GS8208 LED Strip

Not sure about that data sheet - either way it's WS2811 compatible which is 5V based signaling.

I'll have to try a ESPixelStick with it and see if the GS8208 will take 3.3V signaling too. :)
 
Does anyone know if the signalling on GS8208 needs to be close to the 12v supply?
The specification sheet seems to indicate that it does :

On Page 3 - Input Logic Voltage SDI -0.5~VDH+0.5 (where VDH is 12v)

Those figures are the maximum working range (without damage) figures. Page 4 has the VIL and VIH figures although they seem to be somewhat screwed up. I think they mean VIL (logic 0) is 1V max and VIH (logic 1) is 4V min ? Maybe there's a more complete data sheet out there.
 
Hey thanks, I went with a 74HCT08 AND gate as a logic level shifter, and it seems to work OK. Nice clean 5v output.

When I hooked up the board to the lights however, I get no LED output. When the power turns on, all the LEDs shine white for a fraction of a second, so I think they are good strips.

When I measured the output from the ESPixelStick driver (NodeMCU from Banggood) I noticed that the edge to edge speed was 746KHz - is this too much drift from the spec 800kHz ?

That's the only reason I can think of why the lights would not be working.
 
I'm having trouble getting my GS2808 pixels to work. I've tried them on a Falcon, Alphapix, and Joshua with no luck. 2811 pixels work fine on all of these, but not 8208. What is the string type and Kbps setting you use on a ECG-P12R?
 
I'm having trouble getting my GS2808 pixels to work. I've tried them on a Falcon, Alphapix, and Joshua with no luck. 2811 pixels work fine on all of these, but not 8208. What is the string type and Kbps setting you use on a ECG-P12R?

Hmm they worked first time for me on Falcon - just wired in the power, ground and 1 data line.. sorry I can't be more specific I am traveling away from my pixels at the moment.
 
Next was power consumption measurements, with the entire 5M roll (150 LEDs), powered via both ends connected to a 12V Lab Power Supply. I've done this test with Ws2811 too and other lights so I can formulate PSU estimates.

View attachment 12789
Now WTF is going on here! Notice how single R/G/B measurements are off the charts compared to 2811. Mixing colours, which on 2811 would traditionally increase current makes very little difference, Infact mixing colours reduces consumption! I don't believe this is voltage drop related, powering from one end was only marginally lower draw. What probably is at play is the internal series-wiring of the R,G,B, LED elements and their PWM drive circuitry.

Feeding 12V to each LED too won't be helping either - but it's a price to pay compared to dealing the hassle of 5V based voltage drop over longer distances. In reality, it'll mean higher average power draw compared to 2811, just not higher total/peak power draw. This may be a surprise factor for some people upgrading if their cabling gauge wasn't up to spec to begin with.

So this puts 3 single-color LEDs in series, and uses constant current drivers, thus allowing single pixels at 12v. That's a very interesting option and a new design.

Think of it this way. Peak instantaneous current is pretty much the same regardless of color (thanks to the constant current driver).
PWM modulation changes the average current tho. This would be the peak current times MAX(R,G,B) / 255, or perhaps / 256. All colors with the same maximum value would have about the same current.

Suppose RGB = 120, 80, 30. Assume 2, 3, 3 volts Vfwd
30 PWM steps = R+G+B driven in series at peak current (8v across RGB, 4v dissipated in transistors)
50 PWM steps = B shorted, R+G driven at peak current (5v across RG, 7v dissipated in transistors)
40 PWM steps = G+B shorted, R driven at peak current (2v across R, 10v dissipated in transistors)
135 or 136 PWM steps = all off

Increasing Blue to value 100 (or any value <= 120) would not change the avg current (significantly), because it's still turning on for 120 steps.
But increasing Red to value 150 would increase the average current, regardless of the other colors (so long as they are less than Red)

Thus (peak) current is flowing 120/255 of the time (or 120/256 depending on circuit), with different portions used by each colored LED and wasted by the Constant Current transistors.

This makes it less efficient than 5V pixels, but more efficient than 12v BULLET pixels (which wasted 9 to 10 volts of 12). It could wind up being slightly more efficient than 5V pixels when full on.

So it might take more POWER on average. But being 12v, one has a 2.4:1 advantage in current vs 5v so that reduces needed wire sizes or increases allowed length.

So the typical GS8208 current is spec'd at 15mA per chip; at 30 pixels/m that's 450mA/meter @ 12v = 5.4 Watts
Versus 30 WS2812b pixels per meter at 48mA (R+G+B) per chip = 1.44 A @ 5v = 7.2 Watts

However, full blue would be the same as full white for the GS8208 but just 1/3 as much power as white for the WS2811.

Note that wire gauge to handle max current is smaller, since the max current is about 1/3 as much; but the power supply would need to be capable of more total power.

This could be worthwhile, because it allows smaller gauge supply / power injection wires, and can accept more voltage drop.

Where it would really shine powerwise would be individual 12v pixels like bullets. Some people use 12v ws2811-based bullets now for the lower current + greater tolerance (albeit with greater total power). A GS8208 (or GS8206) based bullet would consume less maximum current by comparison. That is, you can run up to 3 colored diodes iin series from the same 15 mA, rather than needing to draw 15 ma/color for parallel. Think 100 RGB bullets drawing 1.5A, rather than 4.5A. (But you wouldn't be able to drop the 12v as low as with WS2811 bullet pixels, since you need to cover up to 3 Vfwd drops rather than just one).
 
After seeing Brads GS8208 strip in person I decided to look into more/further. This is an innovative design which has benefits and downsides. For strip I can see lots of positives but I'm not sure how this would go for nodes even though Ray has it available as 12V nodes. The issue with 12V nodes is that the leds would have to be special as standard common anode wouldn't work. They may actually be GS8206 in https://www.aliexpress.com/item/32876963386.html . No way of confirming which chip is in there without seeing the actual chip or doing some current measurements.
 
Has anyone tried those nodes as mentioned in Alan's last post?
I'm about to place my first ever order of bullets, and for me, it's either these or the WS2811.
 
Has anyone tried those nodes as mentioned in Alan's last post?
I'm about to place my first ever order of bullets, and for me, it's either these or the WS2811.
I have a string of them here that I hope to get tested beside a matching string of 5V and 12V 2811 nodes. I have finally finished a couple of jobs so I expect I'll get the testing done in the next few days.
 
I finally tested a ESPixelStick (at least the older firmware version I'm still running - does not work GS8208. This maybe because the ESP01 is outputting at 3V TTL not 5V.

GS8208 does also not work with J1SYS PPX extenders

And I can also reconfirm that you can't add GS8208 on the end of a WS2811 run. I wish I'd recommended that before I started hacking up my display :p
 
  • Like
Reactions: AAH
Which version of the ESPixelStick hardware did you test with?
I should clarify that I used an ESP01 module directly, wired to a 3.3V regulator & pin2 going straight to the data line on pixels. I was not level-shifting the pin2 output back to 5V either.

Running ESPixelStick version 1.3


I could try it again with a new ESP build & with a level shifter on the output - but haven't had time at the moment to try.
 
Back
Top