Wow, I'm glad to see that this topic is still active, and I'm even more glad to see that I'm not the only fool out there who is attempting to successfully reball and reattach the 360s BGA chips on their own.
Currently, I've made about the same progress as nagamin has. I'm able to consistently remove all of the BGA chips (GPU, CPU, South Bridge, and RAM) with out damaging any of the pads on both the chips and mobo. So far, the GPU is the only chip Ive tried reballing, and I'm at the point now where I'm getting 3 successful reballs for every 4 attempts. I haven't tried the reattachment process yet but I can already tell that its going to be the trickiest part. So I'm holding off until I can figure out a reliable technique for realigning the chip properly on the motherboard and a way to hold it in place while its being heated.
Now to answer the questions I'm sure everyone is going to ask.
The hot air rework station I'm using is a Saike 852D+. Its a cheap clone but works really well.
For a hot plate I'm using a modified griddle with stand offs to securely mount and hold the motherboard in place. It has a max temperature setting of 420F (215C).
I'm using the stencil technique to reball, and the reball station is the KingFull model I ordered off the internet. The kit came with 4 stencils for doing the 360s CPU, GPU, Southbridge, and RAM, along with 2 sizes of solder balls and tacky flux. The balls used on the GPU, CPU, and Southbridge are lead based Sn63Pb37 and are 0.6mm in size. The other balls it shipped with are of the same alloy but are 0.45mm in size and are for the RAM chips. The tacky flux is Amtech RMA-223-TF.
The kit didn't include any kind of directions or literature but its design is pretty much similar to the other DIY BGA stations out there. A solid chunk of aluminum that clamps the four corners of the chip to hold it in place, then the stencil is installed into an aluminum frame that fits nicely over the chip. I was only able to find a couple videos out there that showed how to use a kit similar to the one I own but none of those techniques seemed to work well with the one I have, so I was on my own. Through trial and error, I was able to find a process that seemed to produce consistent results. According to the videos and information I gathered, their technique was to apply a thin layer of flux on the chip, set the stencil on top, pour in the balls, then ever so carefully lift the stencil off, leaving the balls perfectly aligned and ready to heat. For a solid week I tried to get this right, but every time I lifted the stencil, some of the balls would lift with it, and when I would try using less flux, the balls then wouldn't stay in place. I was beginning to get pretty sick of this damn thing when I decided, what the hell, lets see what happens when I try heating the balls with the stencil still in place. Foolishly, I put the hot air right on it and the stencil warped drastically, letting all of the little balls roll under nether it. Well that wasn't going to work, so back to the drawing board.
I would spend the next few days retrying the stencil lift technique again and right when i was about to crack, my dad suggested that I should try heating the balls with the stencil still in place again, but this time I should preheat the station first. So I clamped in the GPU and placed the aluminum block along with the framed stencil, side-by-side, on the hot plate and set it for "Serve" which was about 200F (93C). After about 15min, I took a temperature reading of the GPU with a IR thermometer and it read 140F (60C). I then applied a thin layer of flux onto the chip, insuring all of the pads were covered, then placed the stencil on top. After checking to make sure the stencil aligned perfectly with the GPU, I turned the hot plate up to 325F (162C). I then proceeded to pour in the little solder balls and brush them into place. Once everything looked good, I turned the hot plate up to 375F (190C) and let it sit for about 15min. BY now I figured that the GPU and solder balls should be very close to their melting threshold so I fired up the hot air, set its temp for 220C and air flow to 2.5 (the airflow scale on my rework station goes up to 7). This time around the stencil didn't warp at all. As I applied air to the stencil I used the little scrapper thing that came with the reball kit to lightly push down on the corners of the stencil where some of the balls were sticking to the side of the stencils holes. It was a cool effect watching the balls almost jump into place on their own, centering themselves perfectly on the pads of the GPU. Once all of the balls looked to be in place, I turned off the air along with the hot plate and went to go finish watching the rest of the FLCL series that I had started earlier. When I came back, everything was cooled off, so I lifted the stencil off of the GPU, praying that none of the balls would lift with it, and viola! I had a perfectly reballed GPU.
Below are a few pictures of my results. Unfortunately, my digital camera kicked the bucket before I could take more pictures of the GPU and of the process. The GPU shown reballed was my guinea pig through this whole trial and error process and that's way some of the solder mask(?) is missing in the corners. This picture is also of the first presentable result I had with the reball process, and is by no means perfect. Like I mentioned above my camera has died so therefore I cant show you all the more recent results I've been getting with a freshly removed GPU.
And here is a blurred picture of the BGA station I'm using taken with my cell phone.
I also ask that anyone with experience regarding temperature profiles of the GPU, or any of the 360s BGA chips, to please heavily scrutinize my procedure. I've been on my own through most of the process and the temperature settings I used were those that I found to work during my endeavor.
Oh yeah, I apologize for the super long post. I, as well, enjoy writing way to much. -=D
Author
Topic: GPU Reballing (Read 524 times)