No signal > 680 pixels, fine < 650 pixels?

Hi everyone, having an issue with a new-ish WLED setup.

WLED: 0.13.2
Board: ESP32 WROOM - Amazon AZDelivery 3 x ESP32-DevKitC

Power: 2 x 200W 12v Meanwell PSU
LEDs: Some type I’ve not seen commonly before- 12v, individually addressable, backup channel type tape. - WS2811 compatible ( ‘LC8808/UCS1903’ it claims - 12V Pro Bright Digitally Addressable LED Tape | UltraLEDs )

Issue:

There are 900 LEDs in the run.(it’s a run of about 50, a short break (maybe 50cm),the next 800, another 50cm, then the last 50). If I select 650 LEDs in settings, the tape will partially light up. If I select 670 or higher then there will be no output whatsoever from LED and the whole strip is unlit.

Connected from Pin D16. 12v → 5v buck convertor to power ESP32 and SN74AHCT125N level shifter. 4-5m for data line until first pixel.

I’ve had a brief dabbling of WLED in the past. It used to work fine about 2 years ago in the same setup until my ESP board blew (so an earlier version of WLED, I didn’t have a level shifter then either and I had a few flickering problems), however, I didn’t have time to fix it, so I switched to an amazon SP105E type ‘magic led’ controller. https://www.amazon.co.uk/SP105E-Controller-Supports-Controllers-General/dp/B07V9D9GZV. This works fine now, and I know all the LEDs can be powered successfully with no issue, and the hardware itself is wired ok and the power supplies adequate. This has been in place for the last 2 years. Now I’ve had time to re-install WLED, ESP32, and level shifter.

If I select 670 pixels or greater, the data signal just cuts completely. The weird thing is if I increase the number of LEDs, (from the default of 30) I see the line growing and growing, but it gets darker and darker as the line grows - could this be a setting that needs to be changed, like maximum current?

Question 2, slightly off topic - I used to have the same setup above, and I would put a data line in parallel with another set of LEDs (to mirror the data line) - there was no level shifting setup at the time, however it would flicker and glitch all the time - is this because the voltage would be dropping too low when the data line was duplicated in parallel? Many thanks so much for your help, I find this so interesting and fascinating!

The controller setup is at work so I can’t check, but having a search/think about it, one guess is that if the automatic limiter was set to 800ma (or whatever the default is) would that mean that more than a certain number of pixels would dilute / lower the ma per pixel to be lower than the minimum threshold to light? Next time I’m at work I can turn off the current limiter?

I currently have 8 or so power injection points each fused at 7.5a, and it’s all running a bit of an OTT 2.5mm (sq) cable, the power supplies are 2 x 12v/40a/200w so I think safety wise it should be ok?

The strange thing is though, I did change it from 800/default to 900 / 1000 and it made the LEDs turn off, but going back to default turned them back on, so I’m not entirely sure. I will have to check it when I’m next back at work!

The ABL simply reduces the brightness for all the LEDs turned on so that the “calculated” current is under the maximum you set. It really has no way of measuring the power drawn from your supply.

For 650+ LEDs, if you set the ABL to 1000mA you’ll get a maximum of 1000mA/650 = 1.5mA per LED which will not be very bright at all. Taking the math from the mfg. specs you gave - 23W/m / 96LEDs/m = .24W/LED. At 12V - .24W/LED / 12V/LED = .020A or 20mA for an LED at White maximum brightness.

That means 900 LEDs would need (in the worst case) .02A/LED * 900 LEDs = 18A. Your 2x12V-200W power supplies can each supply 16.66A (not 40A as you posted later) so you have plenty of power to run the setup at full brightness.

Depending on the wire you use for injection, you may need to watch your power injection points a little, but splitting the power across 2 supplies means each only has to deliver 9A max. Inject at the “1/4 points” of the strip - divide the total length by 4 and connect the 2 supplies at 1/4 and 3/4. That gives you 4 sections, A,B,C,D inject at the A to B and C to D junctions. Each injection wire needs to supply 9A worst case (use a 10A fuse on each).
Go to “1/8th points” and you get sections AB, CD, EF, GH for 4 injection wires each carrying 4.5A (5A fuses).
At “1/16th points” AB CD EF GH IJ KL MN OP, 8 wires carrying 2.25A (3A fuses).

I’m not sure what size wire you’re talking about - 2.5mm sq would be pretty huge, the physical copper wire (not the insulation) would each be 1.8mm in diameter and quite the over kill unless you have a really long run.

In the end, you can probably disable the limiter or set it to 18000mA and tell it your LEDs need 20mA each. Your setup doesn’t really need the limiter function and I suspect it’s confusing the situation.

As far as the data line is concerned, there is no “voltage drop” issue as there is very little current required.
The much bigger issue is driving the high frequency data signal over the 4-5m from the level shifter to the 1st pixel. The shifter definitely helps but IMO you’re getting to the end of what works reliably, it’ll get very cabling dependent. You can look up differential TxRx boards (RS485) for a good long distance driver solution.

Given that the current setup has been working reasonably well, you can try to “double up” on the data signal pretty simply by connecting GPIO16 to 2 different Inputs on the 74AHCT125. That will give you two independently buffered outputs, one for each LED strip, from a single data stream.

1 Like

Hi Divsys. Thanks so much! Your reply was truly superb.

Yes, you’re absolutely right about the voltage, my mistake. So I have 2 power supplies, the Meanwell LRS-200w / 5v and the Meanwell LRS-200W / 12V, I was working with the 5v / 40A version on the weekend with SK6812s, so that stuck in my mind… you’re 100% right, the 12V version is of course 16.6A.

Excellent Idea about running the data signal to another input, really appreciate that idea!

The good news is I disabled the current limited today before your post, as I was keen to have another play, and it all seems to be working well on the prototype I did at home with some left over LED tape. I’m hoping that is all that need to be done when I go back into work. Thanks for the advice over the injection points, I actually built my setup 2 years ago and it has been working well, with injection exactly as you say (the tape I used was 144 pixels per metre, not 96, but the data sheet is no longer online for that), with 12 metres of tape. I injected every metre or so.

The reason I chose 2.5mm2 cabling was due to the fact that I was future proofing my wiring if I changed the 12v tape to 5v in future. I actually missed off an extra run of 600 LEDs (that I had disconnected temporarily). This would mean potentially then I would have approx 1500 LEDs at 5v, each taking 60mA (so I guestimate) which would mean 90Amps in total. I would then potentially run 3 power supplies so I could run each at 30A each total. This obviously is a huge load so I would need a super thick diameter wire. Complete overkill at 12v though, but I wanted to make sure it was 5v ready in future. It’s in a reception desk that cost me over £20,000 to manufacture…so I thought it better be ready if I need to make any future changes!!

Thanks for the info on the data cable length - I’ll try and increase the gauge, reduce the distance, and try this out this week. Excited to finally be able to drop the Amazon SP105E device and install the ESP32!

Glad you’ve got a good handle on it.
Personally I would plan on a higher voltage bus so you can use lighter distribution wire and then add a buck converter (or two) down to 5V at each injection point. That should negate the need to ever run 10A over more than a few cm. Stick to a bus of at least 12V and go up to 24V when dealing with long/higher power runs. Using buck converters means you’re much less worried about voltage drops.

WLED is going to open up a world of new possibilities, have fun and keep us posted - maybe post a run down in Projects… :grin:

1 Like