Hello,
I am doing a lighting project and my client wants to use GS8208 LEDs. From what I’ve seen only the difference is whether or not the resistor and capacitor are in the pixel(WS) or outside on the PCB(GS).
Looking through the compatibility list, I am not seeing the GS8208 listed. Are these strips compatible, or do I need to request WS2018 strips?
Thanks in advance!
Mike
Hi Mike,
These are WS2812/WS2815 compatible but often come with default patterns that start up when they dont see data for a while. This can cause undesirable effects/glitches, but they are otherwise compatible. I don’t recommend them. I have heard that these have a variant without default patterns, but sourcing them might be a challege.
I ordered some GS8208 and have a much more favorable view now.
Summary:
I bought this GS8208 strip at 5m, 60/m, IP67. It was $33 ($.11 per pixel in IP67) with free shipping and arrived in 15 days. Here’s the Datasheet. I wired the backup signal path, labeled D2 on the strip, to GND as it says to.
This version/vendor does not have any default patterns built-in. They do, however, all flash white for a quick moment when powered on, which helps scare the 
out of you for a brief moment every time.
It’s really good - much better than the WS2812 or WS2815. It’s perceptually linear across the full range of the global brightness scale. It’s kind of like having APA102 / SK9822 dynamic range but at the lower Neopixel data speeds. The low end has rich minor gradations: Driving at 1/255 of red, green, or blue creates a tiny amount of light, so you can mix for browns and very low whites, and fine tune the color temperature of whites.
On a v2, they are very sensitive to timing. The only output settings that will work are the “Buffered (x2 rate) WS2812 / Neopixel” setting at the slowest possible timing, “1.3us / 2.6us”. This limited output to about 5,500 pixels / sec for trivial patterns (18 FPS for 300 pixels).
On a v3, the default WS2812 driver with it’s only data speed of 800 Kbps 270/630ns worked great. Trivial patterns got 25,800 pixels / sec out the door (95 FPS for 300 pixels)
Like the 12V WS2815 and most 12V LEDs I’ve used, the current usage for any one color being on is the same as when all three LEDs in a chip are on.
As a 12V strip, powering all 300 LEDs across 5 meters at full white did not result in noticeable discoloration or red shift. Vin was 12.0 V, and voltage at the end was 9.4 V.
Nice!
Looks like a very in-depth test.
Do they only flash once, when they are powered on?
Or also periodically doing patterns?
Only once, very briefly, at initial power-on.
Hi @jeff ,
Interesting!!! I wonder what interaction is going on with V2, that setting was added in direct response to the first time I’d seen those type of LEDs and was the only setting I could get them to work on. I hadn’t seen that since (until recently), and figured it was a weird variant. There must be something particular about how the data lines are being interpreted that puts the GS8208 into a low speed 400KHz mode.
I think this can happen in other circumstances too. I’ve recently heard of some issues with these, but it wasn’t easily reproducible. I’d be interested in figuring out what triggers it exactly.
Without knowing more I’m hesitant to recommend these LEDs with PB officially, mainly because it could cause mysterious behavior that might be hard to troubleshoot. Also because if you do get a variant with the built-in patterns, they are aggressive enough on timeouts that it starts to play when there is the slightest pause in data stream (like on V2, loading the interface would trigger it), and it looks glitchy as it switches back and forth. I can imagine a pattern with a long beforeRender might trigger it too.
I think the white flash could be the “test” mode, which I’ve seen referenced in datasheets. I suspect they didn’t really disable “built in patterns” but replaced the pattern with something boring like white (and hopefully increased the timeout). I’m curious what happens if you disconnect data while leaving them powered. Thankfully the power use doesn’t increase with white compared to some full brightness color, but might prevent someone from under-budgeting power as the built in test mode wouldn’t care what max brightness setting you had.
A strip I ordered just arrived.
It was from the same seller as Jeff used.
Flashes white for a short second, but otherwise no other patterns shows up.
Left them on for 30 minutes, and they all stayed off.
In my experience these have much better brightness control than WS2815.
Wondering what the chances are of these getting officially supported?
I’m about to purchase over 50 meters of 12v LED’s, and I have to say I’m tempted to get these over the WS2815…
You have my blessing to use any LEDs you like. Doubly so for any that you test and know to work.
One additional data point: I just tested 20 reels total of GS8208 from 5 different Aliexpress vendors, and while there were some connector differences, all worked pretty much the same as we’ve described here.
Thats great to hear, @jeff! Did you have to specifically hunt for the stuff without built in patterns?
I emailed all five before ordering, but none of the vendors I approached said they were the version with the built-in patterns. I suspect there’s some new ubiquitous vendor or spec for the chips themselves. All were RGB color order and had the white flash.
I am considering buying a pretty large amount of these.
Were some of them of better quality than others?
Can you recommend me a certain brand/model/seller of it? 
I can only comment on my experience with them so far, I switched from WS2815 to GS8208. Besides GS not having an RGBW option, I’m pretty happy. Better brightness control and color range. Just make sure to get the version with no built-in patterns. I’ve gone through about 20k so far, and no issues.
Sorry it took me this long to reply, @Madsmaten - I had to get back to the warehouse where all the strips are installed and turn them back on.
All five vendors shipped working products (we tested them all on the reel). One strip from LUY-LED started gitching once installed - this could be from our rough handling. All products looked nearly identical except the ones from Ray Wu used 3-pin JST-SMs (instead of 4 - both are reasonable design choices for the backup data line) and Ray’s included power injection barrel connectors on one end already soldered.
The bottom five (in green) are the ones we tried. Including shipped price, JAD-LED seems pretty good.
