Progress?

We put gutter guards (i.e. "leaf protectors") on our gutters this spring and now I've got to make new hangers for my RBLs. It seems the RBL's work and look best when hung along the gutters above the garage, but my old hangers won't work now with the guards on. So I'll have to come up with some other way to hang them up there.... I may even have to fire up the 3D printer again....
 
I have a few props with xConnect and they work well too. I've always believed in doing a robust job with the least amount of time and money.

Every year I hear people touting centralized controllers, with props 20-50 feet away and then asking "why do my pixels flash?" My design philosophy has consistently been: One cheap controller, one cheap power supply, one prop and keep everything close together. So, if I can do this for less money but same result, I'm likely to change my process.

I still have a lot of heavy duty extension cords because the LOR users kept telling me to buy lots of extension cords, because you're going to need them. Yeah, I didn't. 😂
 
it's been said a time or two here , 24V -> 5V via reduction . I like this scheme and will hopefully implement it .
I unfortunately have other obligation for this year ( an unforeseen move is in the works ) this is a debbie downer
for anything holiday related .
 
I'll bite.
I do use jst connectors on my new 5v seeed matrix but I use 3 and 2 pin automotive type connectors on
all my 12v bullet pixels. From the beginning I never liked the plug and twist connectors.
I do use Ray Wu connectors on my 10w floods. I have a 3-4-3 T hanging out of them and daisy
chain 4 on either side of the front of my house. I've just never took the time to replace them.
 
I can get away with JST for Halloween, but not for Christmas because the climate just doesn't rain that much for Halloween. No need to solve IP68 problems that don't even exist yet.
 
Mine are laying in the snow and rain. Even in puddles of water. I had a plastic bag on one and it filled with water that the connector ignored. Since then I just use the corrosionX and leave them be.
 
Ironically the same day I touted XConnect pigtails as cheap and plentiful, I went to my stash so I could build up my 2 new receiver boxes and I was 3 short lol. Thankfully I found some in another stash, but man I need to order more next year.
 
MartinMueller2003 said:
Mine are laying in the snow and rain. Even in puddles of water. I had a plastic bag on one and it filled with water that the connector ignored. Since then I just use the corrosionX and leave them be.
Click to expand...
I just ordered a non pressurized bottle of CorrosionX. I'll give it a try, it might save the 1-2 times it really does get immersed in water.
 
I also only use the "Liquid" version. I hat the overspray. While I love how it works, you need to be carefull. It gets on everything.
 
One more warning on CorrosionX: NEVER lose the cap. An uncapped bottle WILL fall over and spill. Once it spills, anything in a 1m area will get a film of the stuff (including you as you clean it). You can clean it up with a bit of work (and lots of paper towels stuffed into a bag), dont ask how I know this. Keep in mind it is designed to find and fill the tiniest cracks in everything and it is very good at its job.

On the plus side, I found a set of diagonal cutters that had been lost in the yard for 2 years. Rusted into a solid block. Put corrosionX on them and put them in a box for 2 months. Came back and worked them free, wiped them down and still use them today.
 
Dielectric Grease, but again it makes a mess if over applied and it finds it way to the outside of the connector where fingers go.
But it's great at keeping moisture out!
 
angus40 said:
it's been said a time or two here , 24V -> 5V via reduction . I like this scheme and will hopefully implement it .
I unfortunately have other obligation for this year ( an unforeseen move is in the works ) this is a debbie downer
for anything holiday related .
Click to expand...
Do you mean to drive cables with 24V data and then reduce it down to 5V at the pixels?

I actually just did a bunch of testing a few weeks ago (watching the data stream on an oscilloscope) on various methods to drive single ended (aka normal) pixel data down long cables. Being an EE / CE, I figured there has to be a solution to this problem, and after digging in and doing some experimentation, there definitely is. Higher drive voltage is actually NOT the answer. There's 2 factors that make for successful data down a long cable:

Drive strength
Proper termination

With weak drive strength, you get rounded off corners of the transitions. This is what weak drive strength (AND bad termination) looks like on a long cable. This is looking at both ends of the same long cable:
weak_drive_strength.png

Notice how all the rising edges are rounded off, and rise very slowly. All the "jaggies" / steps are the bad termination resulting in reflections bouncing back off the ends of the cable. The common "line driver" chips used in most controllers DO NOT have enough drive strength to drive sharp edges into long cables. I did some testing, and found much better chips that have higher drive strength that give MUCH faster edges on long cables.

The other problem is termination. Somehow, the 'norm' seems to be to put a series termination resistor at the CONTROLLER (source) end of the cable. The value that everyone settled on seems to be 33 ohms in series with the controller output. This helps somewhat, but this is actually incorrect as well. Terminating the cable at the DRIVEN end leads to these artifacts as the signal reflects off the far end of the cable. You can get away with it at shorter cable lengths, but when the cable length gets long enough, it causes issues too. This is with much better drive strength, so the rising edges are sharper, but a 33 ohm source side termination, as everyone uses these days:

source_termination.png

Notice the reflection artifacts caused by the signal bouncing off the far end of the cable and having to travel back to the controller end before being terminated by the 33 ohm series resistor. Before the reflection makes it back to the terminating resistor at the drive end of the cable, you end up with a partial pulse at less than the full output voltage, then once the reflection gets to the terminating resistor at the source end of the cable, then the output voltage rises to the full drive voltage. The same thing happens when the data line transitions back to GND - there's a transition to an intermediate voltage until the pulse makes it back to the source end of the cable, and then the voltage actually goes to 0. When the cable gets long enough, this starts to interfere with the sensitive timing of the WS2811 protocol, and you get flickering pixels and garbage data.

Using higher drive strength line driver chips, as well as putting the termination resistor at the FAR end of the cable (aka at the input to the pixels, not at the controller side), I can easily drive single ended data over 80 feet at 5V drive voltage, tested with real cable and real pixels, and have clean results. I'm also not sure that 33 ohms is the correct terminating resistor value either - I was testing with a pot, so I could vary the resistance and find the value with the cleanest waveform (aka matching the cable impedance and thus eliminating all the reflections). The only place I had enough room to have everything setup for doing this was on the kitchen table, and I had to take it down before completely finishing all the testing I needed to do (including measuring the termination resistance that worked the best, at least in my particular setup with the particular cable I was using).

I ended up coming up with something else entirely to solve my problem (HERE) but I feel like the air should be cleared on that it IS possible to drive single ended data on long cables, but not the way everyone is trying to do it currently. The answer is NOT higher voltage signalling, and the answer is NOT putting 33 ohm resistors at the controller side of the cable (even though that's 'easier').
 
I just realized you're probably talking about 24V power and then stepping down to 5V at the pixels, not the data line and driving long cables, but that doesn't change anything about what I said about driving single ended pixel data down long cables
 
Correct. I use 12v and 24v on the feed lines. Then run a tap to a buck converter very closs to the injection point. The buck drops the higher voltage down to 5v and feeds the pixels through a 10cm injection wire. On the plus side, if the feed line has a significant voltage drop, anything greater than 7v on the input to the buck results in rock solid 5v at the pixel. That means I can have a 17v drop on the 24v feed line and everything still works perfectly.
 
MartinMueller2003 said:
Correct. I use 12v and 24v on the feed lines. Then run a tap to a buck converter very closs to the injection point. The buck drops the higher voltage down to 5v and feeds the pixels through a 10cm injection wire. On the plus side, if the feed line has a significant voltage drop, anything greater than 7v on the input to the buck results in rock solid 5v at the pixel. That means I can have a 17v drop on the 24v feed line and everything still works perfectly.
Click to expand...
I mostly (actually I think exclusively - I don't think I have any 5v pixels at all anywhere) use 12V pixels, some of them the "regulated" ones that have the built in 5v regulator, so essentially it's the same - 12V supply can drop to whatever the cutout voltage is of the 5v regulator in each pixel and they all still work just fine.

I did have one regulated pixel in my mega tree light on fire last year, which was a bit shocking! I thought the ones I had were supposedly "fixed" after the fire issues - I guess not.

I've been doing some reading here (haven't been here for a long time), and realized you're basically the maintainer of the espixelstick firmware now. I agree with the sentiment in the other thread that there should be a "release" of V4 proper, as it does tend to feel like a dead project otherwise. I just discovered your fork today, and all the updates you've put in that aren't in the main repo.

I'm just gearing up to use more of them based on the V4 firmware that can play a sequence off the local sd card and just get timing reference over wifi - that's a real killer feature! I want to expand my display to some neighbor's trees on the other side of the street, and wifi is the way to go to get timing reference over there without wires, and without sending all the pixel data over wifi. I also have a crazy idea to expand to many more trees on both sides of the street, if people agreed to have it there (and if all the trees don't get killed by something - so many of the large ash trees down the street have died because of the emerald ash borer beetle, and the remaining few will probably be dead this year)

I have some 'original' expixelsticks that can only stream the data over wifi, and I ran into problems adding more than 7 of them on my network. Things just got choppy and unreliable. I still use the original 7, but my original plans of just sticking an espixelstick on everything and running everything from wifi didn't go as I envisioned, so I had to go to PB16 controllers, and tried to keep everything on wired ethernet when possible, running from a master FPP on a raspberry pi that's the main controller / timing source.

I think you should check out my other project / thread I linked to in the post above - you might find it very interesting :)
 
Time to spark some discussion again.

Before leaving for our camping trip over the 4th i completed two receiver boxes that will control the canes. Almost. Turns out I didn't have a suitable power cord for one of them, so I had to order one off of ebay (it was cheap). It came today, and I had everything else wired up already so I dropped it in and now I have both boxes built and ready for my "full scale cane test".

As you may recall from above, I have 23 out of 25 built and am not stuck waiting for more seeds. I only nned 24 for the display, so I'll probably end up doing my "full scale" test in sets of 12 (each half of the driveway) separately since I'm not sure how long it will take to get the seeds.
 
BillW's story of discovery has inspired me. I have a couple of PX1's (at least one still unassembled) and a box of 10 of the original RobG 10W LED floods with the pixel conversion add-on. I think I will finish those up finally. Right after I fix the bad strings from last year's display....the ones that have been sitting on my workbench since January....yeah, those. :cool:
 
You must log in or register to reply here.
Back
Top