Raspberry Pi 3 Xenon Flash Death

[Treehouseman]

It's back, and this time with vengeance. Seems this time around that the actual Broadcom chip itself is photosensitive, having the flash 6-8" away consistently crashes the pi unless the chip is covered.



Alright, my bad, as pointed out below it was covering up U16, when I tested U16 it must not have covered it properly, it is indeed U16 causing it, the passives must be covering it from the far side and the light was just coming in from the middle of the board, so despite saying they were going to fix this exact issue after the Pi2 they have failed to do so.

[wraper]

Dunno, Broadcom SoC have the package which is not transparent. Moreover you covered U16 with SD card too.

[Treehouseman]

Dunno, Broadcom SoC have the package which is not transparent. Moreover you covered U16 with SD card too.

Crap, apparently I must have not been completely paying attention to exactly what I was covering, it is indeed U16, the passives must be shielding it from outside angles, but the light was coming from the middle of the board when I was testing covering that area, just retested and covering only U16 prevented the crash.

[SeanB]

Same bare die in it, so the same issue. Must not be available in a plastic package, or the cost of putting a blob of COB epoxy on each board is way too high for the return as there likely are very few instances where the board is exposed accidentally to very bright light.

[NiHaoMike]

So what's a good DIY solution for this? The obvious is some dark paint/glue, but there seems to be some concern with "unapproved" materials degrading the bare silicon over time. Is that a real problem or would it only happen long after the board is obsolete?

[SeanB]

Non conductive black electronics grade silicone sealer blob?

[Ian.M]

Silicones usually emit various chemicals as they polymerise.  Electronics grade just means that they don't emit acetic acid or anything else highly corrosive.  I would suggest a brushable conformal coating, well cured, as a barrier layer to prevent any residue from whatever blackout material is used from getting under the flipchip where it could cause problems.

Beware of black paints - many contain conductive pigment particles and if you dont conformally coat the board first, will cause major problems.

[janoc]

So what's a good DIY solution for this?

Put it in an enclosure? 

[SeanB]

How about just a blob of potting compound then, not the rock hard setting but the slightly flexible one which will cover the chip but not rip it off the board as it cures and shrinks.

Of course if anybody knows what is used in the popular COB blob as a rock hard black covering agent, and how to get it in small low cost pack sizes.....

Of course most black pigments are often carbon black in a binder, especially toners, which are a blend of carbon black ( and all the lovely fullerines and buckyballs and nanotubes in it) with black iron oxide powder, nicely blended with some styrene monomer and milled to a fine powder. It might have a high enough resistance not to matter.

[Monkeh]

So what's a good DIY solution for this?

Put it in an enclosure? 

Don't buy a Raspi?

 

[klunkerbus]

black liquid electrical tape? Although I don't know how opaque it is for an application like this.

[XOIIO]

JB weld is an epoxy  It's also an electrical insulator, though I'm not sure if it would have an effect on the silicon.

[Ian.M]

JBweld claims to be steel reinforced epoxy.  It also contains a low percentage of carbon black (according to its MSDS).  I wouldn't want to use it on a sensitive circuit, and if it did give you trouble it would be virtually impossible to remove without damage.

[klunkerbus]

I just have to ask... have you ruled out the possibility that the problem is some form of a noise transient from the flash and not from light?

I used to do elaborate Halloween setups that involved a couple of flash units.  For the strongest flash unit, it didn't take long to learn I had to be careful placing anything else too close to it. 

[wraper]

I just have to ask... have you ruled out the possibility that the problem is some form of a noise transient from the flash and not from light?

I used to do elaborate Halloween setups that involved a couple of flash units.  For the strongest flash unit, it didn't take long to learn I had to be careful placing anything else too close to it.

This is problem known since RPI 2 and is caused by the light.

[MK14]

This is problem known since RPI 2 and is caused by the light.

Dave of EEVblog forums (here!), did a video about it. Here:

[alsetalokin4017]