Cheap LEDs from eBay

[PaulRB]

3 of the LEDs in my 4x4x4 cube have failed already! I only built it 6 weeks ago. I have plenty of spares because I bought 100, but they are fidly to replace. So far the failed LEDs have all been on the outside of the cube, thank goodness.

So if many more go, maybe I should junk the cube and build a new one to attach to my circuit board. But what to buy?

The current LEDs are 5mm blue with a flat top, which gives them a wide viewing angle, important for a cube, and a claimed 9000mcd brightness. They cost less than £3 for 50, from a UK vendor, which seemed like excellent value, but now I have concerns over the quality.

I would not mind paying 3 times as much if I could be sure of higher quality and longer life. Hopefully buying 100 will bring the price per unit down. I would prefer blue, green or even UV to red or yellow.

Any recommendations please?

Paul

 

[geoff07]

Check the drive current isn't too high - that can cause failure. Otherwise, maybe best to soak test them for a week or two to find any infant mortalities. At that price they may be rejects from a fabricator.

 

[Dippy]

I'm not going to poo-poo Ebay for a change as there are many perfectly OK things there I'm lead to believe.

So, speaking for myself, where I really don't want to have to fix or faff around, I only buy things from a long-standing reliable supplier. These people usually have a relationship with the manufacturers and also provide data sheets for you and me to read.

Even for my hobby work I buy small components from Farnell, RS or Rapid etc.
If a supplier can't supply a technical data sheet I tend to look elsewhere.
And just because the vendor is UK based it doesn't mean that they are first-quality items from a pukka manufacturer.

Other people will have other perfectly valid suggestions.
I fully appreciate people are on a budget, but if 50% of things fail that are half-price then you haven't saved much.
Add in p&p , delivery time and fix time (esp if device is awkward) then it's simply a waste.
And with unmarked devices you have to trust that Mr Miggins LEDs Ltd is supplying the right data sheet for the LED you receive...

Searching the BIG suppliers like Farnell, RS, Rapid, Digikey, Mouser etc is a piece of cake - have a go.

All this waffle assumes you have handled, soldered and used the devices properly of course.

 

[Jeremy Harris]

what sort of current limiting are you using and does it always keep every LED If below the maximum specified?

In my experience, the majority of LED failures seem to occur because of current related problems, rather than quality problems. The worst example I've seen was some multi-LED lights I bought (from ebay........) that suffered a few failures fairly quickly. When I did an autopsy on one of them I found that the design was fundamentally flawed, in that several series LED strings were connected in parallel and then fed via a single current limiting resistor. Although the resistor should have been able to keep the current within limits if each of the series strings had exactly the same total Vf, in practice the Vf of each string would be both different (because of manufacturing tolerances) and vary disproportionately with temperature (because the LEDs would inevitably run at differing temperatures across the array). The net result was that one string would hog too much current, causing failure of one or more LEDs, which would then blow and cause a cascade failure in the adjacent strings, as they'd then be running at too much current.

Even now I still see LED circuits on the web where people have connected LEDs in parallel, which seems to show that there is still a lot of misunderstanding about the way they work..

 

[PaulRB]

Thanks Geoff. Not sure how I can check the drive current because the cube is multiplexed by a max7219 driver chip. One resistor sets the current for all LEDs. The minimum value of that resistor is 9.5K and I used 10K.

So you think the vendor may be buying reject batches from the manufacturer? Presumably the if the manufacturer finds an unacceptable numbers of failures in a batch, they sell them off cheap as "unbranded"? Then what, does the vendor quickly test each individual LED and discard the failures before packaging and selling the "good" ones?

 

[Dippy]

"So you think the vendor may be buying reject batches from the manufacturer? "
- Goeff won't know the answer (unless he's the vendor )
He's simply putting forward a theory. It may be wrong or right.

Being a generous person, I doubt if the vendor tests all of them - he may randomly select a few from a bucket load but I doubt that.
Most electronics is statistics and, guessing, I suspect some vendors rely on customers not being arsed to complain esp when a device so cheap that a Postage Stamp nullifies the saving.

This is why I use Farnell etc.
I spend a little more, it works , no worries, end of.
I don't think I've ever had to consult a Forum after buying a component from a proper distributor.
(All my magic smoke situations have been my own fault).

 

[PaulRB]

Thanks everyone. Can't find what I need on Rapid, but will try the other suggested suppliers.

Paul

 

[hippy]

Not sure how I can check the drive current because the cube is multiplexed by a max7219 driver chip. One resistor sets the current for all LEDs. The minimum value of that resistor is 9.5K and I used 10K.

From the datasheet; "The peak current sourced from the segment drivers is nominally 100 times the current entering ISET ... Its minimum value should be 9.53kΩ, which typically sets the segment current at 40mA". So 10K would still be pretty close to that in my guesstimation.

What the LED's are rated for is the big question; can they sustain anywhere near to 40mA ? Maybe put one on destructive testing to see when it does go "pop".

To measure current when multiplexed you probably have to fall back on the old trick of measuring voltage across a known resistance value and applying I=V/R.

 

[Pongo]

Since the led's are most likely not mounted on a pcb or some other way that would secure the leads and prevent stress on plastic package during soldering I would also consider physical damage as a possible reason for the failures.

 

[PaulRB]

Thanks Hippy & Pongo.

There was no data sheet for the LEDs, as you can imagine, but Vff quoted as 3.2-3.8V. Imax quoted as 30mA, but that's continuous. These LEDs are 1:8 multiplexed, and many led data sheets quote a peak current of around 5 times the continuous max, with a 1:10 ratio.

Soldering and stress damage while bending the legs are certainly possible reasons too.

I'll keep an eye out for further failures and be less surprised in future.

 

[geoff07]

Indeed I have no knowledge of where the leds are sourced. But if they cost 1/3 of the price of Farnell leds, or less, for example, what do you think? But it isn't a problem if you soak test them before use and don't over-drive them. Components can be out-of-spec and rejected for many reasons, not all of them suggesting early failure.

 

[Jakob2803]

Hmm 50 pieces for three quid. That wouldn't be from LEDlightsdepot or something like that? I've bought from there before with no problems in the LEDs. He did send me a wrong item once but sent the right one free of charge.

 

[boriz]

Simple rule of thumb. If you want it to last, never drive it a max spec. 20mA should be fine.