I am at a complete loss (Dimming LED anomaly)

[plasmadrive]

This year I updated my redwood trees to LEDs from incans.  These LED are from HLE so they are first rate stock and are all brand new.

 

With a Gen 2 PC controller that I used last year on the incans everything of course dimmed just fine.  This year I put the LEDs on and there was NO DIMMING AT ALL.  Either ON or OFF, nothing in between. 

 

Couple of points:

• There are 9 trees. 
• Each tree has between 1800-2200 LEDs.
• The largest tree has about 2200 LEDs and is drawing about .8 amps give or take a few mils.
• All lights are plugged end to end.

I pulled the Gen 2 controller and put a Gen 3 in it's place with the LED dimming curve (curve 01)

 

Once I put the Gen 3 PC controller in, the LEDs would dim ALL EXCEPT THE LARGE TREE.  It would not dim. ON or OFF only.

 

I swapped outputs with one of the other channels that is working fine and still that tree was ON or OFF only.

 

I experimented thinking perhaps a bad string (no idea how that would effect them all and technically it should not)  When I unplugged the first several strings (lower part of the tree), and plugged the rest of the tree (upper part) into that ouput, they all dimmed just fine. 

 

Then I tried powering just that lower part of the tree, they dim just fine.  However... with the lower part on say about 30% so I could tell it was dimming, I plugged the rest of the lights in and they ALL went to full on. 

 

So, either the top 3/4 works ok or the bottom 1/4 works OK but together they won't dim. 

 

I haven't pulled a scope out to look at the wave form, but what the heck is going on?  Can anyone explain?  LOR ADMIN.. can you have one of your Tech guru's figure that one out? 

 

Is there some sort of capacitive threshold that is a break over point when you have too many lights?  Is .8 amps too many LEDs? 

[-klb-]

Yes, too much capacitive storage in the wiring, without enough linear load. The LED strings don't draw much current at all below 90V, and all that wire adds up to a lot of capacitance. More than even the built in phantom load on the 3G controllers will drain every cycle. Try an incandescent C7 on the same channel, it should solve your issue. A similar high wattage resistance sized to draw a few watts would also do it.

You can actually get the same issue with just enough extension cord, (especially grounded cord) and a single string of LED.

BTW, as I understand it, the forum is primarily for help from other users. If you want corporate assistance without extra delay, you should be submitting a ticket.

[Ken Benedict]

Craig - Try a 47K 1w resistor in parallel with your LED strings.

 

See also: http://forums.lightorama.com/index.php?/topic/11877-lor-units-and-transformers/?p=103169

 

and: http://forums.planetchristmas.com/index.php?/topic/36988-easiest-terminatorssnubbers-to-make/

 

and: http://forums.lightorama.com/index.php?/topic/24329-helpleds-not-fadingdo-i-need-a-snubber/?p=230501

 

and: http://forums.lightorama.com/index.php?/topic/26626-did-gen3-controllers-fix-your-dimming-problems/

 

and: http://forums.lightorama.com/index.php?/topic/25887-solve-this-problem/

 

and: http://forums.lightorama.com/index.php?/topic/23814-huge-fading-problem-please-help/

Edited November 4, 2013 by Ken Benedict

[-klb-]

The C7 is usually faster/easier for testing. I'm not sure that 47K is that much more than what is already in the controller. It should only draw about a third of a watt. But doubling the phantom load just might do it. A 2.5K 20W resistor is probably a closer match to what a single C7 draws, and over rated enough that it can take some thermal insulation in mounting, etc.

[plasmadrive]

Klb, 

 

 

Capacitance in the wire is the only thing that made sense to me as well, but I would not have imagined such a finite threshold if I hadn't seen it with my own eyes.  Litterally 3 or 4 strings of 70 lights makes the difference as to whether or not it works. There was no in between.  There are about 31 strings on that tree.  Makes me wonder what will happen in the rain.  It was the wierdest thing. Even on 5% they still snapped from up to full when I added those extra strings. 

 

I did put a ticket in after I typed this up last night.

 

Ken, 

 

I know about the resistor. I have made many of the dummy loads for my other LED strings.  I doubt that a 4.7K is going to bleed this much charge off, but since I have them already I might give it a try.  I was going to use a bigger load. 

[-klb-]

Are you getting the stage where dimming flickers between roughly the intended brightness and full on? I can get that with as few as 7 bundles of RGBW tightly wrapped on a wire frame mini tree. (The 4 color bundle adds 3 more grounded (neutral) wires in tight proximity, and enhances the effect) Even a single minleon retrofit LED stabalizes that situation, as they act somewhat linear above about 10V. For those, I have not seen any rain issues with 47k loads.

[George Simmons]

Craig - Try a 47K 1w resistor in parallel with your LED strings. http://forums.planetchristmas.com/index.php?/topic/36988-easiest-terminatorssnubbers-to-make/

 

These made all my LED fading problems go away.  I found, like Kevin reports, that the more LED lights on a channel the worse the problem becomes.  The issue, ironically enough, doesn't exist with cheaper, half-wave LED lights.  And on channels with just one string, there never seems to be a problem, regardless of the type of LED.  The terminators/snubbers above in Ken's post have worked perfectly for me.  However, I don't recall having more than eight LED strings on an individual channel, so maybe the trouble mounts significantly when you go farther above that bulb count and you need a beefier dummy load.

[plasmadrive]

Looking at the thing from a logical stand point (there in may be my first problem), The full wave bridge in these light strings should isolate any capacitive charge that accumulates in the string sections.   The charges should not be cumulative over all since they are diode isolated from the feed line.  So why is it either full or zero when the string count gets higher? 

 

I am going to try a few experiments and see if I can't shed some light on this further down the road. 

 

One thing I am going to try is a PWM DC motor drive on this big string just to see if that works for dimming.  I am thinking it will.  No commutation issues. 

 

For now I will try the dummy load and see if that cures the problem in the mean time.

 

BTW, no flickering.. just on/off

[plasmadrive]

OK.. I am flabbergasted.  One single 47K made it work.    There has to be a reason and if I get time I am going to figure it out..  LOL

[-klb-]

Remember that the pass through wires are not rectified. They from plug to socket with no components in path. It is this wiring that contains the capacitance that is causing the impact. The 3rd wire with the LED is not really contributing anything, but the failure to drain off the charge as the channel approaches zero crossing. This charge can't drain back through the triac to the line, as the triggered half of the triac only conducts one way.

[plasmadrive]

Remember that the pass through wires are not rectified. They from plug to socket with no components in path. It is this wiring that contains the capacitance that is causing the impact. The 3rd wire with the LED is not really contributing anything, but the failure to drain off the charge as the channel approaches zero crossing. This charge can't drain back through the triac to the line, as the triggered half of the triac only conducts one way.

Might be the case, but I can't imagine there is that much capacitance in those wires.  Gotta geta scope on it. 

[R Weiland]

Might be the case, but I can't imagine there is that much capacitance in those wires.  Gotta geta scope on it. 

It's not from "Charge"  or "Capacitance" of the wires.

 

The TRIACs in the LOR boxes were designed for incandescent bulbs rated to operate at higher power levels. 

When an LED string is attached, the non-resistive, and low power nature of the LEDs can cause the TRIAC to misfire resulting in flickering, reduced brightness and limited dimming range. The TRIAC misfires because it does not have the required latching and holding current needed to keep it from misfiring. The way to solve the problem is to install a snubber resistance sufficient to provide the necessary current to keep the TRIAC from misfiring. This can be a C7/9 bulb or a resistor as Ken suggested. Bottom line is you want enough current to keep the TRIAC operating within spec.

[-klb-]

Nice theory, but exactly how does it explain that you can have a few strings with just the LED, and no issues, but when you add more, the issue starts, then gets worse as you add more?

If the issue was just low current, wouldn't it be worst with one string, and get better as you add more?

[plasmadrive]

Sorry Guys.. it IS the capacitance of the wire which is holding the charge which is feeding back to the triac which is not allowing it to turn off which is next to the house that Jack built.  Just went thru a very long drawn out ordeal on FB over this very same thing.. (think my family must have put me on ignore since they are not technical.. ha ha).

 

I was not addressing the flicker, but rather why they stay on with more strings.  

[-klb-]

In my experience, the flicker I have seen is between the correct intensity, and full on. So it is the same effect, just not taken far enough to be reliably all on.

The question I have never figured out exactly is that the triacs don't switch off at zero voltage, but rather zero current. And, even if the current half cycle SCR were to stay on past zero crossing, how is it triggering the opposite SCR, for the next half cycle? Is charge leaking back into the conducting SCR, to the shared gate, and this leakage triggering the other half?

I'm not really sure if that is the exact mechanism, or some other characteristic that I have not yet read about, but it is my best guess to date.

[plasmadrive]

In my experience, the flicker I have seen is between the correct intensity, and full on. So it is the same effect, just not taken far enough to be reliably all on.

The question I have never figured out exactly is that the triacs don't switch off at zero voltage, but rather zero current. And, even if the current half cycle SCR were to stay on past zero crossing, how is it triggering the opposite SCR, for the next half cycle? Is charge leaking back into the conducting SCR, to the shared gate, and this leakage triggering the other half?

I'm not really sure if that is the exact mechanism, or some other characteristic that I have not yet read about, but it is my best guess to date.

What happens is the because of the capacitance the current leads the voltage and when the current is zero the voltage is not, therefore allowing the triac to be retriggered with that voltage into the next half cycle.  If these were built with back to back SCRs instead of Triacs, I think you would find the flicker and the full on at dim setting issues would go away.. just a theory. 

 

We always look at triacs as back to back SCRs but in reality they are not because the current goes thru the same junction and they will trigger as a single device with voltage and current in either direction.  They mostly operate as if they were B2B SCRs but not always.  Flicker and erroneous on are cases in point.

[R Weiland]

Nice theory, but exactly how does it explain that you can have a few strings with just the LED, and no issues, but when you add more, the issue starts, then gets worse as you add more?

If the issue was just low current, wouldn't it be worst with one string, and get better as you add more?

 

 Remember the LED is a non-resistive load, In fact it is a non-linear load that distorts the current waveform, this combined with the low current causes the issues. The more LEDs you hang on a TRIAC the more distorted the waveform and hence the triggering. Adding a snubber improves the situation by damping some of the waveform distortion.

[plasmadrive]

It's not from "Charge"  or "Capacitance" of the wires.

 

The TRIACs in the LOR boxes were designed for incandescent bulbs rated to operate at higher power levels. 

When an LED string is attached, the non-resistive, and low power nature of the LEDs can cause the TRIAC to misfire resulting in flickering, reduced brightness and limited dimming range. The TRIAC misfires because it does not have the required latching and holding current needed to keep it from misfiring. The way to solve the problem is to install a snubber resistance sufficient to provide the necessary current to keep the TRIAC from misfiring. This can be a C7/9 bulb or a resistor as Ken suggested. Bottom line is you want enough current to keep the TRIAC operating within spec.

Actually, not so much for the case I was interested in. If you do run too few lights that can cause the flicker, but my OP was about more lights causing the issue.  It has be now proven to be the capacitance of the wire. 

[-klb-]

You listed hold on current as part of the issue. The current drawn never goes down as the number of strings goes up. Plus, I don't think I have ever seen a string flicker lower than what the output would be with some linear load. Only brighter.

With a string or two, the output voltage for 100% on tracks line waveform beautifully. Current looks much like you sliced the middle third out of the waveform. A nice curved positive peak, a flat line, then a nice curved negative peak. And phase angle dimming looks just like you would expect. The leading edges of these waveforms get trimmed off.

When you add more strings, at about 2/3 of the way through the half cycle, you start seeing a deviance. The voltage starts to have an inflection point and a straight line (slow discharge) that does not reach zero by the next half cycle. Current does not change behavior, just the voltage, because you now have more capacitance than the LED strings can drain once below their voltage knee.

It appears to be this stray voltage on the load side of the junction that is causing it to retrigger right after zero crossing, even if the phase angle dimming is not calling for it to be triggered.

The non linear dimming? Remember the current curve being flat line near the ends of the half cycle? That is throwing off the trigger delay to brightness curve.

[-klb-]

Actually, not so much for the case I was interested in. If you do run too few lights that can cause the flicker, but my OP was about more lights causing the issue.  It has be now proven to be the capacitance of the wire.

I've actually never seen a flicker from too few. I have about 80 channels with one string of C6 70 count LED, and never have seen any issues with them.