Just a heads up Calvin, the solder ball size is indeed 0.35mm.
I don’t know why the stencil you’ve got says 0.2mm but perhaps that’s pitch or something else (?)
The cheap low quality 0.35mm solder balls from China are slightly undersized (the BST ones for example) but they work absolutely fine, though I wouldn’t count on them being the alloy which is claimed on the label.
I think due to this undersizing and consistency issue 0.4mm balls would likely fit also, though depending on the brand and quality of the solder balls used it could lead to undesirable issues later on in the process.
I’ve had your exact same experience using solder paste and direct heat stencils, and another user here had your exact issue with the magnetic holder and stencils also.
I never thought after using successful the stencil for the other Switch bga ics, that the soc reballing is such a pain in the a* project. I will try next week at work, to build an extra holder which fixes the stencil extra in the region of the soc and make some release cuts in the edge of the stencil to prevent warping because of thermal extention.
The warping issue isn’t such a huge deal when using preformed balls, as they are much less likely to try and escape below the stencil.
Though I imagine it’s going to be a bit of a pain to get the balls into your stencil and jig type.
The issue I found was just the total size vs pitch, it’s just too hard to keep even and consistent downward pressure over the entire stencil when using solder paste and naturally as a result the paste seaps below the stencil, and the other issue is they have to make the stencil slightly thicker to account for the shrinkage which occurs while using paste too and this exacerbates the problem.
If you wan’t to try with the relief slots, I noticed on a few of my larger direct heat stencils they just drilled holes around the perimeter of the ball pattern, though tbh, this didn’t work too well for the Switch Ram stencil I’ve got, and I wound up just cutting the ram pattern out to a near 1:1 size of the IC itself and it works fine now.
What a mess. I think I m done with soc reballing today. Till the point of reflowing the solderballs, everythink seems fine. After applying heat by a iPhone X preheater from under the solderballs starts to aligne but some balls deside to pair up. After hotair from above it looked like I was drunk during placing the solderballs.
How do you fix the 1:1 stencil at the ic?
And I guess right, that you use the stencil while applying heat for solderpaste and solderballs?
I’m assuming this is during putting the balls on the SoC stage and you haven’t got to putting the chip on the board stage yet? but please tell me if I’ve misunderstood.
I’m always a bit weary of doing this, I try to expose CPUs/GPUs/SoC/APUs to as little heat as humanly possible on the die directly, I know other people do it and use preheaters and hot plates and get away with it but it makes me a bit nervous as the Tegra for Switch in particular is very sensitive heat wise. You shouldn’t need to preheat the underside of the SoC and can just use hot air directly to reflow the balls
I gues this is using preformed 0.35/0.4mm balls? and the pads on the SoC were wicked?
If so, in your case I’d leave the stencil in place during reflow, thin layer of flux on the IC prior to placing the stencil and then the balls (you can use a cotton swab as a brush of sorts to get the flux thin) if it’s too thick the balls will want to jump out. then hot air, I use about 390C at minimum air, start at a distance and come in closer when reflow is anticipated. balls can usually be upwards of almost 50% off the pad and provided there is not too much flux on the SoC they will find their home, this is a non issue if you keep the stencil in situe though during reflow.
I avoid using solder paste for the SoC and instead use a dedicated stencil and jig for this which is a bit different to using a direct heat stencil like yours, the jig takes care of stencil alignment (after setup) and it’s fixed, thin flux layer on the SoC, balls are placed in the jig, swirl it around until they find there home, fixed stencil is removed, then the SoC is removed from the jig and the balls are reflown without the stencil (these dedicated 90x90 stencils are way too thick for direct heating)
details here regarding the jig and stencill I use
In the case of the ram I use solder paste (and that’s about the biggest IC vs pad pitch left which I’ll use solder paste on) and on my cut stencil I’ll just align and put slight downward pressure in the center while filling with paste right upto the point of reflow, as I mentioned though, I wouldn’t use this technique on the SoC again, it’s just way to difficult with paste because of the large size.
I personally only keep the stencil in situe on the IC when using solder paste, but you can do the same with balls - but the stencil has to be a direct heat stencil in both cases as they’re thinner than the ones intended for 80x80/90x90 jigs
You’ll get it man, nobody gets this stuff first time, I remember back when I reballed my first CPU, I popcorned the chip, I tore pads, then came a GPU, I killed it with heat, they’re all good learning excercises at least
Yes. Soldering correct aligned solder balls to the soc is my current goal. I tried the iPhone X preheater for soldering the solder balls to the soc, so I would avoid movment of the solderballs by airflow. I don t like the idea either to heat up a cpu from die side. But I thought 200°C should be ok.
I tried a new round with 0.35 mm after some 0.4 mm solderballs get stuck in the stencil. I applied a thin layer flux with a little fine brush. Maybe after removing the stencil and poking lost balls in place I uneven the flux…
I have a 90x90 jig and I put the direct heat stencil inside close the opening with some tape and it is possible to use it for placing solderballs.
Maybe the solderballs are crap and I have to buy some new ones from a good brand. I should test tomorrow at which degree the solderballs began to liquify.
Couldn’t find any other manufacturer with reasonable priced ones in the EU of this size.
the “Best” or "BST"brand seems to be the best price vs quality but they’re from China and they seem to melt at a 63/37 like temperature, though I still have my doubts on most of the alloys coming from China. Even still they work fine aside from the slight undersizing of the balls.
From my experience, I found out that you would have to heat up one edge first and then slowly move toward the center then finish off. So that thermal expansion happens in the edges first instead of bulking the middle up. I usually have one magnet at one edge holding down the stencil and I start heating from the other side and holding down with a tweezer. I am going to try the 0.35mm (BST) balls soon with my Amoe stencil.
My good solder paste and the used china balls liquefy at the same temperature. So they both are mostly 63/37. So it s my leak of skill and I have to practice more to get it done correctly.
I thought the warping is caused by uneven heat and no room for thermal expanse. I had “luck” instead of only the direct heat stencil, aliexpress send me the whole set with the holder. But the stencil sits very tight in the holder and my first guess was, there is no room for expansion if the stencil is heated up. I will test if the stencil warps when heated outside the holder.
Finally I managed it to reball this tiny monster. I think it is the sixth or seventh attemp, but this time it worked.
I took new flux (original Amtech), place a thin layer of flux on the soc and tried again with removing the stencil after placing the balls and heat with hot air from distance and getting closer. And voilá
And you heat the solderballs direct with the stencil on the soc?
Is it possible to change the soc (with the emmc) to an other mainboard?
I have a mechanical damaged board beyond repair but I know the soc and its emmc are working fine.
Yes, heat the solderballs direct with the stencil on the SoC. Don’t put a lot of flux, to start “boiling” to raise your balls in stencil and the temperature somewhere around 320 degrees and air flow between 5 and 10%
I didn’t try to change the SoC, but it should work fine. It is paired only with eMMC
I tried to remove the SoC from the board, but failed due to pattern delamination.
I was told that too much heat would damage the SoC, but the time to apply the heat was too difficult.
From the back, I heated it at 250 degrees for 160 seconds, then injected flex into the SoC and heated it at 500 degrees from the SoC side.
PREHEAT APU/CPU bottom area to 180celsius then apply hot air at the top side.
Your APU/CPU mostly dead when you desoldering out and resoldering back in your temperature profile
