RGB LEDs Strike Again!
cgracey
Posts: 14,152
Yesterday, the first Prop123-A9 board came off the assembly line and it didn't work. Power supplies seemed dead and would drag the power input to just a few volts. I spent lots of time trying to figure out what the problem was and if something had changed. I couldn't find anything.
Daniel called, who designed the board, to find out how it was going. After about a half hour on the phone, checking everything, he asked how those RGB LEDs were oriented. Indeed, the pin1 indicator was where pin1 should be on the PCB. He remembered that there was something funny in the past about those WS2812B parts having the polarity indicator OPPOSITE of pin1, on pin3. Sure enough, I clipped them off and the board powered up!
We almost decided to get rid of those RGB LEDs on the last PCB rev, but we decided to keep them, barely. We should have gotten rid of them, as they are not only tricky, but can easily die during reflow.
Now, I'm thoroughly testing the design to make sure it does everything, like PLLs and what not. We got rid of the PCI-Express edge connector and just put a 256Mb SDRAM on there, all set to 3.3V. That is what we are going to need for Prop2, since the I/O's only work at 3.3V.
Hopefully, in a few hours, we'll know that everything's all right and on Monday morning, they can crank out 30 of these things. There has been a big rush to get this thing debugged, because it's all set up on the pick-and-place now, and it's a bear to take down and set up again.
Try to guess how many parts are on this PCB! My dad guessed 40, which is what I might have thought, not knowing much about it. Make a guess.
Daniel called, who designed the board, to find out how it was going. After about a half hour on the phone, checking everything, he asked how those RGB LEDs were oriented. Indeed, the pin1 indicator was where pin1 should be on the PCB. He remembered that there was something funny in the past about those WS2812B parts having the polarity indicator OPPOSITE of pin1, on pin3. Sure enough, I clipped them off and the board powered up!
We almost decided to get rid of those RGB LEDs on the last PCB rev, but we decided to keep them, barely. We should have gotten rid of them, as they are not only tricky, but can easily die during reflow.
Now, I'm thoroughly testing the design to make sure it does everything, like PLLs and what not. We got rid of the PCI-Express edge connector and just put a 256Mb SDRAM on there, all set to 3.3V. That is what we are going to need for Prop2, since the I/O's only work at 3.3V.
Hopefully, in a few hours, we'll know that everything's all right and on Monday morning, they can crank out 30 of these things. There has been a big rush to get this thing debugged, because it's all set up on the pick-and-place now, and it's a bear to take down and set up again.
Try to guess how many parts are on this PCB! My dad guessed 40, which is what I might have thought, not knowing much about it. Make a guess.
Comments
397
The chance of an error in any process is directly proportional to the cost of that process.
There are 416 parts on that board! Most are just bypass caps.
BTW: Even regular SMT LEDs are a pain to orient. I have red and green ones from the same manufacture and the polarity marks are on opposite sides...
It has probably ~30 more parts, due to pi filters on the PLL supply pins, plus the SDRAM and its 7 bypass caps.
Once they're built and tested, they can ship. They could be done on Monday or Tuesday.
Chip, killing RGB LEDs during reflow has one common denominator: excessive peak temperature. I don't recall what you guys have for profile validation, but that would be the best place to start.
No. We just need to put them on the right way.
These RGB LEDS have been very problematic for us. I think there's a quality issue with them, aside from our reflow process.
Unfortunately, high temperatures are needed for the new "environmentally-safe" lead-free solder chemistries.
If the parts sit around for too long, absorbing moisture, they can pop during reflow.
The metal contact surfaces also oxidize and lose their wetting ability, becoming unsolderable.
The old tin/lead system was 10x more reliable. Never did soldered parts just fall off boards, before.
This new system came about by EU mandate and, suddenly, Heller reflow ovens were all over the world. Even Parallax needed one to reflow the new solder, so that we could continue to sell products.
This whole thing seems, to me, to be about staid industry using government to force existing players into buying new equipment.
In farming, "Nothing runs like a Deere", until they lobby the government to pass new diesel emission requirements, forcing their existing customers to buy a new, very expensive, tractor that they don't need or want. Oh, and that they can't repair, themselves, anymore.
These things are all scams to redirect the profits of the little guys back into the hands of the big guys. This is why environmentalism is so divisive. Those in small business see that it's relentlessly exploited to undermine them, and favor those cozy with government.
We drank "pure water" ... "comes all of the way from Canada"... which turned out to have levels of radium so high that the good State of Illinois had to rewrite the law to allow us to continue to drink it... and not a word of it in the local paper.
So, anyone believing that environmentalism is even remotely related to a concern for the "little" guy needs to get his history straight.
I am tempted to buy a VW and restart my "global warming thread." Right after I get done playing with my P2:)
Be American about it. Charge extra for the bag. They are non-disposable and you can always give a refund if someone wants to send it back.
Conductive paste seems like a good option here. Who knows maybe we can start a trend:)
Surely, you mean "global taxing". There won't be any denying that.
All right. I'll stop. I don't want to induce nausea in any Prop 2 testers out there.
I got this reply when I asked MasterBond about a suitable substitute for solder in surface mount applications. Might be worth looking at.
Sandy
I think that would be the number of minutes before the glue solidifies.
-Phil
1) Quality of the part itself
2) Storage condition of the raw part from the supplier (sealed moisture barrier bag? probably not)
3) Raw parts baked and sealed for storage upon receipt?
4) If not 3, parts baked prior to use per MSL specifications?
5) Lack of proper reflow characteristics for the component. Their recommendations for lead and lead free profiles contradict each other in terms of what the part can handle.
6) Thermal challenges imposed by the PCB design. In other words, profile requirements for the Altera 484 ball FBGA are different than the LEDs due to thermal mass differences of the PCB where these parts are located.
7) Reflow oven capabilities/limitations. If I recall correctly, Parallax has a 7 zone oven, which makes it more difficult to create a profile that will meet both the Altera chip's needs and those of the LEDs. My first suspicion would be that the LEDs are being stressed a little too much during reflow because the reflow profile is geared towards success of the FBGA
Other than the quality issue of the raw part itself, none of these variables are impossible to overcome.
Chip, as I always tell Ken, I would be happy to swing by and assist with any process improvements if desired. I could bring my SlimKIC 2000 wireless reflow profiler to validate the PWI of your oven for this board.