Bluescreen - Advise on reflow attempt

Link here
https://switchbrew.org/wiki/Panic_codes

So hard to tell them apart, I’m pretty sure the shade is darker still when the fuel gauge is the issue, though, this doesn’t seem to be detailed in that link, so perhaps I’m wrong, or certain LCDs produce a slightly different colour.

So I suppose it could be one of three things, SoC, Ram, EMMC and maybe PMIC, I would still put this one to one side until you get a modchip, I think further reflow attempts will make matters worse as it might just be a case of a corrupt EMMC partition causing this

Can you get the BIS keys from an emmc put into a different device that is in RCM? If so I could open up a backup of the EMMC on PC and see some of the details of it?
Given the original fault was clearly someone trying to snap the thing in half, as seen in the screen damage, I would still put money on it being RAM of SoC… maybe there is enough of a bend in the board that a reflow just doesn’t help…

Unfortunately not :frowning: EMMC needs to be with the SoC to get the keys.

Could be, I’ve had boards like this and when pulling the SoC you can see pads and there connecting traces have been pulled, then the painstaking process of repairing the traces begins, or better yet putting the SoC and EMMC module on a good donor board. I had one a couple weeks ago with the tail end (battery section) snapped off, had to swap the SoC and EMMC over to another board.

That may be the direction to head in, but I don’t think I am ready for an soc re ball yet, so will park it again for now… its not like I dont have enough other boards to get round to looking at!

So, I came across the board again yesterday, and thought… eh, why not, and gave it another re-flow.
I couldn’t get to the point where I could nudge it, and I never saw the meal shield fully melt, so I stayed on the heat for a lot longer than I did previously, but stayed away from the centre of the chip for the most part.
And it booted.
So, the fuse count is clearly ok! This only really leaves me with one question, would you expect this to last? Can I happily re-sell this now with some testing, or do I need to do a full reball to make it a lasting fix?

I personally wouldn’t, in cases such as these where we suspect bad joints are the original cause of the issue then we can assume the reason was either of the following - simply bad solder joints, pad or via issues, board warp etc

Can’t realy determine any of those without pullnig the SoC

in cases such as this where I’m planning on selling, I’ll transfe the SoC to another donor board (which failed for some unrelated reason) and then of course extensive testing afterwards

If you have a personal Switch, I would transfer all your stuff over to this one and use it as primary console and sell your known good instead, that way if it fails later down the line it will only affect you :slight_smile:

I do have a personal switch, it was my first switch fix (a banana switch when it arrived). Hmm, I will have a think…

So, I left this one sitting for a week, and it is now back to a blue screen, though now I can get it to boot if I apply pressure.
I guess the next step is to remove it, re-ball, and take a look under it and decide if I should put it back on (with or without pad repairing) or put it on to my parts board and start moving everything else across?

I have only attempted one SOC re-ball (on a dead one) and it went terribly. Any advise on managing it?
I will have another practice go on the dead one before attempting anything on a live one.

I see, so assuming the SoC is good (that the heat wasn’t temporarily resolving a die issue or something) and it was indeed joint related issues, then I suppose we have to assume it’s a board level defect either as a result of board warp which is causing the balls to detach or not making proper contatct with the board following reflow or that there’s pad/via issues on the board… think the best course of action would be to transfer the SoC and EMMC to another board just to take it out of the equation, unless of course you pull the SoC and can positively identify something along the lines of a detached or soft pad/s etc and conduct the repairs.

Best advice I can give on this is to practice reballing on your other dead SoC, repeat reballing it over and over until it’s perfect every single time, practice makes perfect :slight_smile:

Don’t use solder paste, the chip is too big for it and it never reliably works, even with those wanky magnetic holders :slight_smile: Instead use a dedicated 80x80 or 90x90 stencil with the proper jig (the bronze coloured ones are perfect for this)

When removing the SoC from the board, preheat the opposing side of the board first with your hot air (everywhere) at 150C then step it up to 200/250C for a min or two, then flip the board and add flux around the edge of the SoC while continuing to heat the SoC at 200/250C, when the flux is applied crank your temp up to approx 420C medium air, run your air around the wafer and don’t concentrate on the die itself, nudge the edge of the wafer to ensure the chips is fully molten below and when it is use one of those suction pens to lift the chip, while the board is still hot, tin then wick the pads on the board, then do the same on the SoC, if your iron is struggling on the board have your hot air assist you, try not to remove any solder mask while wicking otherwise this could hinder you when it comes time to put the SoC back on (if you decide to re-use this board)

After chip removal take a close look at the resin/epoxy around the die, if it’s a greyish/bluish colour your good, if it’s a nasty looking brown then the chip is almost guaranteed to be dead.

On the SoC itself, after the pads have been wicked and then cleaned, add a thin film of flux, I usually just use a cotton swab like a paintbrush, it has to be just the right amount, which you’ll only know with practice, too much flux and you’ll cake your stencil up and the balls will not roll into the holes easily or get stuck in the stencil when you remove the upper portion of the jig, too little and the balls will piss off on you when it comes time to reflow them.

when the balls have all been swirled into place with the jig, I tilt the whole jig at a 45 and gently brush over the stencil with a dry cotton swab, this knocks out any double balls etc, then you want to closely look at the stencil to make sure none are missing, after remove the top part of the jig with the stencil, lift out the chip carefully, take another close look and make sure it’s not missing any balls, if it is then repopulate them manually, reflow the balls at a lower temp and minimum aireflow, I use 320C

After it’s similar to the removal, I preheat the opposing side upto 200/250C, add a thin layer of flux to the SoC pads, align the chip , crank the temp up to about 420C and reflow it into place

Hope that helps :slight_smile:

Thanks! I think the missing ingredient here is having a decent jig. I will go hunt one down :slight_smile:

This is the style your after, avoid the silver ones as they can’t grip the chip properly

You can probably find it quite a bit cheaper but Aliexpress search is completely hopeless these days…

The stencil I got off eBay ages ago

The balls I’d go with are the BST brand 63/37, I’d avoid the cheaper no name brands as they are rarely consistent.

Thanks, I have ordered one now. I do have a stencil for it, I’m just not sure its the right size… its like 60mm x 50mm…

Yeah that won’t work, I guess the version you have is a direct heat stencil, your looking for one that’s either 80x80 or 90x90 with the four holes in it

like the following

I’d go with the 90x90 option

Wow, the switch ones seem pretty marked up… couldnt find any cheaper though so have claimed one.

Yeah it’s pretty pricey for a few pennies worth of sheet steel, sure they’re making a killing on the sale of these… be nice to have a laser cutter which is powerful enough to plow through steel then just make my own stencils as and when :slight_smile:

So, this arrived the ither week,and yesterday was my first chunk of spare time since, so thought i would attempt to reball my dead cpu as practice.
I am not sure what I am doing wrong, but there seems to be a gap between the chip and stencil such that the balls can get pushed to the side.
Is there something i can adjust here?

I’m operating from memory here as this is a set and forget sort of thing but I think the four gold coloured knurled screw wheels adjust top stencil height on the jig, there may be another adjustment for the chips height… I forget

you’ll want to keep a slight gap otherwise you’ll cake the stencil up in flux or the balls will roll straight out when swirling

I have it set so the gap is on the larger side which can allow two balls in one hole sometimes and tilt the whole jig at a 45 afterwards and knock them out with a dry cotton swab bit annoying but it makes the swirling process a bit quicker

Yeah, the 4 screws would allow a larger gap,but not smaller. Right now i end up with thousands of balls under the middle part. It looks like i could move the chip up a but with some shims, but thohght there may be anither way. I will try that :slight_smile:

Odd, pretty sure you got the same jig as me so the fit and finish should be pretty on par as mine… do you see any deviation in the jaw height when closed? does the surface of the SoC pretty much (give or take a few points of a mill) come to the same surface level of the jaws?

Mine doesn’t have this issue and no balls slip under the stencil and I don’t need any sort of shim for the SoC :open_mouth:

Can I ask what solder balls your using? some have deviation in size, I’ve taken to using the BST 0.35mm balls as they do genuinly seem to be leaded solder, have a nice shine to them and cheap in price… with the only caveat being is they are a fraction undersized… cheaper no-name balls (Mechanic probably) may well be further undersized and could contribute to the issue of them slipping below the stencil when swirling.

I notice on mine there is a couple of M3 hex screws which go in the otherway which if loose (I haven’t checked this for sure mind) could potentially alter the height slightly, so maybe tightening them up will make a difference, though it seems you’d have to take it fully apart to get to them :thinking:

The other possibility is your stencil is bent/warped, which might explain why it seems to be happening in the middle, I’d take it out of the jig and hold a steel rule/set square up to it and ensure the deviation isn’t to wild, if it is I’d get in contact with the seller for a replacement.

Also, this is maybe an obvious one and you probably know already, but for anyone else, these dedicated 80x80/90x90 jig stencils are thicker and not meant for direct heat (using direct heat on them will cause them to instantly and possibly permanently buckle), after swirling the balls around etc the upper portion of the jig and stencil should be removed, the chip with the now placed balls should be removed from the jig entirely (and double checked for no missing/double balls etc) and the IC and balls should be reflown outside of the jig etc.

Yeah, its weird. The chip sits a little below the jaws, padding it with some card doesnt seem to make a huge difference. I will order some new balls, see if that helps.