36 leds in series through usb ? how is that possible ?

[tooki]

But in LED lighting they’re often driven hard, 30 or even 40mA.

I looked at the datasheets in Google for a few 5mm white LEDs. 30mA absolute maximum (not enclosed but in in a cool ambient) and 20mA is typical for their brightness rating.
I think 40mA would burn out a white LED out soon.

Which is precisely what happens in a lot of cheap Chinesium products! :p

But also, a lot of the LEDs do tolerate it, even if it is out of spec, with reduced lifetime that isn’t short enough to be alarming. (The absolute max ratings are just the maximum the manufacturer guarantees to not cause damage, right?) The SMD versions are a bit more tolerant of overcurrent, in that they can sink a bit more heat than the THT versions.
 
But I did first say to assume 20-30mA.

[respoda]

USB has 4 wires. Some of them require resistors across the two extra wires of the jack to tell it how much current to produce.
Maybe your USB before always provided high current for high brightness but now maybe you are using a different USB that provides low current for low brightness.

I didn't know that about the extra wires (white and green).
Should be fun dabbling with that...

Though I'm pretty sure it's just the red and black being used in this lamp...

[respoda]

Are all of the LEDs lit at all?  Are any of them "out", dead?  If a single one fails short circuit the whole setup is fried and needs to be put in the bin before you fry the device powering it.

It's been running for many hours, and none of them are dead or out, they all stay lit but very dim...
It makes me believe they're just not getting enough power, but it's just a hunch, I'm guessing here.

In fact putting so many cheap LEDs in parallel is asking for trouble, even if they have current limiting resistors which they will need or one single LED will take most of the current and explode showering little bits of burning plastic into your face if you are unlucky.

I'd bin it and move on.

In fact I already switched the 36-led pcb with another 3-led pcb that already had a resistor and capacitor built-in,
tested it and works a treat.

I still have the 36-led pcb, still planning to figure out what made/makes it tick (correctly)...wishful thinking perhaps.

[respoda]

Assuming they’re normal (not high-power) white LEDs, assume 20-30mA per LED. (When loads are in parallel, you add their current.) 36*20 = 720mA, 36*30 = 1080mA.

Many cheaper USB chargers — as well as the standard USB ports on computers — provide only 500mA.

When lit, what voltage do you measure across an LED?

It was 3.3xx something, but I'm using a cheap meter, so it could be a tad off. Let's say between 3.2 and 3.4v.
That was before it got ripped apart. Haven't measured afterwards (my mistake), I just switched to another 3-led setup for the intended purpose.

[respoda]

Don't forget about wire resistance. Thin wires act as a current limiting resistor. Led's don't need 20 mA to shine brightly. Even 10mA per diode is enough to shine brightly, as modern led's are very efficient, even 1mA is sometimes enough to use as a indicator.

As you know the wires in a usb cable are very,...very thin.

[ucanel]

That is why do not wire the leds parallel,
every individual led or series led group
should be used with series resistors
before wiring it with another led or led group.

I guess one of the led in that circuit
had a lower operating voltage,
if you detect and get rid of that one
rest of the leds will start to become brighter again.
The suspected one would be the brightest one,
I don't know may be the dimmest one.

Also:
When a led gets hotter
it's resistance gets lower,
means it draws more current,
if it draws more current
it gets even hotter,
if it gets even hotter
it draws even more current
it is getting nowhere
except killing itself.

So if there is not any resistor for
limiting the current for each of the leds
this problem eventually happens.

You may also search for the hottest led in your circuit
to find the guilty one .

[saike]

I was asked to look at a bunch of large 250Watt (12cm dia. x 20cm long) led lamps recently that had been used outside in the rain and had stopped working. Inside they had a fairly comprehensive step down voltage converter and current limit circuit which showed obvious signs of water damage.
The surprising thing was, there was no sign of any resistors on any of the leds, they just appeared to be connected in series/parallel combinations to work somewhere around 48Volts . I removed the controller and tried  the leds on a 48 volt power supply, trimmed down to 44 volts and the current draw was around 4.5 amps. The rest of the job was done like this with all the lamps running from whatever 40-48V power supplies we could get in a hurry. None of the leds failed.

[tooki]

Don't forget about wire resistance. Thin wires act as a current limiting resistor. Led's don't need 20 mA to shine brightly. Even 10mA per diode is enough to shine brightly, as modern led's are very efficient, even 1mA is sometimes enough to use as a indicator.

As you know the wires in a usb cable are very,...very thin.

The wires in cheap USB cables are very thin. In proper ones, the ground and V+ conductors are nice and thick.

[tooki]

Assuming they’re normal (not high-power) white LEDs, assume 20-30mA per LED. (When loads are in parallel, you add their current.) 36*20 = 720mA, 36*30 = 1080mA.

Many cheaper USB chargers — as well as the standard USB ports on computers — provide only 500mA.

When lit, what voltage do you measure across an LED?

It was 3.3xx something, but I'm using a cheap meter, so it could be a tad off. Let's say between 3.2 and 3.4v.
That was before it got ripped apart. Haven't measured afterwards (my mistake), I just switched to another 3-led setup for the intended purpose.

3.3V sounds typical for a white LED. So those are normal white LEDs with no built-in series resistors.

(By the way, your cheap meter is almost certainly accurate. Even the $2 DMMs usually agree within a few counts of an expensive meter. It’s features, reliability, and safety for mains use that you pay for in a better meter.)

[tooki]

That is why do not wire the leds parallel,

[...]

You may also search for the hottest led in your circuit
to find the guilty one .

Correct.

That said: Please do not break your text into such short lines, it makes it harder to read. (And it looks like a poem, LOL!) Just separate your paragraphs.

[ucanel]

@saike As mentioned, if you can get the chemistry right, you can use leds paralel.
The manufacturing process must be different from many cell power leds vs individual 5mm leds.
And power led you described has very good large heat sink.

@tooki thank you for your warning but it is my habit, I am writing in every forum like that.
It looks more readable to me. If I get more wishes like that I may write as everyone wanted.
And wiriting a poem in English, I couldn't dare even in my mother tongue

[respoda]

That is why do not wire the leds parallel,

[...]

You may also search for the hottest led in your circuit
to find the guilty one .

Correct.

That said: Please do not break your text into such short lines, it makes it harder to read. (And it looks like a poem, LOL!) Just separate your paragraphs.

That was my first impression too

There was one led that seemed to flicker a bit after I took it apart and 'remodeled' it,
I thought it was just because of the lack of a resistor.
Could be it's the cluprit causing them all to stay dim...
I'll yank it out and see if it makes a difference.

[respoda]

...
And wiriting a poem in English, I couldn't dare even in my mother tongue

In fact I am kind of an amateur poet, I even had some poems released a while ago through poetry.com.

Your comment does have the structure for a decent poem. Got all the paragraphs neat and tight :p