Rear Mounting ws2812s (Questions/Advice)
Ian Shannon Weber wrote 08/29/2019 at 00:06 • 0 pointsI am working on a project and I want to use ws2812s diffused by the circuit board, which will involve either mounting them upside down or "oreo" construction. I want to avoid a multi-board assemble, so I am curious if anyone has any advice on mounting ws2812 leds upside down, or any advice on rear/reverse mounting any led. Do I need to do anything special to the pads? What soldering technique is recommended?
Discussions
Become a Hackaday.io Member
Create an account to leave a comment. Already have an account? Log In.
If you're doing a large production run and don't want to risk hand soldering all those LEDs an external LED driver with reverse mount RGB LED's might be interesting.
https://www.digikey.be/product-detail/en/sunled/XZMDKCBDDG45S-9/1497-1257-1-ND/5169746
Stacking PCB's is not possible in your case? https://hackaday.io/project/167548-sao-v169b-hackerspace-logo
Because you could actually load them in a pick and place machine and have them assembled in a factory line as well.
Are you sure? yes | no
There are WS2812s (or ripoffs, anyway) that come in 2020 packages, sort of trapezoid profile - I reverse-mounted those. I know you said you wanted the PCB to diffuse, but a dab of hot glue on top of a hole would diffuse fairly well IMO...
Are you sure? yes | no
If you cut a square 5x5mm and place the WS2812 in the hole. The solder pad would be <.1mm from the LED leads.
Are you sure? yes | no
Cutting a hole would not be possible since I want to have the WS2812 diffused by the circuit board. I will keep the idea in mind though.
Are you sure? yes | no
I'll have to see how much (i.e. too much) diffused the LED output be through a .062 pcb.
Are you sure? yes | no
>diffused the LED output be through a .062 pcb.
There are thinner PCB too. :P
Are you sure? yes | no
You might try "tombstoning" some zero ohm SMT resistors and using those to bridge the LED connections.
Are you sure? yes | no
You could fold the leads of the WS2812 up, so there is no longer a gap between the pad and the lead. Though you'll probably end up breaking a few leads, so make sure you have some spare leds.
Otherwise you'd need large blobs of solder to span the gap. A bit tricky but doable. You'd need a custom footprint, where the pads don't go under the white of the led.
Are you sure? yes | no
WS2812 2020 are available on aliexpress :P these look much easier to reverse mount than 5050 package.
https://www.aliexpress.com/store/group/WS2812-2020/701799_513273278.html
I also want to reverse mount a bunch of WS2812B, we removed them from neopixel rings using an air gun, something must'd happened because they stopped working when we soldered on a pref board and back to the ring.
You can follow the project https://hackaday.io/project/167198-nyan-earrings I will try posting an update in a couple of weeks.
Are you sure? yes | no
Following. Interested in updates. Thanks for the tip.
Are you sure? yes | no
The WS2812B are very sensitive for heat and you must have damaged these with the hot air gun. I always bake the WS2812b or APA102 in the oven for at least two days at around 75C (167F) to get all the moisture out. My success rate using these in the reflow oven without baking them first was around 50%. Even soldering these manually could damage the chip or LED's when you are putting too much heat into these.
Are you sure? yes | no
I used an LED designed for this on my magnetic encoder board. PCB has a hole drilled in the center of the two pads, and the LED just drops in. No issues. You might search to see if there is a variant of the WS2812 that is designed for rear mount. The bracket idea is genius but it's a LOT of hand soldering, not to mention fabricating the brackets. The pads on the LED come around the sides, right? So I wonder if you could use a SMT SIL 2x or 3x header and solder them in thru hole upside down (the part that's supposed to stick up, instead stuck down through the hole next to the LED) and then use the part that's suppose to SMT to connect to the LED pad. https://www.harwin.com/products/M22-5330305/ or use a SMT SIL like this one: https://www.quarndoncomponents.com/product/connector-pin-4-way-2-54mm-pitch-sil-surface-mount-pin-header/ installed as intended, but then mount your LED up on the plastic of the SIL. It would be a bit further from the hole in the PCB, but might shine through ok. Or do a combination of those ideas with thru hole SIL header strips. In any case, if you get the spacing between the headers just right, they will hold the LED in place while you solder it.
Are you sure? yes | no
SILs. I didn't realize that these were an option. Thanks for bringing this to my attention.
Are you sure? yes | no
How about buying something like this and installing the pins upside-down?
https://learn.sparkfun.com/tutorials/ws2812-breakout-hookup-guide/all
If this is too much of an "oreo", you can skip the FR4 and stake the LED to the board upside-down with little stiff shunt wires to the pads/holes. Use a dab of glue on the edges to hold the part in place while you solder the first wire. I'd want a microscope for that operation, and I'd not expect to build more than a few of them that way.
Are you sure? yes | no
I think that I might just have to get a bunch of standard RGB LEDS for my purpose. I think I am going to do a test run with WS2812s and try shunt wires or "brackets" just to see how labor intensive it will be. My estimated end quantity is around 20 boards with 7+ LEDs. So soldering 140 WS2812 upside down with shunt wires might prove to be way to much labor.
edit: Using the breakout would be too oreo. It would double the board count, which I am trying to avoid.
Are you sure? yes | no
A trend I've seen more recent: side mounted LEDs and hot glue as a diffuser. Definitely nothing for a large run, but it works. There are SK6812 LEDs in a 4020 casing that are side mounted. The 5050 cases aren't that robust so I'd imagine soldering them upside down will break some. For my single color LEDs I just solder blob one pad and go from there.
https://twitter.com/mrtwinkletwink/status/1160037616240287744/photo/1
Are you sure? yes | no
SK6812s are a new part to me, thanks for it because I was unaware of anything other than the WS2812s in this LED space. I am not totally sure what I am looking at in the twitter link. Could you explain how the LED is mounted? Does it use that side mounted LED you were talking about?
Are you sure? yes | no
Yes. The badge uses side glowing rgb faders, but the technique would be the same. Have the clear fr4 and put a side gloeing led next to it. Add got glue. Should diffuse enough light into the fr4
Are you sure? yes | no
It might be difficult to reach from the plated pads of the WS2812 to the board when mounted up side down. You might want to consider some "brackets" to overcome the gap: http://californiasteam.tech/images/IMG_1093.JPG
This also allows solder pads farther away from the WS2812, so that you can have a larger area to diffuse the light (when that is required). Like the idea and I am curious how your final solution will look like.
Are you sure? yes | no
I like this concept. I don't think these "brackets" would have occurred to me organically.
Are you sure? yes | no
Are you planning on soldering the tops of the pads to the bottom of the PCB?
I personally think this would work but I'm afraid the LED hanging from its pads would make the circuit more fragile than normal.
I've wondered about doing something similar with WS2812 LEDs and other surface mount parts but I've yet to try it.
Edit: I was thinking you'd want a hole in the PCB large enough for the WS2812 chip to fit in. This way the chip wouldn't stick up as far on the top of the board.
Are you sure? yes | no
I am planning on soldering the light emitting part of the LED toward the board. You are right, the distance (pad to LED top) on the WS2812s is pretty big. I think that if I continue with WS2812s I might use the "bracket" idea that Sander van de Bor mentioned.
Are you sure? yes | no