EEVblog Electronics Community Forum
General => General Technical Chat => Topic started by: tszaboo on February 02, 2017, 11:51:42 am
-
I've just replaced a BGA memory on the product of ours. Testing the double size. It only took the second attempt. I'm surprised, it was so much easier than I thought. Just plenty of flux, and the hot air station (which is an Atten 858, nothing fancy). It is running the memory test, loop 6 ,so for sure it is OK. That is all.
-
Congrats on the work there :-+ To be honest most BGA rework is quite easy really, as long as you have the appropriate equipment etc. I've been
doing BGA replacements for a while now, mostly mobile phone stuff amongst other things with a good success rate.
If the replacement BGA's are supplied pre-balled it's even better. When I started doing BGA I used to use too much flux which can be a hindrance
as what sometimes happens if you use too much is that it boils underneath the IC which tends to float the chip out of place.
-
Spiffy! Did you find any particular tutorial that was helpful?
Asking for a friend ;)
-
Spiffy! Did you find any particular tutorial that was helpful?
Asking for a friend ;)
The memory came with solder balls. So what I did:
Remove the old memory (half the capacity) The setting was 350 degrees, narrow head on the hot air, 3-5 cm distance, small amount of air, lot of patience.
Clean up the pads completely with wick
Clean up the pads with IPA
Put generous amount of flux on the pads, I use Multicore 425-01.
Place the memory in place.
Heat it up, same settings. When it "snapped" in place due to the solder I moved it carefully like 0.2mm with tweezers, and tapped on the top of it on both sides. Maybe this is wrong, IDK, it worked.
Place back the rest of the components which moved.
Try to inspect it with the microscope, give up, and just plug it in, hope for the best.
-
Good job! I've not tried yet and am scared to... Maybe I should give it a go at the next opportunity :)
-
And I have a hard time just unsoldering DIP components.
Good job.
-
The hardest part about BGA rework is getting over the "OMG this looks hard!" stage. Once you actually just do it, it's not so bad. Take your time, use plenty of flux, make sure the PCB is secure and level. Apply heat and allow physics to do the rest.
-
The hardest part about BGA rework is getting over the "OMG this looks hard!" stage. Once you actually just do it, it's not so bad. Take your time, use plenty of flux, make sure the PCB is secure and level. Apply heat and allow physics to do the rest.
No, I think it is actually harder, for sure. You cannot inspect the solder joints. Also, rework is very hard, I've tried salvaging the first try, re-balling it with iron, but I dont think it is possible. The iron left uneven amount of solder on the pins, so they did not solder well. With a regular SMD, I can try doing it several times, with BGA is one try for a part.
And I have a hard time just unsoldering DIP components.
That can be really hard, especially on 4+ layer boards with idiot enginers not using thermal relief and tiny pads. Multiple irons, desoldering iron, etc, does not help, if the design is retarded.
-
Perhaps it was just beginners luck, maybe next time will be harder. I do have access to an xray machine though so inspection is relatively easy. PCBs and IC packages are rather transparent while solder blocks xrays nicely.
-
It is possible to reball smaller chips with soldering iron. When chip starts to float just press it down gently with a tweezer so that solder on smaller "balls" catches onto pads. Obviously it's a totally ghetto solution and reliability of such joints might be bad.
This definitely doesn't work with larger chips (think more than 6x6 balls).
Sent from my HTC One M8s
-
That's great, congrats! :-+
The hard part is to make it work.
The really hard part is to make it reliable.
The really, really hard part is to make it repeatable.
The really, really, really hard part is to make it last on the long run.
:phew:
-
That's great, congrats! :-+
The hard part is to make it work.
The really hard part is to make it reliable.
The really, really hard part is to make it repeatable.
The really, really, really hard part is to make it last on the long run.
:phew:
I just wanted to test the thing with 2x the memory, so we can send it into production. This was definitely faster/cheaper than running a proto batch just for this. ;)