(Workaround Discovered) Finding the faulty node

[tbone321]

These nodes don't have perfect isolation between the input and output sides on the data line and even there, it seems to vary from node to node.  Now much of it could be due to board assembly but some of it may be on the chip itself.  Because of this, each node puts some type of load on the string and when the load exceeds what the chips can handle, they start to misbehave.  Due to the nature of how the nodes work, it can be very difficult to near impossible to determine which node(s) are causing the issue but if it is working the way that you have it, leave it alone.  I would guess that the bad node is in the group after 81 and the added load of the nodes you are putting in front of it are causing it to corrupt the entire strings data line. 

[Corey872]

Each node has a varying amount of isolation, and the ground line has a constantly changing amount of current depending on how far down you are and even what colors are on...bright white being the highest current, bright secondary colors - a bit lower, primary colors - lower still.  If you have a node which is 'right at the limit' little things like its position in the string, and the ground line current can make a difference. 

I was able to make one misbehaving string work by simply by grounding both ends of the ground wire.  If you're seeing a node late in the string acting up - especially around the 'white' switch of the test pattern, this is a likely cause.