Author Topic: Connectors & cables - seeking suggestions on MegaTree design  (Read 737 times)

Offline ags

  • Newbie
  • *
  • Join Date: Mar 2014
  • Location:
  • Posts: 13
  • Kudos: 0
I am upgrading my pixel MegaTree after 7 years of use and refinement. I'm moving away from a uController design (Parallax Propeller) to using the BBB because it has run out of headroom (and is relatively expensive, especially when compared to new SBCs available). I will also be switching from GECE pixels to WS2811 pixels. And doubling the pixel count to 8Ki.


Since this is a specialized design (and for the fun of it) I'll be using my own BBB cape design. The challenging part (for me) is figuring out how to connect so many pixels to a single, central location. The planned (2017) pixel configuration will be 32 strands x 256 pixels each. Using standard Meanwell supplies I'll probably end up with 8 supplies, each driving 4 strands. Everything will sit outside, under the tree, with just a power cord running to it. My 2016 design had a large waterproof box under the tree center, containing a single controller and 4 power supplies. The supplies each drove 16 x 50 pixel strands. I broke the tree into 8 "banks" of 8 strands each, and ran power from the supplies on 8 pairs of #12 wire and pixel data on 8 Cat5 cables (with RJ45 connectors) from the central box to "satellite" waterproof boxes which housed a simple "breakout board" I designed that took the #12 power and Cat5 cable input and broke it out, providing 8 individual power fuses and 8 output (pixel strand) connectors each carrying Power, Ground and Data.


I found that the RJ45 connectors are very convenient, but not very tolerant of even the slightest bit of moisture. I also am wondering about signal integrity: the GECE pixels run at a very slow data rate (~150k transitions/sec) while the WS2811 pixels will run in bursts of ~2M transitions/sec (each strand refreshes in about 7.7 mSec). I see that the 2x10 pin box header connectors are used quite frequently, in the Falcon boards, the Octoscroller cape, and others. I think that's intended to drive a ribbon cable, which (to my knowledge) is only used inside locations protected from the weather (inside a waterproof enclosure).


So my question/concern/request for ideas is about:


* does it make sense to keep my "central w/ satellite" design and try to run the faster data rate down Cat5/RJ45 - one signal per conductor?
* or have everything in a central box, use ribbon cables for data connections to breakout boards also in the central box - but then deal with 32 cables (one from each pixel strand) all going to one location? Not to mention that the power and data feeds to each of the 32 individual pixel strands will now be 10' long, instead of only 8 "distribution" cables from the central box to the satellite boxes and 2' long power/data feed cables to each strand.


I'm wondering if I've been dealing with the same design for so long that I'm missing a totally different (and better) option. (Since I've successfully used a metaphorical hammer so far, am I turning everything into a nail?)


Thanks for any (even wild) ideas on this.

Offline AAH

  • Hero Member
  • *****
  • Join Date: Jul 2014
  • Location: Australia
  • Posts: 523
  • Kudos: 14
  • Blinky blinky blinky
    • I love blinky lights
Re: Connectors & cables - seeking suggestions on MegaTree design
« Reply #1 on: May 10, 2017, 06:53:08 PM »
Depending on how much you want to design yourself then something like the cape from RGB123 http://rgb-123.com/product/beaglebone-black-48-output-cape/ is supported by FPP on BBB. It gives you 48 outputs and then all you have to do is organise power and fusing. I'm actually in the process of building up a compatible board that has power, fusing and the ability to drive differential outputs for longer distances. For the number of pixels you want to drive I would be putting the BBB and power as close as possible to the base of the tree. Cable runs and voltage drop are always problematic.
You can call me AAH, you can call me Al, or you can call me Alan.
http://www.hansonelectronics.com.au/wp-content/uploads/Catalogue.pdf Catalogue of my blinky bits
http://www.hansonelectronics.com.au/

Offline ags

  • Newbie
  • *
  • Join Date: Mar 2014
  • Location:
  • Posts: 13
  • Kudos: 0
Re: Connectors & cables - seeking suggestions on MegaTree design
« Reply #2 on: May 10, 2017, 11:28:58 PM »
Wow. That board makes the BBB seem like the daughter board.


I don't mind designing my own board to get exactly what I want - even if not generic enough for other uses.


The question remains whether I continue Cat5 with 8 data lines and push the 8 data/power "breakout boards" out to the pixels with short individual connections to each strand, or use ribbon cables and have all connections centralized at the center of the tree, with 32 longer cables running to each strand.


Has anyone done any testing or comparison of running WS2811 data @800kHz down Cat5 vs ribbon cable? The power is just power, and cost of copper vs more enclosures, connectors, etc.

Offline AAH

  • Hero Member
  • *****
  • Join Date: Jul 2014
  • Location: Australia
  • Posts: 523
  • Kudos: 14
  • Blinky blinky blinky
    • I love blinky lights
Re: Connectors & cables - seeking suggestions on MegaTree design
« Reply #3 on: May 11, 2017, 01:11:51 AM »
If you use cat5 cable as 4 pairs of gnd and data then you should get 10m with little issue providing you look after it with power at the other end. RGB123 do a RJ45 version of that same board. I wouldn't be too game to try 1 gnd and 7 data or even try ribbon cable beyond about 1m. The twisted pairs of cat5 cable do and fairly good job at protecting the data. If you do do your own board then make sure you provide buffering to get from the 3.3V logic on the outputs of the BBB up to 5V logic used by pixels. You should also provide output resistors between about 33 and 100 ohms to reduce ringing in the data lines.

Offline JonB256

  • Supporting Member
  • ******
  • Join Date: Mar 2013
  • Location: Granbury, Texas
  • Posts: 3,963
  • Kudos: 84
    • Granbury Christmas Lights
Re: Connectors & cables - seeking suggestions on MegaTree design
« Reply #4 on: May 11, 2017, 07:00:42 AM »
Wow. That board makes the BBB seem like the daughter board.


The older F16B (Falcon design by Dave Pitts and Mykroft) also makes the BBB a "daughter board" but it works great.
No longer in production, though. :(

Offline ags

  • Newbie
  • *
  • Join Date: Mar 2014
  • Location:
  • Posts: 13
  • Kudos: 0
Re: Connectors & cables - seeking suggestions on MegaTree design
« Reply #5 on: May 11, 2017, 12:06:31 PM »
@AAH I've heard similar opinions expressed frequently (use pairs for signal/ground, not 8 signals) but I don't understand how that would work. I understand how twisted pair cabling works - and it is predicated on a balanced (differential) signal. But that's not what this is. Either the (signal) ground is left unconnected at the pixel, or it is tied to the ground from the power supply. In the first case, there is no reverse flow in the pair; in the latter, any reverse flow would travel along the (supply) ground as it is a much larger conductor with much smaller resistance.


What am I missing? (serious question, not rhetorical or sarcastic - in case not clear from context)

Offline AAH

  • Hero Member
  • *****
  • Join Date: Jul 2014
  • Location: Australia
  • Posts: 523
  • Kudos: 14
  • Blinky blinky blinky
    • I love blinky lights
Re: Connectors & cables - seeking suggestions on MegaTree design
« Reply #6 on: May 11, 2017, 08:01:28 PM »
I haven't tested it myself and being an unbalanced signal being sent down twisted pairs you're not using the full/proper benefit of what twisted pairs are designed to do. In theory you may get away with using 6 (or 7) signals and 2 (or 1) gnds but I would always prefer to have the additional shielding (of sorts) provided by using the pairs with 1 as a signal and 1 as gnd.

Offline ags

  • Newbie
  • *
  • Join Date: Mar 2014
  • Location:
  • Posts: 13
  • Kudos: 0
Re: Connectors & cables - seeking suggestions on MegaTree design
« Reply #7 on: May 11, 2017, 08:08:16 PM »
Sounds like an experiment is in order.  :)
Unless someone has already tried it...

 

Back to top