Hey all,
I am back and working on a new project and need pointed in the right direction please. The new project has 3 strands of 2800 LEDs so I will need a PB for each strand. I can’t seam to figure out how to mesh the PB’s together and determine if I need any other hardware or expander boards for proper functionality.
If someone can give me a quick TLDR and point me in the right direction that would be great. I have only ever used a single unit in a project.
Best,
E
See lots of detail on the release post about it: Significant feature release: Sync multiple Pixelblazes
The short version is:
This is incredible and significantly eases the potential headaches for this large scale project I am working on. Can I assume that if I want to use a sensor expansion board, I only need one on the lead board ?
That’s right - the sensor board will sync its data to the follower devices.
I wanted to caution you on max frame rate. Most aim for 30-60 FPS. The two limitations are the Pixelblaze calculation speed and data speed - either can be your limiter.
Each Pixelblaze computes about 48 Kpixels/sec depending on pattern complexity, so having three PBs with 2,800 LEDs each would limit you to about 17 frames per second. Doubling to 6 PBs gets you into range for this limit.
If you’re using Neopixels, your data transmission is 800 kbps, so 2800 in one long run will be limited to 12 FPS. In addition, 2800 is a very long run, and depending on your physical spacing, data will likely degrade too much. I’ll assume proper power injection and skip that factor; still, people tend to self-limit the length of any particular run to avoid data transmission problems from EMI, impedance, attenuation, reflections, and jitter. Common limits I’ve seen people use are 250, 500, 800, or “whatever (5-10) meters is”. If I know I’ll be getting up there in length, I’ll always try to validate this with the actual pixels.
If you’re using clocked pixels (4 wires, like SK9822), you can use faster data transmission rates like 2Mbps, yielding a theoretical 30FPS for 2800 pixels in each long serial run, but you’ll still need to experimentally find that max pixel run for data integrity.
If you landed on 1000 pixels as a safe run, then your options are to use an output expander to break each PB’s 2800 (or 1400) pixels into 2-3 runs, or use a PB for each run. Your choice between these is commonly dictated by the physical layout of your project. For example, if you really needed 2800 in one long physical line, like around some big perimeter, then the output expanders aren’t a solution (because you never want more than 2-4 meters of wire between signal source and first pixel unless ALSO adding differential long-range transceivers, which just adds more soldering and complexity).
Some people instead try to inject signal amplifiers within one long run, but this commonly doesn’t work out very well.
I hope this helps, and apologies if some of this is stuff you already know.
Jeff,
You are an amazing human being, thank you for saying everything you did and i will respond shortly to show you what I am planning.
-E
Jeff, Here is a diagram of what I need to achieve. I am not so worried about frame rate as the three strips are right next to each other, diffused and I won’t be doing anything “complicated” with them. Perhaps some slow fade and breathing animations and such. I have power at both ends which I calculated to leave me buffer room on the power supply max and with 8 AWG should avoid dropout. Can a single pixel blaze run a 4m strand of ws2811 with a total of 2800 in the strand? What am I missing? incorrectly assuming? or should know.
Cheers,
-E
Oh! Actually, for this particular config, you might even get away with a single Pixelblaze!
I believe those 24V 720/m strips are going to be COBs with a single data “pixel” always duplicated across 6 physical pixels. Therefore you’ll configure the Pixelblaze for 466 pixels and get great frame rates. I don’t have any experience with max length for data integrity on these.
Seeing that they’re all next to each other, you’ll probably be duplicating the same animation across all three, right? In this case it might even work to just connect the one data output to all three inputs! If that doesn’t, then an opamp buffer should clean up the duplicated data.
So start with one Pixelblaze and see how it goes!
So if I use one PB can I still make a map for it that is 3 x 2800? I was hoping to do this to get the smallest amount of dimensionality out of the strips for effects rather than everything only to run linearly down the length ( back and forth.)
I know its not much height but the strips will be far enough physically apart and diffused so I could maybe get some up down and left right motion.
let me know if this doesn’t make sense, I know its a little weird. I’m just trying to maximize execution outcome.
Ah, ok - so if you do plan to calculate those three strips as having potentially different colors, I think I would go for one Pixelblaze and one output expander.
If I’m right about there being 6 LEDs per “logical pixel” on those strips, you’ll configure the expander for 466 pixels per each of 3 channels, be calculating a total of 1398 pixels, and get roughly 30 FPS out of it, which should be decent.
If there really are 720/m that are individually addressable in those strips, first - please share a link to them, that’s incredible density, and second, the math in my first reply stands. Use three PBs in sync mode (because a single PB limits you to 5-6 FPS, really bad even for slow fades). You might even find the neopixel transmission speed limit of 12 FPS is unacceptable, especially if ever fading to very dim or black.
Here are the strips I’m thinking of using
https://www.amazon.com/gp/product/B0DLGQJG6L/ref=ox_sc_saved_title_3?smid=A2N8TBFCW228PI&th=1
Also using an output expander sounds exciting, I haven’t done that yet.
Oh wow - 36 LEDs per addressable “pixel”.
Each controllable segment will be 5cm long and you’ll configure Pixelblaze for 3 strips x 5 meters x 20pixels/m = 300 LEDs total. The output expander might be good for signal integrity but you’ll have no problem with frame rate.
Interesting, is there another strip you would recommend instead, I need a cobb strip with high density for the projects intended effect but would something with segments smaller than 5 cm be good? Also how did you determine this, seems like something I should learn to look for and understand on the specs.
In the listing is this bullet point:
- Addressable COB Led Strip: The COB light strip uses flip chips and is linearly die-bonded on the PCB board. And this addressable COB led strip lights are designed with Ws2811IC. Because of the integrated IC, you can program each group to be a different color and do some magic effects with these led strips. One IC Chip as a group drives 36LED, with total 20pcs Ws2811 IC and 720LED per meter. Each group is addressable and cuttable.
So with 20 driver chips per meter, you get 5cm per “pixel.”
Some of this comes from the way led strips work at higher voltages. At 24v, it’s fairly efficient for one IC to drive 6 LEDs at a time. This strip would seem to run 6 groups of 6 per IC, which is quite odd.
Not that you can always trust product listing photos, but the photo appears to have 12 identifiable RGB LEDs per pixel. A 720/m density strip would be very difficult to see individual LEDs, but you can see them in the photo.
So my guess is that they are counting each of the red, green, and blue elements as LEDs to do some marketing bullshit. That means you would not have 720 RGB LEDs per meter, only 240. You still only get 20 “pixels” per meter of individual control, spread over a 5cm distance using 12 rgb LEDs.
Wow. Of course. New low! I recognize these now - they’re the ones Vitaly has mentioned in a prior thread.
Nice catch.
The denser versions I know of are 160/m
And 332/m
https://a.aliexpress.com/_mrMvklN
But they’re both 5V strips, so buckle up for a power injection soldering adventure.
So, here’s a question for you - why not use the more traditional 5050 sized chips at 96, 120, or 144/m? You get individual control, there’s 12V options, and you can get similar diffusion results with “neon” style silicon diffusers? What’s your typical viewing distance?
Hey Jeff and Wizard,
Really appreciate both of you jumping in with details—super helpful as always. Based on the specs of my project, here’s what I’m looking for:
I don’t necessarily have to use COB LEDs; they just happen to have a high density and an extra layer of diffusion. For this project, I’m 3D-printing diffusers that will go in front of the LEDs, and it’s absolutely critical that individual chip dots aren’t visible. That’s why I landed on COB, with a high concentration and brightness.
I’ve included a link below to my last project for reference. Just to clarify, I’m not avoiding soldering or wiring out of know-how—it’s more of a logistical thing. For this build, keeping the strips as single segments will simplify things significantly, so I’d prefer not to have to boost signal or inject voltage mid-run. That’s why I opted for 24V, which lets me cover longer distances without power injection.
That said, if you guys have any recommendations for high-density LED strips at 12V—whether from Amazon or AliExpress—I’m totally open to checking them out. Thanks again!
Last project https://imgur.com/gallery/led-music-visualizer-complete-build-photos-NdXFfni
You can perfectly diffuse any of those densities, it’s just a function of how thick your diffuser is and how far it is away from the strips.
When you need it extremely shallow height wise, and you need to preserve the most brightness, that’s when you get stuck in the corner of needing higher densities, And that corner eventually leads to extra soldering.
I’d use the GS8208 for this. It’s the best 12 V individually controllable pixel and it doesn’t have the trade-offs that 24V strips have.
Based on prior projects I suspect it will be fully diffuse with about .8 mm of clear PETG >= 20 mm above the chips at the 60/m density.
We sell them at 60/m and for anything denser you should check AliExpress! For example there’s a 90/m on Amazon but it’s overpriced and takes 3 weeks to arrive.
Jeff,
The issue with this project is I have a pretty shallow thickness to the whole project and the three strips are internally mounted to the parts Im making via 3d printing. You have me intrigued and I am going to check out the GS8208 strips on Ali Express. Can you speak to what makes them the best option?
Cheers and happy friday!
If your 3 strands can be made to start from the same place, a single PB with expansion board is easier to manage though:
Yes! I wrote the product page here about the GS8208 and it’s fair to call it my love letter to them. Check the linked video there.
In short: Gamma curve, and its next best alternative, the WS2815, has had some bad manufacturing batches that resulted in higher failure rates for some artists (the “LEDs Are Awesome” group on Facebook has documented these).
Ray sells 2m @ 144/m and JAD sells 5m @ 144/m if you need maximum density. Ray’s come with 3-pin JSTs and JADs use 4-pin. Either works, it’s just a preference thing. You should inject 12V every 5m if using 144/m.