SS troubleshooting

[RJ]

He is see exactly what the rest of us see. When a node draws too much current form the data line it loads it to the point the current through the resistor is not enough to get high enough voltage. This then adds ringing into the line due to me lowering the resistor value to get more drive current.

 When he tried to add less of a resistor he found out why I told everone not to do this. It simple raise the bottom to above 0v since the pic can npot sink that much current so it does not help and over loads the pic pin.

I believe, and may be wrong the nodes are not being cleaned of flux well and so this leads to too issues.

One is a seconadary current path drain on the data line. And two, getting any mosture into the node shorts them out.

The load from one chip on the data line should not be 20ma-30ma and sometimes more!

The chinese tester has a drive circuit that can handle a large current so some times the string will work with them and not with SSC.

Sometimes it can not drive them when they are like this. 

RJ

[Steve Gase]

If the problem is with flux inside the pixels, there doesn't seem to be a fix.
Are there any suggestions what to do next?

[jnealand]

I have discovered more information on my major bad string of 128 nodes.  If I apply power to the output end of the string I never get any rainbows.  As soon as I remove power at the output end I get the rainbows.  BUT!!!!

I have a second hub with an identical power supply.  If I just move the input cat5 to the second hub with the no power applied to the output end of the string I also get no rainbows.  Makes me wonder what could be different between the two hubs other than the soldering.  Any thoughts on this?

[rdebolt]

I have discovered more information on my major bad string of 128 nodes.  If I apply power to the output end of the string I never get any rainbows.  As soon as I remove power at the output end I get the rainbows.  BUT!!!!

I have a second hub with an identical power supply.  If I just move the input cat5 to the second hub with the no power applied to the output end of the string I also get no rainbows.  Makes me wonder what could be different between the two hubs other than the soldering.  Any thoughts on this?

So if I understand you correctly, all you are doing is using a different hub and power supply with the same string and controller? What happens if you just swap power supplies using the same hub (the one that you get rainbows on)?

[Corey872]

Possibly the voltage level and/or ground line resistance?  I've noted several of my strings will always make the white color change, but miss some of the actual 'colors'.  This is explained as the white draws more power which causes a slight drop in the input voltage and slightly changes the voltage on the data line.

[jnealand]

Well that did not turn out to be the issue either.  I left the string running all afternoon while I was out hanging lights and when I came back in I had my usual rainbows back.  This is very baffling.  I have nothing left to do but start cutting to see if I can find a bad bulb.
If anyone does not know what we are talking about with rainbow results I have uploaded a short video of the rainbow in action.  This is a 128 node string.  You are not allowed to view links. Register or Login

[bcstuff]

Jim,
Follow the string till you get to the last know working bulb and lightly twist it to see if the rest of the string starts working.
If you can cause that pixel to cause the string to work/not work than that is probably your culprit.

I have found a few bad bulbs that were causing a similar issue to what you have, after I cut them out string was good.

Brian

[jnealand]

Jim, Follow the string till you get to the last know working bulb and lightly twist it to see if the rest of the string starts working.
If you can cause that pixel to cause the string to work/not work than that is probably your culprit.
I have found a few bad bulbs that were causing a similar issue to what you have, after I cut them out string was good.
Brian

I wish it was that easy.  I have gone thru every bulb and pulled on the wires and untwisted the wires so that they will lay flat or hang flat when I put them up.  Several folks have sat and watched the string and grab the last good bulb only to have a different bulb be the last good one on the next cycle.  There just is no consistency to this other than it is always in the last half (third) of the string.  I have no confidence that if I start cutting the problem will be solved.  This is a string that was supposed to go up on my roof ridge.  I think I may just put up last years LED strings or not put any roof ridge lights up this year.  I will still have my gutters and rood snowflakes up there.  RJ is looking to test a possible mod to see if that helps, so we will wait to see what happens.

[Steve Gase]

I just went through 42 strings yesterday and cleaned them up.  Whenever I saw a problem, I found the very first issue in the string where a bulb was a different color from what was expected, I cut it out and the bulb immediately before it.  After joining up the pieces before and after the cutouts, I checked the resulting string out, if good -- I then added the two replacement bulbs at the end of the string, and checked one final time.  I found maybe 15 strings with problems out of 42, but when I took out 2 at a time I only once had to repeat for a second problem bulb (pair). 

part of my check-out procedure was to swing the string around a bit, and move it a bit to see if there was anything loose that might cause a bulb to act flakey.

i only think one bulb was bad when I removed 2, but it is hard to know if the first is leaving garbage for the second to pick up, or the second is creating its own garbage.

[rdebolt]

Well after a rough night last night. All I needed was 4 more strings of 60 nodes to finish my 32 needed for my mega tree. The very first string that I plugged in did not even light. So duh, I plug it into another spot on the hub and still nothing. I am really glad that I stopped there, because it blew the fuses in my hub. I have looked at all of the nodes in that string through a magnifying glass and do not see anything out of the ordinary. I'm not sure how I am going to diagnose that string. After everything is said and done it took me 5 hours just to get 4 strings together and working. I hope that everything will still function once on the tree, but for now I am hopeful that I am done for this year anyway. Oh I forgot I still have to replace 2 fuses on my hub and build more dongles since it does not look like the either dongle will be ready for the start of this year's show.

[Corey872]

...I am really glad that I stopped there, because it blew the fuses in my hub.

Been there, done that - though my fuse blows have always been due to accidentally touching a few wires during testing, or the tip of my DMM/Scope probe slipping between two wires.  Don't know that I have seen a string blow fuses due to internal issues. 

I guess if that is the case, cutting it in half and measuring the resistance between the +/- leads should identify where the bad node is.  Cut in half again and you'll know which 1/4 of the string it's in, etc.  At the 1/4 or 1/8th level it might pay to have a good look at all the nodes - you might see something visual inside...it would almost have to be a bridged solder joint or something.

I should have enough ports still operating for this year, but when I order some more fuses, I think I'll order some 12V LEDs too.  Solder those across the fuse and it should be a visual indicator when the fuse blows.  I'd at least like to do this with a few ports for 'testing'... that way I can easily confirm the health of the fuse.  Come to think of it, I might order a few more to solder around various locations of my test SSC - then I can visually confirm power to the SSC, power out to the string, etc.

[rm357]

I've made 2 testers for my fuses - one is an LED from a string of Christmas lights with a 1K resistor (It was an extra laying on my bench) and one of the RJ45 ends that I had cut off of the little ethernet extensions that you get from RJ for the SSCs. I cut the locking tab off of the RJ45 connector so it would not lock in place. With the LED you have to get the polarity right...

The second one uses a 12v "grain of wheat" type light bulb crimped into an RJ45 - again with the locking tab cut off. This one gets a little warm, so I don't leave it plugged in. I also used some pieces of insulation stripped off of the ethernet wires to keep the leads on the bulb from shorting out and to give something for the cable clamp to bite into.

For both I looked at the back of the hub board to figure out which pins to use - 3-5 are +12 and 6-8 are gnd. It doesn't matter which pins you use as long as you have one from each set...

With both of these I can very quickly check any port in question.

RM

[jnealand]

Brilliant RM, I love it.

[injury]

Kind of interested in solutions for this as well. I'm having some of the same issues but I tend to notice a pattern so I'm indecisive on where to cut.

Watching the test pattern I'll have say node 101 incorrect for one green every approximatly 20 color changes (that's about 5 full cycles I believe). Then about 10 color changes in between it's actually further towards the source on the string, once again about 20 changes between it being incorrect once and correct the rest of the time. So they are right most of the time.

This same string does have a section towards the end that's incorrect on almost every color except white. Or at least steady incorrect at evey color change.

Leaves me a bit torn on do I try to fix the single ones up towards the top of the line, or take a stab at the big section first towards the end. Or a better way to ask what I'm wondering is could a bad node 128 cause issues for node 1, or will the problem always start towards the front of the string and work its way to the end.

[Steve Gase]

Problems at the end of the line will have no impact on nodes that come before.
Nodes earlier in the line "could" have an impact anywhere after.  So I have been successfully been doing the earliest nodes first -- I do the one BEFORE the first observable node and the node with the observable problem -- at the same time.

I am likely removing 1 good node each time, but I plan to spend january in sorting out the nodes -- if the effort is justified.

I've found after 16 bad strings, that only one new solder connection is needed mid-string.