availability or product stability. Will they send a PCN or last time buy notice if they decide to change it? seems unlikely.

          2. Well we used them in commercial permanent installations many times, though we don’t usually need to offer replacements for the installation for more than 5-6 years. In the past they have only released fully compatible products and even their oldest released chip variants the WS2801 are still in production.

            I never had issues when it comes to getting really old ICs from China, some factory is always making them as long as there is any demand what so ever.

          3. Are they willing to sign a 7year guaranteed supply contract, and have at least 2 dislocated manufacturing facilities in case of natural disaster etc.

  2. The chinese arduino UNO’s with the CH340 usb serial chip are better in many ways than the original that has the Atmega16U for the usb stability. I think that’s what S’s comment above about FTDI is about as well.

    1. CH340 is not a clone of anything AFAIK.
      On the other hand, counterfeit of cp3102 are real killers, meaning it really, output is not 3.3v like cp3102 is but more around 4/4.4v.
      Ok for arduino use, but I am lucky to have verified it before hooking it to my 3.3v only uC…

      1. No kidding mac, and it’s certainly not a clone of an Atmega16U either. The cloneliness I was referring to was the board itself being an Arduino clone. Just for kicks, right now I plugged in one of my chinese UNO’s and lsusb shows “QinHeng Electronics HL-340 USB-Serial adapter”. So it’s not even a CH340, perhaps it’s a clone of that?! These boards are a fantastic value for the price and have a number of features (in addition to the USB-Serial chip) that make them actually better than the original Arduino UNO.

    2. The main problem with the CH340 is the Windows and Linux drivers – they both suck and they both do it differently! Otherwise it’s a decent chip in my experience. Just remember to remove the DTR AC coupler cap.

      1. I received my CH340-modules yesterday, bought on, 5 pieces for £3.39 (no link because out of stock now). I needed this chip because you can switch voltage between 3.3 and 5V. Altough i did not yet check the voltages the modules seems to work fine. I put two on the PC (with a wire TXD1->RXD2) and was able to put 1MB of data trough them (multiple times) at 400kBaud without any problem, the received data is exact the same than the transmitted data (read/write from/to file + MD5). I used C + Windows API to interface to the serial port. When i increase the baudrate some bytes get lost, but i’m not sure if this is a problem with the IC or the driver or my crappy PC. I didn’t experience any hangup or received garbage, just some lost bytes. Also it seems (to verify) that you can set any arbitrary baudrate, this would be REALLY usefull for µC like the AVR where the UART doesn’t have a fancy PLL or something and you have to mess with baudrate-cristals and so on to get a standard baudrate.
        I downloaded the drivers (for Win XP) on the site of the manufacturer. It’s a bit difficult because i don’t understand chinese but the filenames are english so it’s ok, at this price you can’t complain about such minor problems…

      2. I’ve had no problem with Arduinos using the CH340. Apart from the initial one of expecting it to be FTDI, and having to find the drivers. No hint about that when I ordered, and of course there wasn’t any datasheet or information with it. Drivers installed like any other (actually it’s a bit unusual having to manually install one nowadays), and it shows up as a serial port. No idea about DTR caps, the thing’s just part of the Arduino board. It works fine.

      3. The knock-off Arduino Mega that came with my 3D printer uses a CH340. I plugged it in and Windows recognized it immediately and already had drivers for it. I don’t see any problem.

        It’s remains to be seen how well it works with Linux. After I finish an upgrade, I’ll be putting my printer on a Pi with Octoprint.

    3. Don’t listen to this fool, they don’t know about HoodLoader2 which lets you use the 8u2 or 16u2 (possibly as a USB-HID, MIDI, or even FastLed interface.) Its like Xzibit herd you like arduino, so he put an arduino on your arduino… who’d want a basic ch340 or cp2102 instead?
      -compatible with HID library:
      -and FastLed:

      There’s also specialty projects like:

    1. To me clones are copies of a product. In the early PC days, IBM clones made PCs affordable for the rest of us. I purchased a Laser Compact XT with a single floppy, 512K and 10 mHz in turbo mode for a paltry $1200. I later got a surplus Leading Edge 386SX, and a AST 286 (the AST was every bit as good as an IBM in my opinion). The GW Basic manual was worth the price alone (it was given to me!).

      On the other side, a counterfeit tries to pass itself as as the original. Like cheap “Rolax” watches sold on the streets of NYC. Sometime they make a simple misspelling of the name, but the point is to deceive. I knew my PC cones were clones, not IBMs.

      However, in the spirit of this discussion, most clones are junk. Even if they don’t pass themselves off as OEM parts.

      1. You hit the nail right on the head there Steven-X.

        A manufacturer that makes a clone isn’t trying to pass itself off as the real thing.

        An individual buying what they know to be a clone will make a purchase judgement based on that information. Something that’s not always possible when it comes to counterfeits.

        I imagine it boils down to the intent of the seller.

  3. While it would not be bad for the distributor to suggest a part change and explain the reason, it is very bad for the distributor to change some part because *they* think it is better and not tell the customer.
    If I order some part, I want to receive that part. Not a fake, not a clone, and not something other people think I should use because they think it is better. If a distributor pulls that one on me, either they lose a customer, or at the very least they will have to give me a substantial discount ( say, 80% ) in my next order..

    1. That’s one of my biggest complaints about ordering from AliExpress. (Just for hobby stuff, don’t shoot me.) Vendors care so little about part numbers that they only really use them as searchable keywords. What you’ll actually get is usually more or less approximately what that part number originally signified, and thanks for your order and good luck!

      It’s not malice, just a very different (fast and loose) approach to manufacturing and distribution. If you want stable supplies, it’s up to you to find the right people and establish relationships with them. That’s fine if you live there, I guess, but ordering from Digikey et al is so much easier.

      1. Amazon is fast approaching that as well. I spent weeks hunting down a reliable source for Pink PS3 controllers. I had no way of figuring out which sellers were selling fake PS3 controllers and which were counterfeit. In the end, I finally located the controller at a local brick and mortar store. Amazon lost out.

        When I needed some decent USB cables, not the shitty $1 specials, I hit up Amazon and again I was faced with negative reviews regarding crap cables. I took a chance and got the bad cables exactly as described.

        Amazon lumps reviews for the “same” product together regardless of who the 3rd party seller is. With no way to distinguish bad from good sellers, the bad sellers just keep right at it. Most of which, I strongly suspect, get their garbage right off of AliExpress.

  4. I don’t know if you can really call these clones when talking about products which both are engineered and manufactured in china. They have a different attitude towards IP in China, I would say they are products on their own, compatible to the same software protocol as the product of another manufacturer. If they would use the same name for the LEDs it would be a different story but they use their own brand as do all the other “clones” the SKxxxx aren’t the only leds/ICs compatible with the WS281x and APA102 protocol.

  5. “…the SK version clone is superior. In this case, it implements a data-read-and-latch behavior that’s in the APA102’s datasheet, but that somehow didn’t make it into the APA102’s silicon.”

    I’ve run into that with other things. Years ago I had an Epson LX-180 dot matrix printer and a Star NX-Multitype dot matrix printer, which had an LX-180 emulation mode.

    When attempting to use some functions on the Epson, they did not work as described in the manual, but with the same software the Star worked precisely how it was supposed to. Apparently Star used the programming information provided by Epson, that Epson failed to correctly implement in their own product.

  6. Isn’t this the case with the GD32 clones of the stm32F1s???…This seems to be a trend. Clone an established product….add some improvements to it without documenting them and cut the price. Interesting model…..

    I still prefer the good’ole approach of having proper and unique documentation for any piece of silicon that you produce…even if it is a clone that is ‘better’ than the original.

    make parts sourcing great again

      1. Similar thing happened with the Hitachi HD6309. A licensed clone of the Motorola 6809, but with enhancements added (new registers, new instructions). Added features were not officially documented, I think because the licence was solely for the 6809 architecture.

  7. A few years ago, we tried to make a POV bike light with WS2812s, and they couldn’t update fast enough due to the 400 kbps data transmission rate. It’s not clear to me from the SK6812 datasheet, but it looks like their spec of 800 kHz comes out to be the same 400 kbps, so it would probably work no better for fast updates. Oh well.

    1. The data rate isn’t different, but the PWM frequency the chip is sending the diodes is double.
      The update speed won’t change but you’ll get less flicker. I guess it depends on how fast the display is moving.

  8. It’s not brand loyalty, it’s the fact there’s so much shit about, with Chinese sellers digging stuff out of dustbins round the back of factories, and selling them as genuine items. The only hope you have is to get a known manufacturer, and even then you have to be lucky.

    The guy who comes up with a solution to all this will become incredibly rich. That is if Beijing doesn’t have him assassinated. Seems like selling fake garbage on Ebay is contributing a large part of China’s GDP.

    It’s like the sort of black market you often get under totalitarianism, turned into a global billion-dollar business!

  9. “The distributor came clean, saying that they swapped them out without note because they spoke the same protocol, but were a strict improvement.”

    This is not “clean” at all. This attitude of swapping stuff without letting the customer know is part of the messy supply chain that most DIYers are victims of. It starts with swapping products with “better clones” then just “clones” then just “meh works almost the same” then refurbished products and so on.

    If you say you are selling a WS2812 from Worldsemi, then sell a WS2812 from Worldsemi, or sell a SK6812 and write it’s fully compatible with the WS2812.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.