Smart Strings in Icicle Mode?

[Corey872]

In reading through the pixlenet specs it seems all commands are transmitted through the wires to all nodes - any given node just waits "X" number of commands before picking it's command out of the string.  Based on this, and given enough soldering and joining, would it be possible to run smart strings in an 'icicle' light configuration - ie have a main wire or strand with several 'cicles' of 3, 4, or 5 nodes hanging below?  Or would that cause unforeseen problems?

If that is possible, I suppose if the main horizontal part was also a smart string, the overall array would be limited to 128 nodes?  Though if a person ran heavier wire for the horizontal part, would it be possible to run more nodes off of one SSC?

Just doing that 'thinking out loud' thing again.

Corey

[RJ]

no the controller limits it to 128 nodes.

If I understand no it would not function correct that way.

RJ

[rrowan]

Howdy

The node limit of 128 for pixel nodes is built into the SSC so you don't over drive the current through the cat5 cables and the ribbon cables

You could add welding wire to the nodes and it still would only do 128 nodes max

Cheers

oops, RJ beat me to the post again

Rick R.

[Lucas]

However, you could simulate Icicle Lights 

How high a icicle density are you thinking about?

Using nodes, a section of nodes for the icicle, then a longer section of cable to run back up the icicle and across to the next icicle  (It's only 3 wires) or using ridged strip (10 led version) hanging vertically with cable between strips.

[Corey872]

That is a good thought.  There are several ways to simulate the icicle look.  From the other posts, I gather we are locked into 128 nodes max.  Though I guess I'm not clear if the nodes have to remain linear throughout the string or if they could be branched?  If you will allow me some primitive cave art:

Branched - Would this work assuming all wire polarity/ID was matched and the 128 node limit was observed? and/or what if the top strand is just the 3 carrier wires - no nodes?




Series should certainly work - just require extra wire and more soldering! ?

[loveroflife96]

Branched will not work due to the nature of the DMX signal that is being passed through the wire and how it addresses each node.  The series drawing you have done will work, but would require you to individually turn on/off each node on each icicle which would currently be a programming nightmare. (unless one of the updated programs comes out with some sort of pre-programmed layering tool or something like that)

[chrisl1976]

This is a design I am having made made for me. Although for a different pixel type and controller, yet the same principal should apply as long the SS has no issues with extending the wire length between pixels.

[tbone321]

The problem is that there is NO DMX involved here.  PixelNet feeds the controller and the controller then sends some other protocol to the nodes which is not DMX.  I am not sure of exactly how the nodes process but I can be fairly sure that they do not wait for their address.  How exactly would you address them anyway?  They must be positional devices which means that they respond to the first data set that they see, remove it from the stream and pass on the rest.  Because of this, I'm not sure what your parallel string configuration would do.  I would bet that each icicle would mimic the rest on the string.  While this would make sequencing easier, it would also limit what you could do with them.  Another possible issue you might have is excessive load on the controllers control signal.  In string mode, each node regenerates the signal and passes it on so the controller effectively only sees the load from one node.  In the parallel method, the controller will be sending the control signal to the first node on each icicle and if you have 10 or 20 icicles then you have a much larger load.  I don't know if it would hurt the controller but it might degrade the control signal enough where it simply doesn't work or is inconsistant.  Another issue with the parallel method is how are you going to insulate those T connections.  They can be a real PITA to do.  While the series method requires muc more work in sequencing, you have total control over the icicles and can lead to some amazing effects.

[loveroflife96]

Thanks for the correction tbone on the DMX comment...it doesn't run DMX, but a different protocol....only had one cup of coffee before I started thinking about the issue...this day it takes at least four to get my engines up to full power!

[Corey872]

As I understand from the wiki, all commands are sent to the whole string.  At some point, the nodes are assigned a 'number' within the string [more correctly each channel color within the node is assigned a number] and any given node/channel just waits "X" number of commands before picking it's command out of the string.

Based on that, it would seem things would progress as:

node 1 ...channel 1,2,3
node 2 ...channel 4,5,6
node 3 ...channel 7,8,9

Once those nodes are programmed, then node 2 knows to pick the 4,5 and 6th command out of the string after the start code is issued?  It would not seem to matter where the node is as long as it was programmed with the proper channels in the beginning and kept track of thereafter.

Though as pointed out, light show programming would be a nightmare and I don't have a feel for how 'stable' the node address programming would be.  If it is a 'once and forever' type thing, I guess a guy could always program the string as a string, then cut and solder into icicles.  But if they require periodic reprogramming like some of my X10 stuff, that would further complicate.

Anyway, not meaning to beat a dead horse, here - just passing some time 'till I get my hands on the hardware!

[rrowan]

Slight confusion there.

The Start address is part of the SSC along with total number of nodes.
Each node chip reads the rgb values in number sequence its in.  So the nodes have to be in serial layout to work
Just like the layout that chrisl1976 posted


Cheers

Rick R.

[tbone321]

It appears that you are reading the WIKI incorrectly.  It says that the whole string is sent to all hubs and that the hubs send the full string to each controller.  As Rick said, it is the controller that is given a start address as well as the number of nodes on the string so that IT can send the proper commands to the nodes.   If you also look further, you will see that the controller can be configured for string, pixel, and hybrid mode which is a combination of the two.  While the node addressing thing may work with string or pixel mode, how would you get hybrid mode to work that way.  That would be a nightmare.  RJ also made the statement that when the pixels do go on sale, that they will be in pre-measured lengths and the coop members need to order the closest lenght to the size that they need.  They can then cut them down to the exact lenghts needed and use the extras to make new strings.  This could turn into one mixed up mess if the nodes themselves were addressed.  It is a much easier method to have the node wait for the string control indcator and then use the next three bytes and pass the rest on down the line.  This way the pixel automatically takes its place in the string no matter where you put it.  If you were to take the 5th pixel out of one string and place it in the 20th position in another one, it becomes the 20th pixel with no need to readdress it.

[RJ]

Of course the series shown will work fine but the other would not work as intended. The pixels have no address they have location in the chain and they do not know what it is. So the data moves from one end to the other.

RJ

[CaptKirk]

RJ,

If I understand how the nodes work- when powered up they initialize in series, each one taking the next address in the series- node one grabs address 1 (or 0), the next one then knows it is 2 (or 1), etc.

If you wired up icicles like Corey's first drawing, node one would be 1, the first node on the icicle would be two, but the node just past the icicle would also be two (so now there are 2 node 2s) the second node on that icicle would be 3 but so would the third node on the main line and so on.  This sort of thing would happen on down the string with multiple dups of addresses?

Or am I just full of that stuff I am usually full of??!??!

THX, Kirk

[tbone321]

You are half correct.  First of all, THERE IS NO NODE ADDRESS!!!  The nodes take the first 3 bytes of the command string and pass along the rest.  It really is as simple as that.  If you set up the icicles like Corey's first drawing, then the behavior would be like you said but it would have nothing to do with addressing and everything to do with relative positioning.