What's your WS2812B failure rate? My observations.

T Chap

I designed some boards with this device, followed all the guidelines and have about a 10% failure rate. The individual chips came from Adafruit. I wrote them and they only say they will replace them, and have no idea why they die or what the failure fate is. Some are DOA on first power up after assembly. This means one of several symptoms:

The chip may have one or two LED's that work, but one LED is dead(the dead colors vary), and the chip does still transmit out to the next LED in the chain and the next LED works.

The chip is just dead, no LED, no output to the next LED.

The chip works for a period of time, then dies and all others after it dies.

I have done an awful lot of SMT boards, and have spent far too much time chasing this problem compare to most bugs and cannot find a logical explanation. If you Google WS1812B failure, there are others that claim to have problems of a % failure rate. I have only been using these chips for a few weeks or so on two different boards, and am fast regretting this decision since I have failures on both. Originally I was testing another popular 16 channel PWM chip for LED control and it tested fine, but I was only testing a few LED's to get the code and driver tested. In the first boards I built a few weeks back with the WS2812b, some were DOA, and some died after some time. I modified the board, added the series resister they suggest on the first LED, now on the second revision board am experiencing the same problems. In some rare cases, I have tried powering of and on a bunch of times to see if I can actually blow up some chips with the power on sequence(they suggest power up GND>+5>Micro in that order), and strangely there are times that the dead colors come on just from power up, even though the dead colors will not turn on when you command them to, yet other colors do turn on on the same chip via commands. It's like the LED is alive, but not getting told to turn on internally. I can't see anything obvious on a scope starting from LED 1 to 50.

I would love to hear from anyone that has used this chip in production. As it stands with my testing, this chips seems like it would cost tons in warranty and service on a real product and would make for a lot of unhappy customers, since there may be no end to the repairs of these LED's. Maybe it is best as a hobby LED. The problem with daisy chain signals is that your product is potentially useless if there is a failure that kills all LED's down stream, but a discrete driver can at least keep the rest of the product afloat if one LED goes out. I would much prefer someone say that they have these working perfectly over long periods of time and get some encouragement that I need to revisit the design for hardware bugs.

Publison

I see you have participated in the threads I was going to link to:

http://forums.parallax.com/showthread.php/158590-WS2812B-and-series-resistor?highlight=WS2812B

http://forums.parallax.com/showthread.php/158429-WS2812B-Propeller-Issue?highlight=WS2812B


I have yet to solder up my freebies and participate. One more day maybe.

erco

Publison wrote: »

One more day maybe.

Surely tomorrow's the day, Jim! Will expect a full writeup by 6pm.




EDT.

Duane Degn

T Chap wrote: »

some were DOA, and some died after some time.

When you say "DOA", this is after they were soldered right?

Could your oven be too hot? (I'm sure you've considered this.)

I'm pretty sure JonnyMac has done multiple WS2812B boards and he has mentioned he has never had a failure.

So far the boards I've made have been hand soldered. The only times I've had failures was when I abused the chips when attempting to power then using a DIY slip ring.

The three 16 pixel rings I purchased from Adafruit a year and a half ago all still work and I've used them a lot as active feedback with encoder knobs. I just remembered the Adafruit rings aren't the "B" variety.

Are your boards lead free? If so maybe the higher temperature needed to melt the solder is causing a problem? (Just making a wild guess.)

T Chap

DOA means right out of the oven, apply power send a signal and it is dead. I don't use unleaded and use the same oven profile I always use. I have thought a lot about heat, but even using care with an air pencil I have had them die.

WBA Consulting

Happened to catch this thread while waiting for my lunch order and although digging through the datasheet isn't easy on my phone, I would recommend looking to see if the datasheet has any specifications for min/max rising or cooling slopes through reflow. This comes from first hand experience with some 3 watt RGB LEDs in the first design of the video wall for the Armani store in New York. We had a fairly high fallout in the first run and the LED manufacturer told us it must be our profile. After running our KIC profiler, we found we were close to the max slopes for heating but not over. This spawned more questions to the supplier and over the course of the week, the actual cause turned out to be an incorrect die bonding adhesive used to manufacture the LEDs which was only spec'd to 220 degrees C (SAC305 lead free solder melts at 217) so obviously the LEDs weren't capable of surviving lead free assembly. Needless to say the LED manufacturer covered all costs to that point and our customer redesigned the board with a different LED.

Sorry for the long story, but my point is check your profile and the part datasheet for any discrepancies.

T Chap

Peak/Classification Temperature 215C I am hitting this too hard in the oven based on their Lead profile, but their no Lead says I am fine on temp as far as not exceeding max values.

Keep the product before using it in the oven and bake 65~70 degrees 24 house (interpreted to mean 24hours, C or F??)
After coming out from the oven within 2 hours immediately complety; ( assumed to mean solder the boards within 2 hours of the 24 hour bake)
Use of the product is not finished in time back in the oven; (possibly means if you don't reflow within 2 hours of the 24hour bake, then bake again)
When the shift, the patch and then completed the furnace had finished work, there is no patch back into the oven in time (no clue)

The datasheet is all in Celcius, would anyone assume that the 65-70degree bake at 24hours would be F since all other values are shown in C? I most certainly don't do a 24 hour bake.

I will try to bake the chips at 70 Celcius and see if there is improvement. Possibly moisture is killing it.

Cluso99

FYI: LEDs have always been fairly susceptible to failures when soldering temperatures were too high or too long.

WBA Consulting

That's essentially an MSL 6 rating, "bake before use" which would follow suit with the failure rate you are seeing. As you mentioned, bake at 70C for 24 hours and try a few.

Were they sealed in a moisture barrier bag with dessicant and a humidity indicator card when you received them? They should have been. A moisture barrier bag looks like a thicker ESD bag and will be vacuum sealed.

I agree with your interpretation of their verbiage including the no clue for the last sentence. My only guess is they mean baking is not required for any process after reflow.

T Chap

Adafruit ships with the cut tape parts in non sealed bags with no moisture indicators or silica. I will test some tomorrow that are baked for a few hours and check the results. Part of this post is to find out the viability of this LED for production and to see what others have experienced.

Publison

erco wrote: »

Surely tomorrow's the day, Jim! Will expect a full writeup by 6pm.




EDT.

Not to pollute this thread about manufacturing, see:

http://forums.parallax.com/showthread.php/157643-WS2812B-Fun-Board/page2?highlight=WS2812B

Done 12 hours early.

ziptai

T Chap,

Do you have any details of your results you can share? Did you see a marked difference after baking before soldering?

Thanks,
Bill

T Chap

I don't have real test results, but the reality is that without baking you get a high rate of failures(assuming you did not receive the parts in a seal bag). It is rather easy to know if there is too much moisture, just put hot air on it for a few minutes and you will hear it pop if there is steam inside. I only ran a few batches, but after baking the problem seems to have went away. I will be back on this endeavor first of the year and will have more experience with reflow results then.

MikeDYur

Does anyone know if these knockoffs have Propeller potential. It's a serial interface I know, voltage and logic levels I think would have to be addressed, so how else could it be specific to Ardwino.


http://www.ebay.com/itm/WS2812-5050-RGB-LED-Module-Chainable-Board-Arduino-/272048601847?hash=item3f575c3af7

mindrobots

I have some that are similar that I've used on the Propeller. If I recall, the 3v3 output on the signal pins drove them. You power the string itself independently, so that's not an issue.

They've gotten ridiculously cheap! (I'll reserve "inexpensive" until you prove they work and last past the unboxing.)

Duane Degn

MikeDYur wrote: »

Does anyone know if these knockoffs have Propeller potential.

I don't see any reason why those would be different than the other WS2812B LEDs.

I think it's odd Vdd is labelled 4-7VDC. I thought the upper limit was closer to 5V and the lower limit is at least as low as 3.3V.

These LEDs don't always behave well when powered with 5V and 3.3V logic is used. The LEDs work fine powered with 3.3V.

Buck Rogers

Nearly all of my devices, two freebies, (don't ask!) and a batch of similar ones from Limor all worked. I say "nearly" because I've got a crowd that I'm yet to use.

Of course its natural that the Prop would certainly be considered for that planned but yet to be accomplished project.

Wow this thread has a long life time.......
---
Erco? What's your wandering robot doing in the diner where Andrew ordered his lunch nearly a year earlier? I thought it was still in Publison's garage playing poker. And it was last seen in Oz trying to arm wrestle a kangaroo, while six koalas watched. Or even in Scandinavia skiing.

MikeDYur

Thank you for your insights. There's a fine line on what you put together for a child, buy inferior components and have them fail, the toy ends up on the bottom of the toybox. Or spend the money and buy top quality, to have the toy end up on the bottom of the box anyway.