More Xmas Lights

[Dave_H]

Just thought to start another thread about LED Christmas lights. I use
them througout the year as single-colour low-level light sources, putting
away the festive multi-colour strings sometime after New Years.

I found some decent cheap ($2-$3) 50-strings of red, blue and white
"rice" lights. They're fine, just indoor and maybe old stock, house-branded.
My inline full-wave rectifier worked on the red string, but only half of
the blue and white strings lit, indicating two sub-strings on opposite
half-cycles. This different configuration must be related to higher forward
voltage drop on blue and white LEDs. This limits the use of full-wave
brightening/de-flickering to certain colours/lengths of strings.

Another topic: since introduction of LED Xmas lights I've had several cases
across different vendors of blue LEDs in particular dying before their time.
Individual lights go dim or into a weird flickering and then dark, fortunately
the rest of the string carries on. I gave my brother a few multi-colour
outdoor strings a few years back, and all the blues are gone but all others
are OK. With cold/dry winters here I attribute this to static electricity.
Checking specs for blue LEDs I found warnings that they are static-
sensitive and require special handling. By the look of it, string vendors
didn't account for this. Maybe the problem has been fixed, or are people
seeing this with recent product?

Actually, a string on newer "emerald" greens suffered similar fate, but the
older greens, which are a dimmer yellow-green, seem immune.

Dave

 

[JohnR66]

Indium gallium nitride LEDs which are the true green, aqua, blue, UV and white LEDs are static sensitive. Unless the light string is in contact with a static producing source, It is probably the line voltage itself that is killing them. The power supply is notoriously noisy with spikes from disconnecting inductive loads that dumps large voltage transients on the line and can zap the LEDs.

The better designed strings are recified with a filter capacitor that absorbs the spikes. Also never remove or insert LEDs from the string while energized. The sudden spike produced can zap a LED. I've done that a few times.

 

[Dave_H]

Indium gallium nitride LEDs which are the true green, aqua, blue, UV and white LEDs are static sensitive. Unless the light string is in contact with a static producing source, It is probably the line voltage itself that is killing them. The power supply is notoriously noisy with spikes from disconnecting inductive loads that dumps large voltage transients on the line and can zap the LEDs.

The better designed strings are recified with a filter capacitor that absorbs the spikes. Also never remove or insert LEDs from the string while energized. The sudden spike produced can zap a LED. I've done that a few times.

That makes sense, except I've never had problems with white, even
early product; and aren't white LEDs inherently blue or blue-green
under that phosphor coating?

One set of failures happened on an old metal floor lamp,
where I'd taken out the regular bulbs and wrapped the shade
with a few strings of different colours. Looked neat, but touching
the metal support (with the ac wire unside) with static discharge
must have induced a spike in the wires. This wasn't
the only failure scenario though.

Newer strings might be more resistant though I can't see that
components at the plug end will protect all LEDs from static
which has a very fast risetime, and with the inductance of all
that wire...

Incidentally, what is the mechanism to keep the string lit
with multiple bad (open or high-resistance short) LEDs? I figure
it's a shunt zener of ~4v integrated into each LED.

Dave

 

[JohnR66]

That makes sense, except I've never had problems with white, even
early product; and aren't white LEDs inherently blue or blue-green
under that phosphor coating?

Yes, it is typically a blue LED. Can't say why if they were the same brand/quality.

One set of failures happened on an old metal floor lamp,
where I'd taken out the regular bulbs and wrapped the shade
with a few strings of different colours. Looked neat, but touching
the metal support (with the ac wire unside) with static discharge
must have induced a spike in the wires. This wasn't
the only failure scenario though.

Newer strings might be more resistant though I can't see that
components at the plug end will protect all LEDs from static
which has a very fast risetime, and with the inductance of all
that wire...

The faster the rise time the better for capacitive filtering. It all depends on the energy in the spike. Too much energy and the cap charges up to too high of voltage. Normal static build up on people should be no problem, but the protection is of no help if the discharge occurs around the LEDs themselves. Each LED would require its own protection cap or diode (usually a diode at LED level).

Incidentally, what is the mechanism to keep the string lit
with multiple bad (open or high-resistance short) LEDs? I figure
it's a shunt zener of ~4v integrated into each LED.

Cheap LEDs don't contain any other components. Typically a non lit LED has a "dark current" conducting channel in the die that keeps the voltage just low enough so the rest of the die does not light. Sometimes these bad LEDs can be brought back to life by giving them a shot of high current. This opens the conductive channel and the rest of the die will light up again usually at something less than full original brightness. OTOH, the manufacturer may be doing something. I have never seen this though. I hacked many a string of lights for the LEDs and never seen anything but an ordinary LED.

 

[brickbat]

...Incidentally, what is the mechanism to keep the string lit
with multiple bad (open or high-resistance short) LEDs? I figure
it's a shunt zener of ~4v integrated into each LED.

A zener would work, though I've never seen it used - too expensive, I'd guess. I've hacked some 2008 GE Soft White strings, and they used a series-parallel arrangement. Each LED was in parallel with a 2nd LED. So if one of them opened, the string would still light. Now, of course the issue is that the other LED of the parallel pair would see twice the current. But, given the simplicity, and the high reliability of these LEDs, I think it's a very good solution. I've had a couple of strings running 24/7 since last January, and not a single LED has failed. They're still bright, too. All in all, those GE strings were top-notch. Full wave rectification, too...

 

[JohnR66]

A zener would work, though I've never seen it used - too expensive, I'd guess. I've hacked some 2008 GE Soft White strings, and they used a series-parallel arrangement. Each LED was in parallel with a 2nd LED. So if one of them opened, the string would still light. Now, of course the issue is that the other LED of the parallel pair would see twice the current. But, given the simplicity, and the high reliability of these LEDs, I think it's a very good solution. I've had a couple of strings running 24/7 since last January, and not a single LED has failed. They're still bright, too. All in all, those GE strings were top-notch. Full wave rectification, too...

Same here. I hacked a 2007 and 2008 GE string for the warm white 3mm LEDs. Funny thing, The LEDs used in the 2007 were of better quality. I tested them at 30ma for 192 hours and the LEDs from 2007 did not fade as much. However, both of the 3mm LEDs were better than a 5mm WW LED I got out of a generic string. These faded severely. Dig through my posts in the LED forum from a year ago to see the results with beam pictures.

IIRC, the GE string were full wave rectified for 120Hz flicker rather than 60Hz. They seem to be the better quality strings out there in the B&M shops.

 

[Dave_H]

.

Cheap LEDs don't contain any other components. Typically a non lit LED has a "dark current" conducting channel in the die that keeps the voltage just low enough so the rest of the die does not light. Sometimes these bad LEDs can be brought back to life by giving them a shot of high current. This opens the conductive channel and the rest of the die will light up again usually at something less than full original brightness. OTOH, the manufacturer may be doing something. I have never seen this though. I hacked many a string of lights for the LEDs and never seen anything but an ordinary LED.

Interesting. I didn't think discrete protection would be wired per
bulb, too expensive, but something internal to each LED as you
describe. This sounds like an "antifuse", something which creates
a low-resistance path when it blows, sort of what was used in
early FPGAs (made by TI etc.).

I haven't hacked any strings for LEDs yet, tend to use them as-
is. What do people typically use the LEDs for, individual LEDs or
just mod'ing the string?

Dave

 

[Dave_H]

Anyway, slightly different topic, I picked up a LED "ornament",
big plastic frame with 50 red/green LEDs arranged in a pattern.
Looks like it's full-wave rectified by the pod inline. Very bright,
looks nice, no need to take it apart .

Odd thing, I tried this on an AC remote control to see if it
could be dimmed (full power is 3W). As mentioned in another
thread, the remote module won't stay off, something related
to load-sensing. This doesn't happen with most "regular"
strings. I wonder what would make this one different.

Dave

Update: dimming works, down to ~zero which is essentially
off. Steps are a bit coarse though. Problem solved.

 

[Dave_H]

Looks like I'm not done with the blue LED failure thing.

I stretched out a string of blue<>yellow colour-shifters on
the carpet (which is very short and not synthetic) and the end
lamp is now shifting between yellow and nothing. Seems
to be no coincidence the end lamp is where it was handled.
All others are OK so far. Guess I should stick to red<>green.

Dave