GS8208 12V pixel strings - A wildcard third option for the 5V vs 12V WS2811 question?

ryanschristmaslights

Senior elf
Administrator
Joined
Jun 30, 2010
Messages
951
Location
Adelaide, Australia
'5 volt versus 12 volt, which one do I choose?'

How many times have you seen someone ask the above question? More than likely they are not sure whether to buy WS2811 5V or 12V RGB pixels. Well how about a third contender, GS8208 12V pixels? Not all 12V is the same, as it turns out! Spoiler: The power draw is lower, especially on white.

In 2021 I began investigating possibly increasing the amount of GS8208 in my display, alongside existing GS8208 IC 12V addressable RGB strip that was bought in 2018 for window/door outlines and leaping arches. After using this kind of addressable RGB strip for three seasons (now four) it is still working great. The GS is because this IC is from Genesis Systech instead of the currently more commonly known WorldSemi (e.g. the WS in WS2811). Unlike most other 12V IC strips, each 5050 LED is individually controllable. Other differences?
  • The R/G/B of each individual LED is wired in series instead of parallel.
  • A built-in test mode when no data is present, just apply power (although one third-party site indicates there may be a newer/alternative IC revision with this test mode omitted).
  • Automatic 12-bit gamma correction from 8-bit data for enhanced RGB colour mixing/blending abilities.
  • A fourth wire that is used as a backup data line. Potentially helpful if you have a single pixel failure that would have otherwise prevented the data signal making it to any subsequent pixels further down the line. This should also work with multiple failures in a string, so long as the problem does not occur with consecutive pixels.
Does your pixel controller need to explicitly support GS8208? Not necessarily. This is supposed to be compatible with WS2811 data. I have successfully had a couple of WS2811-configured PixLites drive these pixels when they are directly connected to a controller output. I have not tried mixing WS2811 and GS8208 on the same output. You can read more about GS8208 generally from some past forum posts and wiki page:

This message is split across multiple posts for easier referencing.
 
Last edited:

ryanschristmaslights

Senior elf
Administrator
Joined
Jun 30, 2010
Messages
951
Location
Adelaide, Australia

Finding a supplier​

2021 was the year I introduced GS8208 12V bullet pixels into my display. These were manufactured by a new-to-me supplier. I did ask several of the more well known sources within the Christmas lighting community for buying lights from however they couldn't make them anymore or the minimum order quantity was far too high. At one point Ray Wu at Rita Lighting had GS8208 in bullet pixel form listed on AliExpress I think for around US$0.23 per pixel, but the item has since been removed. After enquiring about them I was told these strings were no longer available. ETOP LED's response was similar. ScottLED could make them, however there was a minimum order quantity (MOQ) of 5000 pixels at $US0.40 each due to needing to specially source the IC. I'm grateful for Felix (benbrown) investigating whether ScottLED could source the requested IC, however I didn't need anywhere near half of the amount required let alone the full 5000. So the hunt was on for a new supplier with a lower MOQ.

GS8208 photo of individual bullet pixel both sides of pcbSearching Alibaba, GS8208 pixel strings in either a bullet or square pixel form factor were advertised by a small number of potential suppliers. None of them I'd heard of before now. I enquired with one of these suppliers who had advertised US$0.19 per pixel, with IC in stock and a low minimum order quantity of only 50 pixels (1 string). It turned out that the actual per-pixel price would be almost double, apparently due to increased supply costs for the GS8208 IC (just as has happened to many other electrical components). Initially this supplier said US$0.40 (the same as ScottLED but with much lower MOQ), however I negotiated this down to US$0.35. It was the last week of August at this point so I decided I would order with this supplier and not look any further.

Who was the supplier I settled on, you ask? Shenzhen Shiji Lighting Co., Ltd. I asked them for a quote for 50x GS8208 pixels per string with black insulation and 10cm spacing edge to edge. After advising their sales rep I'd be installing the pixels outdoors they advised that they can manufacture with 'waterproof' 4-pin connectors and suggested making strings with 2-pin power injection connectors at the first pixel. The photo below is from another customer's order. This seemed like a neater solution than using T pieces if additional power was needed, so I accepted their recommendation.

GS8208 bullet pixel string example with connectors
 

ryanschristmaslights

Senior elf
Administrator
Joined
Jun 30, 2010
Messages
951
Location
Adelaide, Australia

Ordering the pixels​

I placed an order with Shiji Lighting for 16x strings of 50 12V GS8208 pixels with 10cm spacing edge-to-edge (default is 8cm), black insulation and the supplier's own 4-pin 'waterproof' connectors. Each string also has a 2-pin connector at the first pixel for optionally injecting power (I've not needed to do this). Also added to the order were some 4-pin and 2-pin pigtails, a few premade 4-pin extension leads and end connectors for sealing the output connector of the last string. The terms were 50% deposit with 15 day leadtime and the remainder due once the order was ready to ship. The sales rep supplied production photos at a few points along the way, such as when the epoxy waterproofing was drying and the completed strings ready to ship. The total time between ordering and receiving was 4 weeks. The courier was DHL.
  1. 06 Sep 2021 Order placed with initial 50% deposit payment.
  2. 23 Sep 2021 Strings have been manufactured and epoxy is in the process of drying
  3. 24 Sep 2021 Order ready to ship. Paid the remaining 50% balance.
  4. 24 Sep 2021 Shipment picked up - SHENZHEN, CHINA, PEOPLES REPUBLIC.
  5. 04 Oct 2021 Shipment delivered - ADELAIDE, AUSTRALIA.
GS8208 bullet pixels manufactured with epoxy drying GS8208 one string of bullet pixels (50 per string)
 

ryanschristmaslights

Senior elf
Administrator
Joined
Jun 30, 2010
Messages
951
Location
Adelaide, Australia

The pixels have arrived ... a closer look​

They're here, they're here! Here are a couple of close up photos. The red wire in the second photo is from the 2-pin optional injection connector.

GS8208-Shiji-Order---close-up-of-pixel-(1).jpg GS8208-Shiji-Order---close-up-of-pixel-(2).jpg

The 4-pin connector is slightly smaller than Ray's 13.5mm one that I use in my display (see third image, below left). Also there are some cosmetic issues with about four of the pixels across few strings where the LED doesn't stick out as far. In practice though it didn't really matter for where I planned to use them as I just pushed those pixels further through the coro to almost make up the difference. There is less of a buffer between the exposed wires and end of the epoxy, though it looks like there is just enough to still do the job. But 4 out of 800 isn't too bad, right?

Shiji 4-pin connector side-by-side with Ray Wu's 13.5mm 3-pin connector. A few LEDs are not properly extended. Out of the 800 ordered, I think this applied to around 4.
 

ryanschristmaslights

Senior elf
Administrator
Joined
Jun 30, 2010
Messages
951
Location
Adelaide, Australia

What's the power draw like compared to 12V WS2811 bullets?​


Ok, ok you made it this far. A key advantage for GS8208 is its reduced power requirements, compared to other 12V IC bullets. For the sort of brightness that looks good once it is dark, I opted to limit the GS8208 brightness to either 40% or 60% depending on where the lights were destined. Below are how things went with 100x pixels connected to power and cycled through White, Red, Green and Blue. Just like @BradsXmasLights's findings with GS8208 in strip form, the power draw on white versus R/G/B is not what you might have expected. For comparison I measured a string of 105x WS2811 12V icicles, too (also bullets). Yeah, yeah, I know... 100 vs 105 is not exactly fair but oh well 🤷‍♂️. These measurements are at the controller output side.

BrightnessAll On WhiteAll On RedAll On GreenAll On Blue
100x 12V GS8208 40%3.83W3.77W3.77W3.78W
100x 12V GS8208 60%6.63W6.62W6.63W6.65W
105x 12V WS2811 40%20.63W9.58W9.60W9.60W
105x 12V WS2811 60%30.08W12.65W12.68W12.68W

There was little difference in power draw for White vs individual R/G/B with the GS8208 - slightly more for white at 40% but near enough to identical at 60%. For the sake of this post I also measured the GS8208 at 100%:

BrightnessAll On WhiteAll On RedAll On GreenAll On Blue
100x 12V GS8208 100%12.19W16.56W16.55W16.61W

At 100%, white actually draws less power than single colours. 100x GS8208 full brightness (100%) white pulls less than half the amount of watts that WS2811 does on white at 60% brightness.
 

ryanschristmaslights

Senior elf
Administrator
Joined
Jun 30, 2010
Messages
951
Location
Adelaide, Australia

Miscellaneous thoughts/final comments​


A total of 700x GS8208 12V bullets were added to the 2021 display (Split into strings of 200, 200, 150) + 100x spares. With around 30-35 metres of existing GS8208 12V strip added in previous years, you might say that's a lot of GS8208! o_O

As far as test mode? It doesn't really bother me. I power the pixels via the controller's fused output. Shortly after the time of applying power to the controller, there is a momentary flash of white and a couple seconds later a red flash. A further few of seconds and the pixels begin responding to the incoming data stream from the sequencer or show player. I timed it once and it seemed to take around 5-6 seconds from power to outputting a playing sequence - keep in mind this is with an Advatek PixLite 4 MK1. With the PixLite powered but no sequences playing, this does seem to prevent the test mode from activating, at least from what I saw.
 

ryanschristmaslights

Senior elf
Administrator
Joined
Jun 30, 2010
Messages
951
Location
Adelaide, Australia
Where in my display did the GS8208 strings end up? The five coro props (thanks @Troy ELD) along the left side fence, the zig-zag lines on the roof (Pixel mounting strips - also from Troy ELD) and the three front porch poles (zip-tied to gutter guard that is then wrapped around the poles and secured with wire) with the leaping arches out of GS8208 LED strip sitting in-between the poles. See if you can spot the GS8208 in the below video starting from 1 minute, 17 seconds:

View: https://www.youtube.com/watch?v=CKzmWFm-ztI&t=1m17s


So who out there in 12V land is tempted to grab a set? 😎
 

i13

Dedicated elf
Joined
Jul 5, 2013
Messages
1,172
Thanks for posting about these. Am I correct that, for example, a GS8208 pixel running a single colour is about the same brightness as a GS8208 pixel running white? I'm thinking that what you're describing is a type of pixel that lowers the brightness of its R, G and B so that when more than one colour is on, it outputs the same total brightness as when a single colour is on. This is why the current draw is approximately the same. Is this correct?

WS2811 pixels are brighter on white than on a single colour because they allow you to turn the R, G and B on to 100% brightness at the same time.
 

TerryK

Retired Elf
Joined
Feb 9, 2020
Messages
655
Location
West Central Ohio
... Am I correct that, for example, a GS8208 pixel running a single colour is about the same brightness as a GS8208 pixel running white? I'm thinking that what you're describing is a type of pixel that lowers the brightness of its R, G and B so that when more than one colour is on, it outputs the same total brightness as when a single colour is on. This is why the current draw is approximately the same. Is this correct?
...
I do not think that the GS8208 maintains a constant brightness across its colors. Rather its PWM generators are only trying to create a linear relationship to a colors command value and the current in that colors LED where of course 100% (binary 255 command) is the full LED current. A human eye's light intensity response is not linear; logarithmic I think. And to increase the complexity, the eye is more sensitive to some colors than others. I also do not think LED lumen output is linear with current.

The links to the chip specs and previous Forum Threads detail why the current per pixel does not sum with more than one color turned on as in a WS2811 for example. A GS8208 works very similar to the WS2815 except that the WS2815 is a strip only while the GS8208 can be incorporated into a bullet pixel.

Without all the detail, the GS8208 RGB LEDS are in series while a WS2811's are in parallel. This means the LEDs used in a WS2811 and GS8208 must be different. That is a WS2811 LED would have the RGB and a Common Anode connections. A GS8208 would (going by the the GS spec sheet) a +R, -R, -G, and -B. Still 4 pins but definitely different internal connections in the LED itself.

As a side note: I really like the WS2815s mostly due to the lessor amperage draw per pixel. I really wish GS8208s would be available from US vendors as bullet pixels. I would immediately move away from the WS2811s.
 
Last edited:

i13

Dedicated elf
Joined
Jul 5, 2013
Messages
1,172
I must have had a slight brain fade. If they're in series then the same current will flow through all three colours when they're all turned on. This will be the same amount of current but more total brightness than when only one colour is turned on.
 

Cranzy

Full time elf
Joined
Dec 24, 2018
Messages
264
Location
Perth
I have some GS8208 strip which i use for 2 of my spiral trees. The colours are fine.
Thing is, I also have some WS2815 strip for my other 2 trees, and if i put all 4 next to each other with both 30% brightness, the WS2815 are noticably brighter than the GS8208.
Another thing to note is I also had to go from the controller to the GS8208 then to WS2815/WS2811 as it doesn't work going from Controller -> WS2811/WS2815 -> GS8208.
 

ryanschristmaslights

Senior elf
Administrator
Joined
Jun 30, 2010
Messages
951
Location
Adelaide, Australia
Am I correct that, for example, a GS8208 pixel running a single colour is about the same brightness as a GS8208 pixel running white?
GS8208's Green and white look the same to me whereas red appears a little duller. Maybe that's just my eyes, though, as I see the exact same thing with WS2811 12V icicles.

I also have some WS2815 strip for my other 2 trees, and if i put all 4 next to each other with both 30% brightness, the WS2815 are noticably brighter than the GS8208.

Could this perhaps be the gamma correction at play? Maybe try a slightly higher intensity limit or try fade up/down to see if whether GS8208 has more linear changes in brightness vs WS2815 (or vice versa).

@i13 @Cranzy perhaps the following unpolished short video is of interest. It demonstrates the momentary flickers of the built-in test pattern at powering up through a PixLite 4 MK1 with the intensity limited to 45% and then 100%. My camera only picked up on the white flash at 100% (also please excuse the unsteady one handed handheld clip here - I cut this short due to too much wobbling). After that is a side-by-side comparison of GS8208 vs WS2811 at 45% (camera exposure locked). The white temperature looks a little cooler on the GS8208 but that might just be because the WS2811 they're being compared to are a few seasons old, not sure.


View: https://www.youtube.com/watch?v=JzdDf8rCmCI


I really wish GS8208s would be available from US vendors as bullet pixels. I would immediately move away from the WS2811s.

Perhaps if the word gets out and enough people are interested? 😉
 

TerryK

Retired Elf
Joined
Feb 9, 2020
Messages
655
Location
West Central Ohio
Yes, the WS2815s do seem brighter than the WS2811s. I've always attributed it to the diffusing or lack thereof; at least the WS2815's silicon protection does not seem to diffuse much. Have not had any problem driving them, controller type/firmware perhaps. I've been using a Falcon F4V3 to date.
 

MichaelF5

Apprentice elf
Joined
Nov 23, 2016
Messages
76
Location
West Hoxton
one of the selling points of these seems to be being able to do longer runs without injection.. anyone tested the limits of that?

anything with a faster Serial frequency on the way?
 
Top