Hello, I just ordered a PixelBlaze, hoping to use it for the project below and would love to get feedback on whether I’m on the right track. I’m new to LED projects of this scope but I’m very eager to learn!
Some questions I have:
WS2815s are 12v. Do I need to do anything else aside from the buck converter going from 12v->5v to power the PB+Expanders?
I have the basic expander pictured. Should I get the Pro Expander? What are the upsides to it?
Anything wired incorrectly?
I’m planning on using 12awg and 14awg wiring for much of this to avoid voltage drops, as this project will spread over a 25’ x 75’ area. Aiming to have 14 different zones of 730 LEDs, so 10,220 total.
Regarding the pro expander - I find when I’m building things this big, the power wiring can be a really big drag. It’s nice to be able to troubleshoot quickly and easily swap in good strips for ones that go out. The pro expander lets you do that. It can handle 15A per board, which is enough to directly drive ~1600 WS2815s directly off its pluggable headers (and more if you limit the brightness setting or never run all pixels at max intensity). It has a built-in 5V buck converter for powering the Pixelblaze, replaceable fuses, and several helpful troubleshooting LEDs.
Looking over your proposed wiring diagram (nice, by the way!) one thing jumps out to me, and that is possibility for a fairly low frame rate. That could be OK for a slow fading ambience project. The current-generation Pixelblaze v3 can generate about 48,000 pixels per seconds, depending on the pattern complexity. That works out to a little under 5 frames per second. On the forum we covered some more considerations for a similar build situation in this thread.
Your best options to boost frame rate are either to wire for symmetry (for example, if the left and right sides are always symmetric, you can split the data line and have half the number of pixels being computed), or use multiple Pixelblazes that are sync’d via Firestorm. Adding Pixelblazes should scale roughly linearly (4 Pixelblazes ~= 4X frame rate), but can be a little extra work to figure out if you need to rewrite patterns and coordinate your pixel maps to work well across the entire project.
Last, if you haven’t purchased your LEDs yet, consider the GS8208. We have some other threads here about them. On the LEDs Are Awesome FB group (and in my own experience), the WS2815s have a bit of a reputation for frequent failing, depending on the batch they were manufactured in. The GS8208s also have noticeably better brightness response curves.
Thanks so much for the detailed response! This is extremely helpful. I hadn’t looked at GS8208 before, but I’m doing that now. Based on your other thread, it seems like the “JAD-LED Store” on Aliexpress would be a good choice if I go that route?
Regarding power supplies: This installation will be at Burning Man, so those pictured power supplies will actually be IP67.
Main question: Would I be crazy to not get Mean Well branded PS’s? Obviously other brands are significantly cheaper. Is this a clear get what you pay for situation? I’m thinking through different wiring and zone options based on different PS amp ratings. Here are some lower cost options I was looking at:
Would you happen to know a good aliexpress Mean Well vendor for PS’s with IP67 options? I see non-branded ones on there but have no clue on how to gauge quality.
Thanks for the heads up on the frame rate, I hadn’t considered that at all. Since I’m very much a newb tackling Firestorm makes me a little nervous. I’ll think through symmetry options and whether we can split some data wire and run sections in parallel.
Random other question: Do you happen know if IP65 LED strips run cooler or hotter than IP67?
That’s exciting! I’m working on a large LED project for burning man as well. I can’t wait to see your work. Do you know where it will be or have a project home page?
I don’t think consumers have great data for power supply reliability - we mostly have anecdotes. I was placing an order for 170 Meanwell power supplies so I asked a few sales reps this exact question. A main metric is MTBF: Mean Time Before Failure. Long story short there’s little way to trust the figures that manufacturers publish (and before years of warranty return data is collected, it’s just a composite statistical prediction).
The feeling out there is that Mornsun power supplies will let you save 20% with similar failure rates reported from datacenters that buy hundreds of thousands of these. Mean Well still enjoys a good reputation built over years, so I went with them, especially because I want my project to survive for over 5 years. I’ve become an unwitting expert in their product lines. Check out this sheet for a comprehensive look at their 12V products. HLG, HEP, and ELG can survive outdoors without an enclosure for years. FWIW - I’m using 170 Mean Well supplies in ammo cases for weather protection. People with more experience than me say you don’t really need IP67 power supplies at burning man if you give them a little dust and rain protection.
If I was doing a one-burn project, I think it’s perfectly reasonable to use off-brand power supplies and maybe buy a spare with the cost savings. An AliExpress one I’ve have good experience with (two years nightly operation, outdoors in an enclosure) is Ray Wu’s 300W 12V model. It’s fanless and supposedly IP67. Keep your orders under $800 total to avoid the tariffs under the current deminimus rule.
For GS8208: Yea, JAD seems good; Wushi as well. Sample size: 4 strips from each vendor * 5m of 30/m. Allow 40 days for shipping - air freight right now is insane and blows away your cost savings.
IP65 strips do shed heat a little better (through the backside, especially on aluminum) but their durability and actual rainproofing after being bent a little is much worse, so I always stick with the IP67 tubes. My guess is you’d start to see any differences in heat-related failures after hundreds of hours of operation, and it’s generally pretty cool at burning man during the hours that projects are running their LEDs. Also, the rule of thumb is to run these things at 60-80% of their max current (in Pixelblaze, use the limit brightness setting). You’ll barely notice the brightness loss, but you’re definitely save 20-40% power and heat.
This is amazing, thank you! This installation will be at our camp, In Queso Emergency, located at 4:15 and A this year. Our main gift is food, we’ve made tens of thousands of breakfast tacos from 2-4pm for the past 3 burns (under different camp name IBC). We also have iced cold brew coffee and a 50’x75’x10’ public shade structure (where this project will be mounted) with a cafe area, cuddle pit, and dance floor. Definitely stop by to to check it out and say hello!
No project homepage for us at the moment. What about your project? Can you share any more details on it? I’m very curious what 170 Mean Wells will be powering
I’m diving into the links you sent. One concern: we may have to bite the bullet and use sources shipping from the US if it’s 40 days for shipping on Aliexpress. Our camp needs to have this packed up by Aug 7, so that doesn’t leave enough time to build and tinker. With that in mind, do you think it’s a big mistake to go with WS2815s as originally planned? Those are easily sourced with shipping from the US from BTF or Alitove.
When you say under $800 to avoid tariffs, is that per vendor? Including shipping charges or excluding?My Aliexpress cart is around $2k right now with about $1200 in GS8208 from JAD.
Oh, amazing! We’ll have to drop by and see each other’s work! I’m working on an art car called Titanic’s End; we’ll be at the 3&B plaza this year. We’re also primarily known as a food gift camp - we steam Bao (pork, red bean, and egg custard dim sum buns).
Our resident Wizard here (the creator of Pixelblaze) has been helping out as well.
I think the $800 is per shipment. Double check, but I believe it’s the declared value of goods on a document called the bill of lading, which is like an international packing slip.
Also, if you need more time, I bet you’ll be fine with WS2815. Just buy 5-10% extra strips.
@jeff Wow small world. I’ve had many a bao at bao chicka wow wow - love them!
I’m going to try make a final call on GS8208 vs WS2815 today. I’ve been reading that the GS8208s require an external capacitor on the power line and external resistor on the data line. Any insight on best practices here or recommendations? Does the PB take care of the data line protection already?
@cnidus Awesome, where will Bang Bang be this year? I’ll stop by for a cocktail
Huh, I’ve never needed either on my GS8208, and I’ve driven it from Advatek and Pixelblaze. Pixelblaze definitely includes a 75 or 100ohm impedance matching resistor after the level shifters, probably Advatek as well. So that advice might be geared towards someone coming straight from a data out pin on a micro.
Interesting, I’m sure that’s what it was. I’ve got my PB up and running and playing around with the different patterns. Mapper looks very cool
Another question: If I wanted to use Firestorm with multiple PBs to achieve a higher framerate, would that obviate the need for the expanders in my picture? IE could I just get 5 PBs on Firestorm with 2,000 LEDs off each PB output? This should get me 24 frames per second?
Follow up: If I’m interested in music reactivity, do I just get one sensor expander to connect to a single PB and it communicates with the other PBs?
You can do that, but not wirelessly. Your options are to wire one sensor board to multiple Pixelblazes, or to use multiple sensor boards (which will sample slightly different audio if using the onboard mic, which can be cool, or you can chain a line-in to all sensor boards). In general, the line-in produces better results IMO than the onboard mic (the mic is fine but this is true of all mics due to the proximity effect).
Yes! But two things to consider (as someone who did this with 6 Pixelblaze) - First, on the non-clocked “neopixel”, WS281X/GS8208, the output expander can increase framerate because it can accept 2Mbps in while each string of neopixels can only be driven at 800Kbps, and the expander can output in parallel (so using 3 of the outputs is most efficient).
Second, the complexity of adapting patterns to 6 Pixelblaze took an investment of time. If you have one main pattern you want to work on and adapt, this isn’t too cumbersome. In my case, I was adapting about 40 patterns and sometimes had to do a lot of copy-paste of small snippets between the 6 Pixelblazes on different tabs. It also really requires an understanding of the mapper and the map offset tricks if you want 2D patterns to work well. This is totally doable, but I’d want you to practice on 2 before investing in like 4 or more.
Any chance you’re improved patterns could be made available somewhere? and/or tips & best practices for this optimization process? I’m working on a thing now that will probably be 5 pb’s when it’s done, and would much rather build on your work, if possible, than starting from relative scratch.
@jeff that car looks really amazing and I really love all the environmental work you’re putting behind it! My first LED project will be really near you, a croquet set at camp Cosmic Giggle at 3:30 & D
