Author Topic: MSL parts storage and care for the hobbyist/small business (Read 7482 times)

funbags · « **on:** July 11, 2018, 02:47:27 pm »

Hey All,
Lately I've been fighting popcorning of some 5050 LEDs. I see about a 20% fail rate per panel of 72 LEDs. I ordered them from China and they don't come in moisture protective packaging, unlike my MSL parts shipped from US suppliers, and I dont know how long the LED's were in the warehouse conditions. At least popcorning is the only thing I can think of, with that kind of fail rate.

Profile / solder:
Hold @ 225C (Max 230C) for 50 seconds in a Controleo 3 BD toaster oven. Chipquick SAC305 solder paste.

WS2812B 5050 LED on a reel, specifically. (they have an IC built into the chip)
Data sheet doesn't cover MSL.
https://html.datasheetq.com/pdf-html/191078/ETC/5page/WS2812B.html?lang=en

I was thinking about getting one of those Ziplock tupperware boxes and a large desiccant device for storage, but I'm not sure how to bake them, and I think they NEED to be baked. I believe 40c is the max temp because of the reel, but any recs on how long? I found a lot of info for IC's but not much for large LEDs. The supplier is no help (I don't speak Mandarin).

Thanks for any info!

(additional search on topic)
https://www.eevblog.com/forum/projects/ws2812-reflow-soldering-failure-rate/

elecman14 · « **Reply #1 on:** July 11, 2018, 04:11:57 pm »

I would look at J-STD-033 section 7 for bake time. Based on a super quick google search for "WS2812B MSL" it looks like they are a MSL level of 5A.

I would get humidity indicator cards if you are going to store them in tupperware. Not sure how well that will work or how often you will need to re-bake. descant in a moisture barrier bag is the industry standard way to go. Those also have shelf life issues as well.

funbags · « **Reply #2 on:** July 11, 2018, 05:04:32 pm »

Quote from: elecman14 on July 11, 2018, 04:11:57 pm

I would look at J-STD-033 section 7 for bake time. Based on a super quick google search for "WS2812B MSL" it looks like they are a MSL level of 5A.

I would get humidity indicator cards if you are going to store them in tupperware. Not sure how well that will work or how often you will need to re-bake. descant in a moisture barrier bag is the industry standard way to go. Those also have shelf life issues as well.

Thanks for the help. You know, I searched for a while and I couldn't find anything. I used your simple search terms and found everything I need. Thanks!

funbags · « **Reply #3 on:** July 11, 2018, 05:46:41 pm »

For posterity I found this in an updated data sheet for the WS2812b. Basically, you should treat them as MSL 6 and bake before use no matter what.

4. Dehumidification process:
LED products which is beyond the prescribed time limit,or for getting damp for some other
reasons,we recommend the customer first do the dehumidification process,then use
them.Dehumidification method:under 75 ℃ / 22 ±2 hours

Cheers,

jmelson · « **Reply #4 on:** July 11, 2018, 10:31:53 pm »

Quote from: funbags on July 11, 2018, 02:47:27 pm

Hey All,
Lately I've been fighting popcorning of some 5050 LEDs. I see about a 20% fail rate per panel of 72 LEDs. I ordered them from China and they don't come in moisture protective packaging, unlike my MSL parts shipped from US suppliers, and I dont know how long the LED's were in the warehouse conditions. At least popcorning is the only thing I can think of, with that kind of fail rate.

I'm not doing this on components yet, but I did have bad blistering on some blank boards that had been in stock for a long time. So, now any boards that are more than a month or two old go in the oven and are baked for something like 1/2 hour at 50C, 1 hour at 60 C, and 2 hours at 75 C. Then, I have zero problems when reflow soldering them. I suspect a similar bake cycle will prevent your problem.

Jon

Kean · « **Reply #5 on:** July 12, 2018, 12:26:47 pm »

About 4 years ago I did a production run with 25,000 of the WS2812S (the 6 pin version) and we had a high failure rate, despite seeing no issues during prototyping or small scale production in Australia. We went back and forth between WorldSemi and our PCB & final assemblers in China, and everything initially pointed to moisture damage due to the rather poor MSL of these devices. So we had another batch of 2,000 LEDs put through production, this time with prior baking and a lower temperature reflow profile. Same issues were found with this batch.
So I hopped on a plane to Shenzen to meet with WorldSemi and the assemblers, and review all production processes. The actual issue turned out to be ESD. This was partly due to a recent change by WorldSemi to shrink the die in their LEDs, but mostly poor handling during the final assembly. I've had a lot of people come to me with LED strip with failed WS2812's, usually the first one on the strip and almost certainly damaged by ESD.

Siwastaja · « **Reply #6 on:** July 12, 2018, 12:40:14 pm »

If you do some googling, you'll find that everyone's having hard time using WS2812 and similar in production. Failure rates typically around a few %, but up to 20-30% of LEDs are widely reported, and the total lack of "they are working just fine for me" type of responses, which we usually see in similar discussions, is quite revealing.

Some report that certain counterfeits perform much more reliably than the genuine ones directly sourced from Worldsemi. Even though, this is quite an expensive experiment to make in production.

The obvious answer is: just don't use them.

funbags · « **Reply #7 on:** July 12, 2018, 02:07:02 pm »

Thanks everyone.
Unfortunately I'm locked into these crap LEDs for the short term. I started a bake of 100 LED's last night (75C for 24 hours) in prep for working late tonight.

I'll populate some panels and take extra ESD precautions and let you all know how I fared. I wont change my paste (SAC305) or reflow profile. I want to see if a ridiculously long pre-bake changes anything. Even halving my current fail rate of 15% would be plus.

Thanks!

funbags · « **Reply #8 on:** July 12, 2018, 03:43:04 pm »

Also, I tracked down a reel of the SK6812 clones. Going to try those in the next run. I understand they have better thermal characteristics and better light emitting characteristics.
Cheers

funbags · « **Reply #9 on:** July 13, 2018, 02:13:12 pm »

Baking the WS2812B for 24 hours at 75C seems to have worked. All LEDs on a panel of 5X16 LEDS worked (well two end LEDs on one bar seemed to flicker/ripple but I cant figure out why). I replaced them, their caps, and the data line resistor and still they flicker. I suspect its a mnfg defect on the PCB.. So into the fail pile it goes.

Treat ws2812b LEDs as MSL 6 and you can limit popcorning.

Cheers,

Siwastaja · « **Reply #10 on:** July 13, 2018, 02:36:57 pm »

If you google a bit more, you'll find out that the flickering is just a foreboding of them totally failing soon.

You could have found out by a bit of googling that the issue is not about the moisture, and now that you have confirmed it by yourself, calling another failure a "success" won't change the fact that it's a failure and you can't use/sell those either. Making optimistic assumptions that problems with knowingly bad products are problems with your PCB is stupid, because such PCB issues are very rare. Your PCB is most likely fine...

Using a nonfunctional product "because you don't have anything else" isn't a strong argument. You need to admit you don't have a product. You need to start from the beginning and design one. Believing in an easy miracle product, even after you have confirmed what everybody already knows - that it's a bait-and-switch unusable crap - doesn't help you. You still don't have a working product!

The more you waste time, the harder it will be. I have been through this so many times. The next step after optimism, belief and self-blame ("it must have been something I did") will be blameshifting, which is equally unhelpful.

But everything starts from accepting the reality. Then you decide if you give up completely, or do the redesign. The quicker you get at it, the better. Stop wasting time now.

funbags · « **Reply #11 on:** July 13, 2018, 04:05:44 pm »

I replaced all the components associated with the flickering LEDs and the flickering persisted. I considered the flickering a fail. I don't understand the point of that comment. It just seems unnecessarily aggressive. I took your advice tried something different, it seems to have worked fairly well, though not perfectly. As a precaution I also tracked down a reel of the clones to try out (as you suggested). I'm not ignoring your advice if that's what you think. I'm taking it all into consideration and acting on most of it. I can't think of a higher compliment.
Cheers,

Siwastaja · « **Reply #12 on:** July 13, 2018, 06:31:38 pm »

Didn't mean to be aggressive, sorry. I don't think you are ignoring my advice, just want to be as clear and straightforward about it as possible. Personally, I have just went through a wasted half year of development time and competitive edge while playing "let's use this faulty part for now" - "but it doesn't work at all!" - "let's still use it, we have to!" games, which finally resulted in them giving up and letting me start doing my job, but it could have happened so much sooner.

julianhigginson · « **Reply #13 on:** July 14, 2018, 02:14:09 am »

What do people here use for MSL part baking when they do their own assembly?

Do you have an actual dry cabinet that also does humidity extraction, like you see in assembly houses?
Or do you just whack your stuff into a toaster oven? (or something in between?)

xaxaxa · « **Reply #14 on:** July 14, 2018, 08:45:37 am »

Quote from: funbags on July 12, 2018, 02:07:02 pm

I'll populate some panels and take extra ESD precautions and let you all know how I fared. I wont change my paste (SAC305) or reflow profile. I want to see if a ridiculously long pre-bake changes anything. Even halving my current fail rate of 15% would be plus.

I would recommend also trying with leaded solder and lower temperature; I recall seeing a comment somewhere that crap LEDs tend to use bonding compounds that are only rated to 220*C which is too low for SAC305 reflow profiles. Personally I don't ever do lead-free because reliability is important to me.

funbags · « **Reply #15 on:** July 16, 2018, 02:14:11 pm »

Quick update: I've now made 3 panels with zero dead LEDs. Baking seemed to do the trick, but I'm also going to look into Leaded solder (mentioned below) and also try the clones (still in the mail)

Quote from: Siwastaja on July 13, 2018, 06:31:38 pm

Didn't mean to be aggressive, sorry. I don't think you are ignoring my advice, just want to be as clear and straightforward about it as possible. Personally, I have just went through a wasted half year of development time and competitive edge while playing "let's use this faulty part for now" - "but it doesn't work at all!" - "let's still use it, we have to!" games, which finally resulted in them giving up and letting me start doing my job, but it could have happened so much sooner.

Thanks Siwastaja. That's totally understandable. And I appreciate the input. I would be frustrated too. I'm glad they at least eventually made the right decision. Funny story, but in the animation industry we have the opposite problem. Where our leads/sups keep changing their minds so often, we run out of bid days to complete the artwork. Sometimes we wish, even if it was the wrong direction, they'd just stick to it. haha. Cheers,

Quote from: julianhigginson on July 14, 2018, 02:14:09 am

What do people here use for MSL part baking when they do their own assembly?

Do you have an actual dry cabinet that also does humidity extraction, like you see in assembly houses?
Or do you just whack your stuff into a toaster oven? (or something in between?)

Hey julianhigginson,
I used a 2 pronged approach that seemed to work really well. I used a Black and Decker toaster oven with a Whizoo Controleo kit, and I baked the reel of LEDs at 75C for 40 hours. And I also bought a large desiccant bag from Amazon (it's for gun safes), and a Ziplock airtight tupperware container. And I threw in some MSL cards as an indicator. So far, the fail rate has been moved below 1%.

Quote from: xaxaxa on July 14, 2018, 08:45:37 am

Quote from: funbags on July 12, 2018, 02:07:02 pm
I'll populate some panels and take extra ESD precautions and let you all know how I fared. I wont change my paste (SAC305) or reflow profile. I want to see if a ridiculously long pre-bake changes anything. Even halving my current fail rate of 15% would be plus.
I would recommend also trying with leaded solder and lower temperature; I recall seeing a comment somewhere that crap LEDs tend to use bonding compounds that are only rated to 220*C which is too low for SAC305 reflow profiles. Personally I don't ever do lead-free because reliability is important to me.

Thanks xaxaxa,
I've been super reluctant to use leaded solder paste (pregnant wife and a 3 yo), but I might have to if I want perfect LEDs. How do you minimize contamination of your work area? I noticed that my gloves do get a little paste on them inevitably. I live in an apartment, so my work area is in my living space unfortunately. Or maybe it might be worth my while to have them assembled by someone else?

Thanks All.

mikeselectricstuff · « **Reply #16 on:** July 16, 2018, 09:28:45 pm »

Metallic lead is notoriously unreactive - the fumes from leadfree fluxes are much worse.

I really wish that reputable companies like TI & Osram would team up to make a decent quality LED with internal control. Even if several times the cost of WS2812s, it would still be cheaper than the current solution I have to use on any permanent installation of seperate LEDs and drivers.

funbags · « **Reply #17 on:** July 17, 2018, 01:56:52 pm »

Thanks mikeselectricstuff! I'll do some more research and see what I come up with. I'm just overly cautious right now, with the baby and all.

I do all the soldering outside on our balcony (luckily it's summer and not freezing in New England), so fumes aren't really an issue. I carry the oven outside for every reflow.

I totally agree. I wish someone would make an alternative to the ws2812/clone chips...or at least have a factory that respects MSL 5a parts. Having to mandatory-bake the LEDs before use is kind of ridiculous. I'm getting the hang of storage and handling, but I would easily pay double for a more rugged package.

mikeselectricstuff · « **Reply #18 on:** July 17, 2018, 03:12:59 pm »

Quote from: funbags on July 17, 2018, 01:56:52 pm

Thanks mikeselectricstuff! I'll do some more research and see what I come up with. I'm just overly cautious right now, with the baby and all.

I do all the soldering outside on our balcony (luckily it's summer and not freezing in New England), so fumes aren't really an issue. I carry the oven outside for every reflow.

I totally agree. I wish someone would make an alternative to the ws2812/clone chips...or at least have a factory that respects MSL 5a parts. Having to mandatory-bake the LEDs before use is kind of ridiculous. I'm getting the hang of storage and handling, but I would easily pay double for a more rugged package.

If I could be sure they'd work 100% after a bake I would be less averse to using them, but it's lack of confidence that it's the only problem that means I can't risk it.

reboots · « **Reply #19 on:** July 17, 2018, 09:34:49 pm »

Quote from: mikeselectricstuff on July 16, 2018, 09:28:45 pm

I really wish that reputable companies like TI & Osram would team up to make a decent quality LED with internal control. Even if several times the cost of WS2812s, it would still be cheaper than the current solution I have to use on any permanent installation of seperate LEDs and drivers.

Quote from: funbags on July 17, 2018, 01:56:52 pm

I totally agree. I wish someone would make an alternative to the ws2812/clone chips...or at least have a factory that respects MSL 5a parts. Having to mandatory-bake the LEDs before use is kind of ridiculous. I'm getting the hang of storage and handling, but I would easily pay double for a more rugged package.

You may be interested in the Inolux IN-PI554FCH and IN-PI556FCH. These appear to be very similar to the WS2812B/WS2812. So similar, I can't tell if they're an original workalike or just a rebranding. (They certainly have name-brand pricing.) The timing characteristics differ slightly from the Worldsemi datasheets. If they are rebranded, maybe the characterization and QC are better.

https://www.digikey.com/product-detail/en/inolux/IN-PI554FCH/1830-1106-1-ND/7604647
https://www.digikey.com/product-detail/en/inolux/IN-PI556FCH/1830-1107-1-ND/7604648

mikeselectricstuff · « **Reply #20 on:** July 17, 2018, 10:18:19 pm »

The other problem is that they only have 256 grey levels and no global power setting, so they're only useful over a fairly narrow band of brightness - the APA102 types are a little better, with 5 bit global but still far from ideal.
TI have a nice 1-wire protocol in their TLC5937, and r, and esistor-set current, but only in a fat SO8. This die combined with a good RGB LED would be nice.
I currently default to the TLC5971 12-channel driver, as it has a nice 2-wire SPI protocol, software-settable current and inbuilt shunt regulator. 12 is a nice channel count as it works for RGB and RGBW. Must have specced about 100K of them over the last couple of years.

funbags · « **Reply #21 on:** July 18, 2018, 02:15:23 pm »

Quote from: reboots on July 17, 2018, 09:34:49 pm

Quote from: mikeselectricstuff on July 16, 2018, 09:28:45 pm
I really wish that reputable companies like TI & Osram would team up to make a decent quality LED with internal control. Even if several times the cost of WS2812s, it would still be cheaper than the current solution I have to use on any permanent installation of seperate LEDs and drivers.
Quote from: funbags on July 17, 2018, 01:56:52 pm
I totally agree. I wish someone would make an alternative to the ws2812/clone chips...or at least have a factory that respects MSL 5a parts. Having to mandatory-bake the LEDs before use is kind of ridiculous. I'm getting the hang of storage and handling, but I would easily pay double for a more rugged package.

You may be interested in the Inolux IN-PI554FCH and IN-PI556FCH. These appear to be very similar to the WS2812B/WS2812. So similar, I can't tell if they're an original workalike or just a rebranding. (They certainly have name-brand pricing.) The timing characteristics differ slightly from the Worldsemi datasheets. If they are rebranded, maybe the characterization and QC are better.

https://www.digikey.com/product-detail/en/inolux/IN-PI554FCH/1830-1106-1-ND/7604647
https://www.digikey.com/product-detail/en/inolux/IN-PI556FCH/1830-1107-1-ND/7604648

The only thing I can think of as a drawback, is that the timing has slightly shorter intervals, so it might not be 8mHz MCU-friendly... I'll pick up a few when I have a chance and test if its drop-in, or if I need to tweak the protocol library. Thanks for the heads up!

funbags · « **Reply #22 on:** August 03, 2018, 07:12:12 pm »

Hello all,
I just wanted to give an update.

I have switched over to the ~~SK9822~~ SK6812 LED and it is far superior in almost every way, except for minimum brightness. Its resin window is much clearer (the WS2812 was actually a bit yellowish and hazy. Though, it helped blend the colors better). It has a higher maximum brightness. It has better thermal characteristics. It's cheaper. And the manufacturer *actually* labels it as MSL 6 and it arrived in a thick vacuum sealed bag.

And this is the best part!!!!! It's a drop-in replacement. No change to my board, no change to code. It just works.

I've made 3 panels of 5 LED bars with 240 LEDs and I have gotten 0 fails. I tested one panel of 5 for 24 hours; cycling all the RGB combinations at max brightness, and none failed yet. To put it in perspective, I would see failures with the WS2812 after just 5 min.

I'll keep testing, because I'm still very skeptical, but the initial results are very promising.
Thanks for the input, all.

mikeselectricstuff · « **Reply #23 on:** August 03, 2018, 07:17:44 pm »

Quote from: funbags on August 03, 2018, 07:12:12 pm

Hello all,
I just wanted to give an update.

I have switched over to the SK9822 LED and it is far superior in almost every way

Where did you get them?

funbags · « **Reply #24 on:** August 03, 2018, 08:04:07 pm »

From Aliexpress. (Sketchy AF, I know)

EDIT: I mean SK6812 (not SK9822) I'll fix in other post.

One vendor sells them by the reel for $70 USD shipped. Which puts it $15 cheaper than the not-vacuum sealed reels of ws2812 I was getting from LCSC (china).

Confirmed Clones:
https://www.aliexpress.com/item/SK6812-adressable-full-color-RGB-5050-LED-light-source-with-built-in-chip-1000pcs-bag/32365923614.html?spm=a2g0s.9042311.0.0.4a824c4dLu74H3

Cheers,


EEVblog Main Site	EEVblog on Youtube	EEVblog on Twitter	EEVblog on Facebook	EEVblog on Odysee

Author Topic: MSL parts storage and care for the hobbyist/small business (Read 7482 times)

Share me