Help needed for Noob.

[Brodey]

Hey,

New to the hobby, I've always just used dumb LEDs, but the guy up the road whom I have unofficially always competed with has now gone this route, so now I need too in order to be better. (yeah, I'm that kind of guy). So, I'm starting in January 2024 to have everything ready for Christmas.

Wiring is my biggest issue; I have most things or will buy most things. I have decided to go down the route of 12v WS2811 LED Strings with my connector of choice being Ray Wu connectors, and WT32-ETH01 with WLED flashed and a Raspberry Pi 4 running FPP.

The ESP32 device (WT32-ETH01) is 5v for reference.

I plan on having ~4,500 pixels (nodes) by Christmas; research on the WT32-ETH01 says for perfect performance, I can run 512 LEDs per pin for a total of 2,048 LEDs per WT32-ETH01 controller, so I'll run 3 of these. Reference: https://kno.wled.ge/features/multi-strip/#esp32

I have a couple of options for power supplies with what I have available to me already, but happy to buy something else if it'll make my life easier. I had old ETH mining rigs I had a bunch of power supplies for, some old desktop ones, and some server PSUs.

Server PSU (DPS-1200FBA) 100a: With a 12x 6 Pin Power Supply Breakout Board Module, which is 3x 12v pins and 3x Negative pins per connector.
1500W Desktop PSUs from Silverstone (SST-ST1500-GS) 120a: I can convert 6 pin adapter cables to 3x 12v positive and negative.
AC/DC Adapter (JCY-1250) 12v 5a: This is just a cheap transformer that came with 1 set of LED strings I bought, probably useless.

I think my biggest issue with wiring is the 12v lights and 5v controller, and then power injection.

I managed to stumble upon this tool: https://wled-calculator.github.io/. If I say I want to run 1,000 pixels, I will need a 20a fuse. I've never used Fuses before, and if a project required one, I typically deemed it too challenging for me. But, yet, here I am still trying.

I think I need a 12v to 5v Step Down Adapter to convert my 12v negative to my v5 controller. This is my current wiring diagram.

I am using the transformer in the top left for now; the top right device is the WT32-ETH01 device, bottom left is a 5v 1a USB phone charger with the USB positive and negative wires only in use.

So, other than the ground wire going from the 12v PSU to the GWD on the controller, 5v and 12v are separated. I think this is where I need my 12v to 5v adapter to go.

In this configuration, I can get it to work if I have a 50 LED string attached and turn the controller on after I turn on the 12v power. However, the 12v transformer makes a high pitch noise, which doesn't make me feel safe. If I add another 50 Strings to the end of the other 50, I can't get it to work at all.

This is just level 1, obviously; as it gets more complex, I am just going to run 3 controller boxes with 1 controller in each, and then 5m Ray Wu cables from the box to the LEDs.

Here is a diagram of my house and my proposed setup:


So, the thick black line is my house, blue are LEDs, red is saying which controller will power which lights. With some help from Google Bard (AI), I only really need 594w of power or 5amps per string, so 1x 1000PSU would be enough, but I will run 3 to simplify wiring.

I will do a power injection every 200 pixels. This is where I'm confused. Do I run a cable from the same PSU up to the 2nd string of 100 pixels and just boost it? Because to boost, I carry a positive and a negative wire up to that point, then a positive and negative to the PSU, do I need to run a negative and another 12v to 5v adapter to run a negative from the PSU to the controller again? Doesn't make sense why I would; based on my first diagram, I already have this cable existing. Or should it not exist?

That raises the question: If I use a 12v to 5v adapter (if this is the correct thing to do), do I need fuses like the website calculator suggested I did? If so, how do I wire the fuse based on several strings and power injection?

Sorry for the long post; I only have a year to figure this out and buy all the lights, so the guy up the road doesn't show me up again.

The software side I am good at the, I'm a programmer, but I program web applications, not hardware. This is my first major hardware project; I can, however, solder, strip wires, etc, and I will be having an electrician inspect the wires etc, before running them over Christmas to ensure I'm not going to start fires or electrocute some kid riding past on the footpath (even though the kids around here might deserve it sometimes).

 

[Skymaster]

So, other than the ground wire going from the 12v PSU to the GWD on the controller, 5v and 12v are separated. I think this is where I need my 12v to 5v adapter to go.

You do not need to convert the "grounds"
Voltage is a reference - a potential difference - between two points.

So, Your 12V wire is 12 volts above it's ground, or 0V reference.
And likewise your 5V wire is 5 volts above the 0V reference wire.

By joining them together, you are providing a common 0V reference point. Which is what is needed because the data line when transmitted is referenced against it at the controller end, and at the pixel end, is read as a reference to the Gnd.

Going to your first picture, your 12V to 5V adaptor would be used instead of your iPhone charger. It would convert the 12V positive line down to 5V, (with still the same shared 0V) to provide the lower voltage the ESP32 needs.
The picture you have there, is actually 100% correct. You would join it exactly as shown, without any converter in there.

The fusing is to protect your wires. It is not dependent on the LEDs you run.
A 18AWG / 0.75mm wire is good for about 7 amps, and you would fuse as such.
That then limits the number of pixels you can use.
If you have more pixels - say 20A worth, then you need to run a larger gauge wire that will take 20A (there are two factors to come into play here, see below) - and fuse appropriately.

The two factors for cable size calculation are:
1. Voltage Drop - a thinner cable has a higher resistance, and as such will drop more power for the same current flow
2. Maximum current flow through the cable.

#2 is actually an extension of #1 - the voltage drop is dissipated as heat, and the max current is selected to ensure the cable doesnt melt.

 

[Brodey]

You do not need to convert the "grounds"
Voltage is a reference - a potential difference - between two points.

So, Your 12V wire is 12 volts above it's ground, or 0V reference.
And likewise your 5V wire is 5 volts above the 0V reference wire.

By joining them together, you are providing a common 0V reference point. Which is what is needed because the data line when transmitted is referenced against it at the controller end, and at the pixel end, is read as a reference to the Gnd.

Going to your first picture, your 12V to 5V adaptor would be used instead of your iPhone charger. It would convert the 12V positive line down to 5V, (with still the same shared 0V) to provide the lower voltage the ESP32 needs.
The picture you have there, is actually 100% correct. You would join it exactly as shown, without any converter in there.

The fusing is to protect your wires. It is not dependent on the LEDs you run.
A 18AWG / 0.75mm wire is good for about 7 amps, and you would fuse as such.
That then limits the number of pixels you can use.
If you have more pixels - say 20A worth, then you need to run a larger gauge wire that will take 20A (there are two factors to come into play here, see below) - and fuse appropriately.

The two factors for cable size calculation are:
1. Voltage Drop - a thinner cable has a higher resistance, and as such will drop more power for the same current flow
2. Maximum current flow through the cable.

#2 is actually an extension of #1 - the voltage drop is dissipated as heat, and the max current is selected to ensure the cable doesnt melt.

Champion!

For some reason, I think my issue is my data cable that goes from the controller to the LEDs. I used an 18 AWG wire for it when I disconnected my controller after setting a colour in WLED. It stops flickering and holds a solid colour like it should, but with the data cable connected, it just flickers.

Research has told me that by using an 18 AWG wire, interference can happen, and it loses connection. Even though it's less than 10cm away from it, it's losing signal and data. My research (https://g.co/bard/share/06ee3a340999) which is just Google Bard.

My question is, is there a specific wire type I'm meant to use for data? as opposed to the one I am using for power?

2 days into plugging things in, rewiring things, tinkering, settings, adjustments, and different power supplies. Literally 20 hours in the past 2 days just to get to this point with no noticeable improvement other than my wire management.

 

[merryoncherry]

"happy to buy something else if it'll make my life easier"? Based on your yard, try a HE123D, F48, K40, anything like that and some receivers (with proper PSUs with a voltage that matches the pixels). Assume the receivers will go in the spots where you have controllers now, even dumb ones will be fine. As a fellow software guy, gotta say it's nice to have the flexibility to just reconfigure stuff in the field if I made any mistakes or change my mind.

 

[Brodey]

"happy to buy something else if it'll make my life easier"? Based on your yard, try a HE123D, F48, K40, anything like that and some receivers (with proper PSUs with a voltage that matches the pixels). Assume the receivers will go in the spots where you have controllers now, even dumb ones will be fine. As a fellow software guy, gotta say it's nice to have the flexibility to just reconfigure stuff in the field if I made any mistakes or change my mind.

Thanks for your input; I had actually looked at these, and whilst they are in my price range, in my head. I can't come to terms with how these are better options. If I consider that each WT32-ETH01 can handle 2,000 pixels perfectly, I can push it to double that without very noticeable drops. The organisation and wiring seem a lot more simplistic.

It creates somewhat of a module approach or a "pack". For every ~2,000 nodes I want, I get 1 controller, 2,000 lights, a PSU, and a 12v to 5v converter, and I'm in business. I can place it anywhere in the yard with 1 power cord going to it and an ethernet cable. Coming out of it is either 3 3-wire cable that carries data or my 2-core wire for power injection, and I'm good to go.

Both XLIGHTS and FPP both actively support direct WLED integration as well.

In my eyes, the wiring will be so much more tidy, and when I do choose to add lights or a mega tree, I simply add a new "pack", and I'm plug and play.

I just can't see how any other way is actually easier. Costs aside, this way is better. Until you get to like 20,000+ lights, and your yard turns more into a controller than a yard. But I have to think to myself, am I ever going to have more than 20,000 in a yard my size? I may as well just turn my house into a lightbulb at that point. My yard isn't big enough to have 20,000; I am nearly sure at 10,000, I'd have taken up every square centimetre.

Of course, I could add a mega tree and a couple of matrix walls, but I think matrix walls don't look that nice anyway unless it's the garage and a mega tree most people normally get their own controller for anyway, so it's not even a better solution for that because I am aware I am hosting a mega tree, I can just throw several WT32-ETH01 devices into the same box and no one would ever no.

I'd love to hear a valid counterargument as to why those devices are better, other than good marketing, and an argument that doesn't include the fact that you might be defending it because you had already bought it and now have to justify the expense. I'd love to hear it. I am new to the hobby, and whilst I've already got the hardware, I'm unlikely to change. I'd be interested in what I have missed in my research.

It seems the only issue I'm having is poor wiring. Once I have worked that out, I'm in the clear.

 

[Skymaster]

Have you considered something already based on that platform, rather than re-inventing the wheel?

SmartAlecLights for example, sells poduct based on the WT32-ETH01 ESP32 which handles everything for you.

SAL32-Pixel

smartaleclights.com

You wire your 12V into it, it handles the buck conversion for the ESP32 power, and provides the four fused pixel outputs.
For $23 it's a bargain.

 

[someoneAUS]

SHEESH thats significantly cheaper than a quinled which is what I always seem to end up recommending for anyone who just wants to have a play and get a feel for it.

 

[Skymaster]

For some reason, I think my issue is my data cable that goes from the controller to the LEDs. I used an 18 AWG wire for it when I disconnected my controller after setting a colour in WLED. It stops flickering and holds a solid colour like it should, but with the data cable connected, it just flickers.

Research has told me that by using an 18 AWG wire, interference can happen, and it loses connection. Even though it's less than 10cm away from it, it's losing signal and data. My research (https://g.co/bard/share/06ee3a340999) which is just Google Bard.

My question is, is there a specific wire type I'm meant to use for data? as opposed to the one I am using for power?

The wire gauge for the data signal is irrelevant. It's such a low current that it doesnt matter.
The gauge for the power is much more relevant as it's handling the high current.

The issue is the ESP32 only has 3.3V logic outputs. The pixels are expecting a 5V logic signal. You need to introduce data buffers/boosters; which are what are present on a normal controller.
You also need to ensure the grounding is correct - and all connected properly; between the pixels, powers supply and controller.

 

[Skymaster]

SHEESH thats significantly cheaper than a quinled which is what I always seem to end up recommending for anyone who just wants to have a play and get a feel for it.

You need to add the ESP32 on to that cost as well.
The 5V version is here: http://smartaleclights.com/shop/index.php?route=product/product&path=69&product_id=154 - which comes with an optional ESP32 changing the price from $23 to $50

 

[merryoncherry]

I like @Skymaster's suggestion if you want to do the ESP32 route. That would also be better*.

In my reply, I had made several fundamental assumptions in error.
1. I have a lot of lights, perhaps "turned my house into a lightbulb". I thought you might be interested in carpeting the yard, since you said "the guy up the road whom I have unofficially always competed ... (yeah, I'm that kind of guy)". Guess that was wrong.
2*. I thought "quicker" and "easier" might be "better" for you because you were saying things like "Literally 20 hours in the past 2 days just to get to this point with no noticeable improvement other than my wire management".

I will not attempt to argue a RPI w/FPP + a handful of ESP32-based controller boards vs. a controller with differential ports. These are both going to work well, and be cheaper and easier than how you started, and at the end you basically have little satellite boxes that drive a few strings of pixels each (a receiver would do 4x800 = 3200), connected to some central point with CAT5 cables.

You were saying WT32-ETH01 - that's US$17, then a stepdown to 5V, and were starting to ask about wire and fuses, and spending a lot of time. What I'm suggesting is one of dozens of available receiver boards, usually priced between $20-$30, depending on whether they're smart receivers that daisy chain. Those couple bucks get you fuses, far less wiring to do (some of them screw onto popular power supplies and need zero wires), and support for 5, 12, or 24V.



So you're saving a lot of time and a bit of money on the little satellites by using a premade board. (At $23, I think the SAL32 is similarly going to save you a little money and a lot of time.)

Then there's the question of the central controller. I had proposed something like a HE-123D, similar things in USD are ~$200 with the BBB and SD card. Those would run FPP, and be a full player if you need that... otherwise you'd be getting a Pi and an Ethernet switch - probably a bit less than the controller but not that much more in the grand scheme. I tend to use controllers that have a lot of local ports in addition to the receiver ports... if you can make use of 16 or 32- local pixel strings then the gap pretty much gets closed. And it is very simple, all the boxes you need could have been assembled in the 20 hours you spent so far.

You could mix and match these approaches too, of course... some larger and smaller controllers together... pretty much how my display ended up.

 

[fugley]

Thanks for your input; I had actually looked at these, and whilst they are in my price range, in my head. I can't come to terms with how these are better options. If I consider that each WT32-ETH01 can handle 2,000 pixels perfectly, I can push it to double that without very noticeable drops. The organisation and wiring seem a lot more simplistic.

It creates somewhat of a module approach or a "pack". For every ~2,000 nodes I want, I get 1 controller, 2,000 lights, a PSU, and a 12v to 5v converter, and I'm in business. I can place it anywhere in the yard with 1 power cord going to it and an ethernet cable. Coming out of it is either 3 3-wire cable that carries data or my 2-core wire for power injection, and I'm good to go.

Both XLIGHTS and FPP both actively support direct WLED integration as well.

In my eyes, the wiring will be so much more tidy, and when I do choose to add lights or a mega tree, I simply add a new "pack", and I'm plug and play.

I just can't see how any other way is actually easier. Costs aside, this way is better. Until you get to like 20,000+ lights, and your yard turns more into a controller than a yard. But I have to think to myself, am I ever going to have more than 20,000 in a yard my size? I may as well just turn my house into a lightbulb at that point. My yard isn't big enough to have 20,000; I am nearly sure at 10,000, I'd have taken up every square centimetre.

Of course, I could add a mega tree and a couple of matrix walls, but I think matrix walls don't look that nice anyway unless it's the garage and a mega tree most people normally get their own controller for anyway, so it's not even a better solution for that because I am aware I am hosting a mega tree, I can just throw several WT32-ETH01 devices into the same box and no one would ever no.

I'd love to hear a valid counterargument as to why those devices are better, other than good marketing, and an argument that doesn't include the fact that you might be defending it because you had already bought it and now have to justify the expense. I'd love to hear it. I am new to the hobby, and whilst I've already got the hardware, I'm unlikely to change. I'd be interested in what I have missed in my research.

It seems the only issue I'm having is poor wiring. Once I have worked that out, I'm in the clear.

i understand where your coming from on cost as i went down a similar road with esp32 as id always used them around home with home assistant for lighting, fans, switch etc , about now when i should stopped and had a rethink!!, staying with a format i knew pretty well (ESP32 / WLED) i then dipped my toes with three dig quad and then later a couple of dig octa because at the time controls were in short supply, but what you'll find is all the stuff you could or should of added starts to mount up in additional costs ... then a couple of years deep in a rabbit hole you'll start to wounder was this the right move when you look at what you've spent

best advice that was ever given to me and should of stuck to on this one was ... big a little bigger than you need as then you'll never break it, and it'll last you a life time

 

[thewanderingpine]

Thanks for your input; I had actually looked at these, and whilst they are in my price range, in my head. I can't come to terms with how these are better options. If I consider that each WT32-ETH01 can handle 2,000 pixels perfectly, I can push it to double that without very noticeable drops. The organisation and wiring seem a lot more simplistic.

It creates somewhat of a module approach or a "pack". For every ~2,000 nodes I want, I get 1 controller, 2,000 lights, a PSU, and a 12v to 5v converter, and I'm in business. I can place it anywhere in the yard with 1 power cord going to it and an ethernet cable. Coming out of it is either 3 3-wire cable that carries data or my 2-core wire for power injection, and I'm good to go.

Both XLIGHTS and FPP both actively support direct WLED integration as well.

In my eyes, the wiring will be so much more tidy, and when I do choose to add lights or a mega tree, I simply add a new "pack", and I'm plug and play.

I just can't see how any other way is actually easier. Costs aside, this way is better. Until you get to like 20,000+ lights, and your yard turns more into a controller than a yard. But I have to think to myself, am I ever going to have more than 20,000 in a yard my size? I may as well just turn my house into a lightbulb at that point. My yard isn't big enough to have 20,000; I am nearly sure at 10,000, I'd have taken up every square centimetre.

Of course, I could add a mega tree and a couple of matrix walls, but I think matrix walls don't look that nice anyway unless it's the garage and a mega tree most people normally get their own controller for anyway, so it's not even a better solution for that because I am aware I am hosting a mega tree, I can just throw several WT32-ETH01 devices into the same box and no one would ever no.

I'd love to hear a valid counterargument as to why those devices are better, other than good marketing, and an argument that doesn't include the fact that you might be defending it because you had already bought it and now have to justify the expense. I'd love to hear it. I am new to the hobby, and whilst I've already got the hardware, I'm unlikely to change. I'd be interested in what I have missed in my research.

It seems the only issue I'm having is poor wiring. Once I have worked that out, I'm in the clear.

As someone who had 2520 pixels off a WLED board this season - the drops were noticeable. 2000 is probably your actual limit.

 

[Martin Mueller]

You are most likely going to have some WiFi lag caused by that many pixels going over your WiFi infrastructure. I suggest using an SD card at the ESPs and running them in FPP remote mode.

 

[thewanderingpine]

That was e131 over Ethernet. FPP remote + SD card makes a lot of sense for these (assuming your board has an SD card which mine did not)

 

[Brodey]

The wire gauge for the data signal is irrelevant. It's such a low current that it doesnt matter.
The gauge for the power is much more relevant as it's handling the high current.

The issue is the ESP32 only has 3.3V logic outputs. The pixels are expecting a 5V logic signal. You need to introduce data buffers/boosters; which are what are present on a normal controller.
You also need to ensure the grounding is correct - and all connected properly; between the pixels, powers supply and controller.

This, this actually was the key! The 3.3v logic output I was not made aware of. I added a booster, and the problem was solved. My nightmares are solved. Thank you for that!

i understand where your coming from on cost as i went down a similar road with esp32 as id always used them around home with home assistant for lighting, fans, switch etc , about now when i should stopped and had a rethink!!, staying with a format i knew pretty well (ESP32 / WLED) i then dipped my toes with three dig quad and then later a couple of dig octa because at the time controls were in short supply, but what you'll find is all the stuff you could or should of added starts to mount up in additional costs ... then a couple of years deep in a rabbit hole you'll start to wounder was this the right move when you look at what you've spent

best advice that was ever given to me and should of stuck to on this one was ... big a little bigger than you need as then you'll never break it, and it'll last you a life time

The costs are less of a factor here; I have money. I have time. I think what I am looking for is the hobby, to be honest, the enjoyment of building, constructing, designing and challenging myself. Everything a good hobby offers whilst also putting a smile on a lot of people's faces when they get to see the output.

I run a business, so I know the lesson you're trying to teach here, I, too, learnt it the hard way. What I'm trying to get at here is that going down the route of a Falcon, for example, seems like a shortcut that'll get me to the end, but the entire fun of the hobby is the journey, not the destination. Do I really want to go get the best of the best premade kit, hook it up and claim it as hard work and be proud of it?

I think that is why I'm going down the other route, creating a modular approach that I can attach to when I'm ready. If it costs a little more or is a little more complex in the beginning, I get to enjoy the journey, and when I see the output, I can take actual pride and excitement out of it; I'll take that route every day.

As someone who had 2520 pixels off a WLED board this season - the drops were noticeable. 2000 is probably your actual limit.

Noted! I think I plan on doing a lot less per board anyway; that's why I have several going. I think the average board will be running a lot less. But it's good to know real-world numbers, not just numbers I've read online, but something someone has tested and provided feedback on; thank you!

You are most likely going to have some WiFi lag caused by that many pixels going over your WiFi infrastructure. I suggest using an SD card at the ESPs and running them in FPP remote mode.

The model ESP device I went for is the WT32-ETH01, which is an ethernet-driven device. I knew I would face WiFi issues, so I went for an ethernet model. It also has its own modem and switch, so it'll be run independently from my home network. I "shouldn't" experience any latency; everything is the latest. I've got CAT7 cables 10gbe Switch with a Modem that supports those speeds. Whilst this is overkill, this is just stuff I had laying around from other projects, so may as well utilise it.