advertisement


Anyone reballed a BGA chip?

C

Cesare

pfm Member
I've got a broken Tektronics scope where i've traced the problem to the BGA processor - it's unclear whether the processor itself has developed a fault, or whether it's a problem with the solder balls - it's old enough to be from that first batch of lead free soldering which may or may not have suffered some sort of thermal stress over the years and ended up borked.

Symptoms were a tripping power supply. This turned out to be a excess power consumption on the 3.3v rail. Using a bench supply to tickle the circuit a bit, the power consumption turned out to be happening within/under the processor (given where the heat was being dissipated). Heating it to reflow the balls made things worse, so it definitely feels like a short in the ball grid array. It's an MPC860, so a PowerPC based microcontroller, no longer made, with enough variants that finding the right one as new old stock is tricky. Best bet is probably to salvage one from another board if the processor has failed.

So, since obtaining a replacement processor seems to be somewhat problematic, i'm first going to have a go at reballing the existing processor in case this resolves the issue. I've never undertaken such a process before, so i'm sort of winging it.

I've got as far as removing the processor with hot air + a heat pad underneath, and have cleaned up the board and processor. No pads lifted, and the short has gone, so there's a chance i'm onto the right cause.

The socket is top right on the circuit board in case this isn't obvious:

XxF2rEQ.jpg


And here's the back of the processor:

BpuYdzk.jpg


So at least the BGA is regular, so a standard stencil may align well enough to get some solder balls in the right spots. I've got some solder balls and stencils on order from china, so i'll see what happens when they turn up. I'm somewhat unsure what size balls to try, what flux etc, so it's going to be a journey of discovery.

Has anyone attempted such a feat? Any useful advice other than don't do it? ;)
 
Cesare said:
it's old enough to be from that first batch of lead free soldering which may or may not have suffered some sort of thermal stress over the years and ended up borked.
Click to expand...
This was REAL common in some computers after the switch to RoHS in the early 2000s. It's quite likely the source of your problems. I wouldn't tackle it but If you feel you can do it, good luck.
 
you can make the balls by squeegieing solder paste through a stencil onto the cleaned chip. Then heat with gently hot air to form the balls. Solder paste comes in differing solder to flux ratios depending on your stencil thickness (I didn’t know this until I used the wrong one and bridged everything).

You want a good amount of flux under the reballed chip if you are positioning by eye (sounds like you are), and the chip will float on the flux while the solder balls melt and their surface tension pulls it into alignment.

Best of luck! I’ve done a few bigger chips with proper infrared solder stations, but I’ve also mounted smaller BGA chips to boards using a frying pan as my oven and IR thermometer to check when it’s up to temp, with a decent success rate!
 
Puggie said:
you can make the balls by squeegieing solder paste through a stencil onto the cleaned chip. Then heat with gently hot air to form the balls. Solder paste comes in differing solder to flux ratios depending on your stencil thickness (I didn’t know this until I used the wrong one and bridged everything).

You want a good amount of flux under the reballed chip if you are positioning by eye (sounds like you are), and the chip will float on the flux while the solder balls melt and their surface tension pulls it into alignment.

Best of luck! I’ve done a few bigger chips with proper infrared solder stations, but I’ve also mounted smaller BGA chips to boards using a frying pan as my oven and IR thermometer to check when it’s up to temp, with a decent success rate!
Click to expand...

Yes, I saw the solder paste + stencil option vs the solder balls approach, and it felt to me that if I had the official stencil then that's probably the right approach. However, without the right stencil I felt that going the solder ball route was more likely to succeed, as I can poke the balls around under a microscope if I can't get things quite lined up, but i'm totally up for changing direction if this seems daft.

I do have a small hot plate that will allow me to reflow the chip itself, and I think i'll be able to attach it to the board with hot air + preheating the board with the plate as it came off pretty easily.

I hadn't considered a frying pan though :)
 
doesn’t seem daft, just not the way I did it as I had stencils and solder paste available that I was comfortable using (for non BGA stuff at that point).

Stencils should be very standard, it’s the ball size and pitch between them that matters. Your stencil could be a massively bigger array and still be perfectly useable.

I don’t know how much experience you have with stencils and solder paste, but tape everything into alignment, chip to something, stencil to the same something to maintain alignment while you squeegee the paste in.

One advantage I can see with the all paste option, is once you have cleaned the chip and board, you know all the solder from chip to board is identical.

Best of luck with it!

Edit: oh and the frying pan thing, we were camping and needed to make up a PCB for something that broke. We had VERY limited options. I’m not suggesting it’s the best way to make up pcbs, but it does work, just to prove you don’t need a clean room and all the gear just cos it’s surface mount or bga (to be fair if the board had any through hole stuff to populate we would have been screwed).
 
Puggie said:
doesn’t seem daft, just not the way I did it as I had stencils and solder paste available that I was comfortable using (for non BGA stuff at that point).

Stencils should be very standard, it’s the ball size and pitch between them that matters. Your stencil could be a massively bigger array and still be perfectly useable.

I don’t know how much experience you have with stencils and solder paste, but tape everything into alignment, chip to something, stencil to the same something to maintain alignment while you squeegee the paste in.

One advantage I can see with the all paste option, is once you have cleaned the chip and board, you know all the solder from chip to board is identical.

Best of luck with it!

Edit: oh and the frying pan thing, we were camping and needed to make up a PCB for something that broke. We had VERY limited options. I’m not suggesting it’s the best way to make up pcbs, but it does work, just to prove you don’t need a clean room and all the gear just cos it’s surface mount or bga (to be fair if the board had any through hole stuff to populate we would have been screwed).
Click to expand...
All for your "Fire building and advanced electronics assembly" Scouts badge ?
 
Tw99 said:
All for your "Fire building and advanced electronics assembly" Scouts badge ?
Click to expand...
Was supporting a student race team one of the safety circuits for our EV had failed so we worked out what was wrong and populated a new PCB (always take plenty of spares) while we were cooking up dinner that night. We were camping as lower hotel bills = more money to spend on the car!

Stick all the components on with solder paste, sauté until medium well done (chip is at 220deg on the top and all the solder paste looks shiny) and you are good to go.
 
Puggie said:
Was supporting a student race team one of the safety circuits for our EV had failed so we worked out what was wrong and populated a new PCB (always take plenty of spares) while we were cooking up dinner that night. We were camping as lower hotel bills = more money to spend on the car!

Stick all the components on with solder paste, sauté until medium well done (chip is at 220deg on the top and all the solder paste looks shiny) and you are good to go.
Click to expand...

Formula Student? or similar
 
gintonic said:
Formula Student? or similar
Click to expand...
Formula Student exactly, we did a 350v 80kw (limited power) single seater EV. This was one of the student designed boards to convert analogue sensors to CAN with some onboard DSP (to sort out the boomy resonant room modes of the wheel speeds).

I support a few different student competitive teams for my current employer!
 
Puggie said:
Formula Student exactly, we did a 350v 80kw (limited power) single seater EV. This was one of the student designed boards to convert analogue sensors to CAN with some onboard DSP (to sort out the boomy resonant room modes of the wheel speeds).

I support a few different student competitive teams for my current employer!
Click to expand...

Control theory stuff - i've no idea about that, always sounds tricky to me. I'll stick to audio DSP :)
 
Ok, got the parts to attempt the reballing. The part turned out to be 0.76mm balls with 1.27mm spacing, so one of the larger masks worked for this.

DSlurg8.jpg


hdbN8Yt.jpg


The heating pad worked for this, 240C and they melted beautifully, and gave a decent looking chip.

Re-attaching it to the board went well, it was a little tricky to make sure I got enough heat into the chip and board, but it snapped to the right spot and felt good.

Switch on, exactly the same problem as before :(

So, i've now tracked down a possible replacement processor, so i'll be trying this. Of course, it's possible i'm chasing the wrong problem, but I don't think so. This is very much a learning exercise for me, so it doesn't really matter too much if I chase my tail, as time spent diagnosing is really the aim not the resulting fixed component.

I think I have the technology to reball stuff now, it's not totally easy to get enough heat into the board, especially if it's populated (as this one is). Desoldering the surface mount components on the underside would probably be the way to go to make this smooth, as there's only a few capacitors there to smooth power rails, so this is easy enough to do given the small 30x30mm area of the heater. I'll try this next time, actually, i'll try this to get the chip back off the board.
 
impressive stuff,

Have you considered heating the chip from the top as well. A low airflow heat gun can work well with a hand held IR thermometer (chip tops are pretty good black bodies).

These are pretty good when they are in stock:

hobbyking.com

Digital Heat Gun (EU Version)

Buy digital heat guns online at HobbyKing for the lowest prices. This quality digital heat gun is a great tool for desoldering and shrinking heat tube.
hobbyking.com hobbyking.com

Someone else must sell them too, I’ve just not looked hard for another seller.
 
Puggie said:
impressive stuff,

Have you considered heating the chip from the top as well. A low airflow heat gun can work well with a hand held IR thermometer (chip tops are pretty good black bodies).

These are pretty good when they are in stock:

hobbyking.com

Digital Heat Gun (EU Version)

Buy digital heat guns online at HobbyKing for the lowest prices. This quality digital heat gun is a great tool for desoldering and shrinking heat tube.
hobbyking.com hobbyking.com

Someone else must sell them too, I’ve just not looked hard for another seller.
Click to expand...

Yes, i'm using heat from above as well as from below using a cheap hot air gun, but i was trying to work out how to more accurately and reliably know i've got the solder pads all connected, reducing the risk of burning the board.
 
An update.

Having decided that since the exact same failure occurred having reballed the chip, either I was chasing the wrong problem, or the chip itself was the issue. I decided to take a chance on some old stock from a chinese supplier, which turned up today.

I soldered it on, heat pad below, hot air from above, and this went surprisingly well - I think i'm getting better with these larger chips.

PxXcyjd.jpg


Replace in the instrument, attach to a bench supply to catch any shorts, and bingo, it powers up:

TbzCRCd.jpg


Unfortunately, it's not behaving at all - I can't get a sensible trace on any of the channels, but the UI is working, the clock is keeping it's programmed time, and there are no glitchy artifacts on the screen, so i'm pretty sure the processor is installed well and is functioning. Ok, it's possible at an outside case that some important pin is not connected that happens to *only* affect the acquisition side of the instrument, but that's kind of unlikely...

Here's the board:

w0kNLLU.jpg


There are 4 voltage regulators bottom right, only one is needed for the 3.3v rail supplying the CPU(s) which are now working, so it's possible that one of the other regulators is dodgy, so that's probably the first step.

It's unlikely that all 4 of the hybrids have suffered the same fate, so i'm going to rule out the 4 ceramic packages in the metal casing bottom left. Top right is the CPU and memory i've fixed, so this leaves top left.

This leaves 9 chips above the hybrids, which must contain the ADCs, the trigger logic and the logic to capture display data. I'm going to suggest that the 4 chips directly above the hybrids are the ADCs, the central chip (ADG365B) is the trigger, and the top chip (ADG360C) is the capture/display logic, with memory either side. Saying that, this scope has very little memory, 4k or something like that per channel, so it might be integrated into this chip.

So next up, some debugging and investigation on some surface mount chips with actual legs I can probe and check, so a result! I've definitely made progress though, so that's ace.
 
Nice work. Now it boots has the software got any self-diagnostic routines at all? I’d have expected modern scopes to have various diagnostic and calibration routines tucked away somewhere even if it involves holding a number of buttons at power-up.
 
That's a good point, i'll investigate. There is no public theory of operation manual for these that I can find, so even if it did, it wouldn't necessarily point me at the source of the problem but it might do...
 
On a slightly related note and coincidently I was in a charity shop the other day and there was a bag of metal disks with holes cut out for various chips for doing ball soldering. Curious things!
 
You must log in or register to reply here.


advertisement


Back
Top