2

We are working on a new project in which we have designed and prototyped a new custom PCB to include an ESP32 in which we are controlling a WS2811 12VDC LED strip via Bluetooth. I am the business owner trying to manage the project, not the Electrical Engineer who designed the circuit.

The LED strip will have 85 controllable nodes via the data line. (12VDC LED strip)

A 3.3V to 5.0V level shift has been built in to the circuit.

Currently, we are working on firmware, etc. and also the mechanics.

We are finding that when we have a length of cable that is greater than 8-10 ft, there begins to be a significant amount of distortion in the signal line causing random colors and flashing with the LED's.

As mentioned above my EE has a 3.3V to 5.0V level shift built into our board. But, we are still managing to get a 'dirty or weak' data signal causing random colors and flashing.

We have tried many different wire combinations i.e. two twisted pair, independently shielded, (20, 24 gauge, USB 2.0 cabling, etc. It does seem that a twisted data line set is better than not twisted.

Most applications will be within 15-20 feet. So, my goal is to do what is necessary to obtain solid power and signal lines at this distance without requiring any additional external boosters, etc.

I realize that without looking at the circuit and components used it is difficult to make any solid analysis. But in general terms, should this be 'doable' with the given components of a ESP32 + Level Shift?

What other additions / modifications to our main board should we consider in order to deliver a solid signal at this distance.

What wire / pairs / shielding do you recommend for power and data?

My EE is suggesting for extreme long lengths to add a small transmitter at one end of the cable and receiver at the other end.

I would like to avoid this if possible, especially on the 15' long or shorter cables. Maybe use a transmitter/receiver setup on longer cables only.

Any suggestions or recommendations for board considerations and/or cabling would be greatly appreciated.

Update:
Thanks so much for all of your input.

As mentioned in the original post, the strip is 85 nodes of WS2811 LED strip.

This is a 12V power strip with low voltage data line.

As you mention, yes worse case is static white LED's with high brightness. We are attempting to design for this worse case.

You are also correct that each LED strip will require a lot of power. From our measurements, this 85 node strip utilizes about ~ 3 amps of power at the brightness we are running.

We don't believe that we are not having problems with the 'LED strip'. That is, when the strip is attached to the control board with a 'short' cable 3 meters or less, the LED strip works great. All colors and brightness seems to be fine.

It is when we try to move the 85 node LED strip away from the control board with a longer cable ..for example, 8 meters, there becomes issues.

We can only conclude that the data signal at this point has become weak and distorted. The LED's are doing what they are told from a dirty/weak signal.

According the the EE designing the circuit, the data line is level-shifted from 3.3 to 5V. The data line amperage is 2.5mA base on the voltage and resistor being use.

Let me make this a little more complex. As mentioned, we are using an ESP32 so we have bluetooth control, I/O, etc. We are actually not controlling only (1) 85 node strip, but rather (4) LED strips independently of each other. That is, 4 different outputs that we control independently (all the same kind of 85 node LED strip). So, I am wondering if we are running up against the wall as far as how much drive the ESP32 + Level-Shift can handle. As mentioned above though, we can operate all 4 of these strips at solid white, full brightness each attached to the control board with a short 3m cable. As soon as we extend any of these cables, the LED strip on that output is affected. The other strips on short cables still operate fine.

Currently, we are experimenting with different resistor values that alter the data line amperage. Hope to be back with some news soon.

Is this a control/ESP32/Level-shift issue? Or, is this just as much signal that we can expect to get from this setup and we need to incorporate a transmitter/receiver setup (+data/-data) and its own pair for each cable?

It seems based on the application, that the WS2811 LED strip is best for 12V applications when it comes to addressable LED's.

We are open ears....

SamGibson
  • 17,231
  • 5
  • 37
  • 58
MKAL
  • 21
  • 2
  • Is that 10 feet the total length, or between each pair of chips? – Wouter van Ooijen Feb 08 '22 at 18:07
  • You mention 12V, but the ws2811 is a 5V chip. How do you power the chip chain? – Wouter van Ooijen Feb 08 '22 at 18:08
  • (I make one remark per comment to structure the discussion) Beware: there have been discussions on reddit/emedded about the reliability/reproducability of these LED chip strips and the consensus was that they are unsuitable (too unreliable) for mid-volume product. – Wouter van Ooijen Feb 08 '22 at 18:10
  • 1
    If you put 85 RGB leds on 20 mA each you have 5.1 A. That can seriously screw up your ground level. – Wouter van Ooijen Feb 08 '22 at 18:12
  • if it's borderline working, i find dropping the voltage from 5v to 4.5v or so often fixes it by reducing current and bringing the data line closer to the power rail. it slightly dimmer but hard to notice in most setups... – dandavis Feb 08 '22 at 18:49
  • Please post the schematics to see if there is something wrong there. – Justme Feb 08 '22 at 19:44
  • @rdtsc the data rate for this type of strip is fixed (at least, the high pulse part is fixed) – Wouter van Ooijen Feb 08 '22 at 21:55
  • MKAL - Hi, Part of the "Stack Exchange approach" is that questions and answers are in different places - unlike typical internet forums. If you want to add new information to your question, you must click [Edit] below the question, *not* write the new information in the box below labeled "Your Answer". Therefore I have deleted your "non-answer" (since it was not the solution to your question) and moved the new information into the question, as an edit. There's more info about the site rules in the [tour], the [help] and the [FAQ](https://meta.stackexchange.com/questions/7931). Thanks, – SamGibson Feb 09 '22 at 00:31
  • Sounds like the issue might be ‘ground bounce’. Estimate and measure the voltage drop across the 0V wire between the start of the led strip and the source (esp32/level shift). Better still - use an oscilloscope with the probe on the led strip end. Ensure the scope earth is tied to the esp32/driver 0V. I’d expect you’ll see the led 0V jump up. If it is over 0.5 V, then that’s your problem. You really need the driver at the start of the led strip to remove any voltage drop from the power wiring affecting the data signal. – Kartman Feb 09 '22 at 03:47
  • If you think the length of the signal line is the problem you could try the 8 meter line with single ws2811's each let's say 0.5 meter as signal re-shaper. – Wouter van Ooijen Feb 09 '22 at 08:01
  • You mention 3 A with all LEDs fully on. That is much less than the 5.1 I calculated. This is something worth investigating. Does the problem remain with less LEDs (let's say 10) but still at the long (8 m) distance? – Wouter van Ooijen Feb 09 '22 at 08:03
  • Another experiment: use RS485 to transmit the signal. This should be much less suceptible to ground currents. – Wouter van Ooijen Feb 09 '22 at 08:05
  • Another experiment I'd try: with the lonng distance, power the LED strips *from the far end with an independent power supply*. – Wouter van Ooijen Feb 09 '22 at 08:07

2 Answers2

1

The WS2811 re-transmits each bit it receives, so there shouldn't be any data reflection issues. But you stated that there is distortion when the LED strip reaches 8-10ft. How much current is the system pulling at that point? Can you adjust the current per LED down? If you can, maybe you'll find a clue about where the distortion is coming from. As one of the commenters points out, you can get to 5 Amps pretty fast. How is the power and return layout? Is there enough copper/wire gauge to prevent voltage drop on the ground line?

Aaron
  • 7,274
  • 16
  • 30
1

WS2811 draws significant power (0.3 W/led, when white color and full brightness) so you might be experiencing voltage drop at the end of line as well as the ground level rising due to resistance on the led strip. As you said, it need s robust power supply.

The actual power supply you are using might be able to provide enough, but the tracks on the LED strip are too thin and therefore resistive. You need to route the power +5 VDC and GND -connections every so many meters of strip you have. I would recommend to run extra wires every 3 meters, but if you are only experiencing trouble with 15 ft, that would suggest you could do with longer distances.

One option that sounds like an easy solution is to use a WS-series led strip that uses 12 volts, as then you would draw less current with same power. But that will easily lead to thermal management problems (overheating) as WS-chips are not as efficient with higher voltages as normal LED-strips are.

Also personal experience: There might be faulty chips on the strips that might produce random errors. This occurs also on the 4-pin strips. In those cases I desolder them and solder in a 1-led size part of a new strip. I don't want to blame the faults on the strips, they are just a type of device that gets handled a lot and might have experienced ESD-damage.

Ralph
  • 3,026
  • 1
  • 4
  • 22