Reballing discussion and tips

[Dannyx]

Re: Reballing discussion and tips

SolidState Relay.

usually a block with 4 screw terminals and an led on it.

This must be what turns the lamp on/off. I'm not sure what that screw with the resistor at the end is for - haven't taken it out to analyse. Must be some thermal probe or something, though that resistor looks awfully big for that....

[Dannyx]

Re: Reballing discussion and tips

Hi everyone. Sorry for digging up such an old thread of mine, but I thought I wouldn't open up a separate one to talk about this, since the title would be almost identical, so hopefully someone is still watching...

I had an argument today with one of our super-skilled "engineers" here about a Vaio lappy which has no picture on the internal screen, but works on an external monitor. Now here's this chap insisting the issue is a GPU issue which should be reablled, while I suggested it's a screen/cable issue. Is this even possible ? If there's a GPU fault, shouldn't ALL video outputs be dead ? I'm just curious. Whether we're talking about a laptop where the chipset is handling graphics, or one with a dedicated GPU, can this happen and would a reball fix anything or would it be a collosal waste of time, possible f**cking it up even more afterwards ? ? ? :|

[televizora]

Re: Reballing discussion and tips

It can be a GPU issue. I have seen several laptops with Nvidia 8400/8600 that had image when connected to external display, but no image on the internal display. Also, in about half of the cases there were also artifacts in the picture when they were connected to external display. Reheating the GPU solved the issue in the short run. So, it may be GPU issue. But firstly try with other ribbon cable and LCD display.
Also, if someone tells you that he can reball, most definitely he would heat the motherboard with heat gun, nothing more. There are people that can reball and replace the faulty GPU, but there are few of these. One needs serious equipment, expertise and experience to successfully perform this kind of repairs. And no, partially faulty GPU can have signal on some of it's outputs, but no image on other. No image on both internal display and external display means bad GPU, mobo or hard failure, but image on external display can mean bad internal display or partially damaged mobo.
For example, short circuit near the LVDS connector on the mobo could blow a fuse on the mobo and there will be no image on the internal display, but external will work just fine - IBM T4x series. It's hard to repair computer equipment without a schematic. Of course, there is info for older stuff in the web, but for anything recent....

[Dannyx]

Re: Reballing discussion and tips

From what I gather, he ended up reballing the thing and it's working, so it turns out a loose chip in need of reballing CAN cause only one video output to fail...

[Dannyx]

Re: Reballing discussion and tips

Well guys, believe it or not, I'm back on this thread. If you remember, a couple of years ago, I was talking about different techniques for reballing and the state of this operation in our shop. A lot has happened since and through a series of crazy events I won't go into just now, I ended up doing it myself just today for the first time from start to finish ! Yes, that's right, after all these years, I had a go on the T890 machine, since the guy I was talking about apparently lost more and more of his eyesight to the point where the asked me for help because he can no longer see such small details...

Anyway. I wick, I flux, I ball, I flux some more until it's finally time to put the damn thing back on the board. It was all going well until disaster struck: after the 8 minute profile ran I went to inspect the chip and it was a good 1-2mm out of alignment ! For some reason, it slipped to the top-right of its boundaries ! (see the doodles, exaggerated for clarity) WHY ??? I lined it up by eye as best I could, so it sure wasn't perfect, since I'm not a friggin' machine and there's no way to SEE underneath it, but I wasn't THAT far off for sure ! I was expecting it to snap in place on its own and even the old guy told me I should just be "close enough", but it didn't happen - it was a sh!t fest ! Not even close ! What's going on ? Too much flux, unequal heat from the lamp, some problem with the balls...slanted table ??? :| I left it like that over the weekend since it was getting late, but it sure ruined my day.....such a disappointment, since I almost had it...>_> I didn't SEE it happen, but I imagine it happened when the balls melted. Truth be told, I asked G if the lamp should be carefully positioned above the site, but once again, he told me "close enough" is "good enough", which leads me to believe it was off-center and heated up one side of the board more than the other, so the chip was pulled over there instead of equally from all corners...

This is the kind of thing that never seems to work for me but always seems to work for others and it's extremely annoying: components lining up on their own once solder melts ! NEVER been able to do it, even when I'm dealing with QFP stuff which in theory is more manageable since it has actual leads you can SEE ! Imagine how insane it is for me to try and land all those tiny 0.5mm balls without even seeing them ! Even with QFP, 9 out of 10 times, the chips slides to one side, like it's not heating up evenly, even though I try to hit it as close to the center as possible. Either my equipment or my technique sucks...or both....

[diif]

Re: Reballing discussion and tips

You don't know what happened as you weren't watching what was going on!

[momaka]

Re: Reballing discussion and tips

You don't know what happened as you weren't watching what was going on!

Which sucks even more, because now the experiment has to be repeated again and see what goes wrong for the -2ND- time in a row.

A few of the reasons I can think of for this to happen:
- someone bumped into the station or the table on which it is on
- thermocouple wire sitting on the chip? Maybe it tugged onto the the chip? I've seen this happen when doing RAM, so I never place a thermocouple directly on the RAM chip itself, but only on the PCB close to it.
- new balls on the chip were never fully soldered and as a result, chip "rolled" once the flux heated (especially if the table has a slight incline.)
- G was messing with ya the whole time.

My advice: whenever you do something for the first time on a machine that you have not used before, never let the machine "do its thing" and assume all will be good afterwards.

[Dannyx]

Re: Reballing discussion and tips

A few of the reasons I can think of for this to happen:
- someone bumped into the station or the table on which it is on

There was nobody around but me at that time, plus the table is reasonably sturdy, so even if it did happen, it would have to be a goddamn earthquake to move it THAT far off the target

- thermocouple wire sitting on the chip?

G taped one thermocouple close to the corner of the chip using aluminium tape, but it wasn't touching it, plus it's taped to the side where the chip actually drifted, so if anything it should've STOPPED it from going that way

- new balls on the chip were never fully soldered and as a result, chip "rolled" once the flux heated (especially if the table has a slight incline.)

This is the more likely scenario. After I applied the balls through the stencil for him, G used his old, rickety, nasty hot air station he probably got when the joint opened 30 years ago, to melt the balls in place. The thing barely seemed to put out any air, though I could see the element glowing inside, so I'm not sure if the balls stuck properly, or they were just barely hanging there. They appeared to be shiny and none of them fell off when he peeled off the stencil, yet this is a more likely scenario, combined with the uneven heat from the lamp. Not sure how precise this has to be positioned above the chip....

- G was messing with ya the whole time

No, he was out smoking a fag at that time, so he couldn't have, plus I was the one who positioned the chip, since he can't see properly. He insisted I must've F'ed up and didn't position it right to begin with, claiming there's no way it could've drifted, even though both me and another guy looked at it beforehand...I protested, he kept on going, so we left it at that - the damage is done....it's the cause I'm most interested in. Next time, I'll let him do it so it hopefully drifts again for him

[Dannyx]

Re: Reballing discussion and tips

Here's the actual result. I took it off again, but had to leave shortly after, so haven't had the chance to put it back to see if reoccurs. It's probably a good idea I took the picture because some of the components to the right also got knocked out of place, so at least I know where they should go...provided someone else didn't mess them up even further...

[dicky96]

Re: Reballing discussion and tips

Why do you have all that tin foil on the board? When I use my BGA rework station it never desolders (as in moves) anything nearby, it just removes the intended part, and I don't put anything on the PCB to stop it from happening

[Dannyx]

Re: Reballing discussion and tips

Why do you have all that tin foil on the board? When I use my BGA rework station it never desolders (as in moves) anything nearby, it just removes the intended part, and I don't put anything on the PCB to stop it from happening

Not my idea - the old guy put it there. It's sticky tape usually used for taping ducts together, which makes it worse IMO because you can peel it off with other components and it leaves a nasty residue behind to boot....

The thinking behind it is not to blast heat into nearby components...if that matters at all...

[lotas]

Re: Reballing discussion and tips

Usually, for such purposes, heat-resistant scotch tape is used koptan .

[momaka]

Re: Reballing discussion and tips

Not my idea - the old guy put it there. It's sticky tape usually used for taping ducts together, which makes it worse IMO because you can peel it off with other components and it leaves a nasty residue behind to boot....

The thinking behind it is not to blast heat into nearby components...if that matters at all...

This reminds me of way back in the early days of BGA "rework" and Youtube videos when the "trend" with the aluminum foil/tape started and hardly anyone knew what they were doing. Mostly it was people doing the "oven baking" tricks on Xbox 360 motherboards and video cards... but it spread to laptop boards too. Most competent repair shows will NOT use this technique.

Looks like it's time for old G to retire from BGA work. (no offense to him.)

In any case, DO NOT use aluminum tape or aluminum foil on boards.
As you noticed, indeed it leaves a nasty residue and will usually cause more harm than good (actually, I don't think there is anything good that it does) by moving SMT components around as the board heats up. The only time I ever use it (aluminum foil, NOT aluminum tape) is if I want to protect nearby electrolytic capacitors. But if that's done, don't wrap the foil tightly around the board or any components. Just simply put it loosely around like a heat barrier/wall/shield. More often than not, though, you should never protect any nearby components from heating up - especially with this technique, where the foil is cut only around the part your heating - BAD idea. Essentially, you're trying to heat up a large chip but not the nearby area, and that can cause the chip to loose heat through conduction through board traces. This in turn will usually make one crank up the heat, because it just won't desolder otherwise... and that's the danger - you're running the risk of overheating the chip. You'd be much better off letting everything heat up. Also, you're less likely to cause board warping. Now, if there are any nearby plastic connectors that might melt, then use Kapton tape, as lotas suggested above. It's heat-resistant and won't leave residue when removed.
On that note, if you ever have to remove or install CPU sockets, this is how you do them: cover all of their plastic parts in Kapton tape and then use hot air to solder them on the board. Blasting them directly with hot air will usually cause the socket to melt. IR may or may not melt the socket... but it's still a good idea to cover it.

[Dannyx]

Re: Reballing discussion and tips

Speaking of which, another chap in the shop and I had a debate about this T890 machine and whether it really IS an IR machine or it's simply a quartz heating element at the top. This guy, we'll call him T, recently got himself a plate heater, which looks exactly the same as the bottom part of the T890 machine G has, just without the lamp head above. T slowly wants to get into reballing too, since G's getting close to retiring and he's having more and more trouble seeing stuff up close each day anyway, so it will be up to us younger folks to take over any odd reballing jobs that may come in, even if infrequently.

T practiced on an old graphics card with his new plate heater and removed the chip with his hot air station, by simply holding it above the chip and waiting patiently until it came loose. He even used the K probe of his multimeter to monitor the chip temperature as he was doing this. He claims there's no point bothering with the T890 because it's a plain heater, not an IR lamp, so it's not that much different from what he did with the hot air. It's still a pretty basic machine and doesn't offer that much automation - it just switches off the lamp when the chip reaches the maximum temp you've set, but it doesn't turn off the hot plate too AND at the end of the day it still relies on K-probes that YOU place on the board, so that unavoidably introduces errors. Granted, when T put the chip back on, also with hot air, the board didn't POST, but it could've been broken beforehand, since he didn't check - he got it from one of our other guys who just had it laying around. After having a few tries with the T890 himself, T said he doesn't quite like the way it heats up the chip, which seemed too much to him, so he went on his own quest to find a better method. I however would stick with the T890, just tweak the method slightly: ditch the aluminum tape and maybe dial back the temperature a bit so it doesn't kill the chip. If I remember correctly, the maximum is set to 200c towards the end of the profile. Not having done this before, I just rolled with whatever G said worked for him all these years...

[momaka]

Re: Reballing discussion and tips

Speaking of which, another chap in the shop and I had a debate about this T890 machine and whether it really IS an IR machine or it's simply a quartz heating element at the top. This guy, we'll call him T, recently got himself a plate heater, which looks exactly the same as the bottom part of the T890 machine G has, just without the lamp head above.
...
T practiced on an old graphics card with his new plate heater and removed the chip with his hot air station, by simply holding it above the chip and waiting patiently until it came loose.

I can't fault "T" for ditching the machine and doing the rework "manually" himself with hot air. The difference is, with hot air in hand, you need to constantly watch the temperatures. Thus, you may actually get a better feel of how the board heats up.

Simply putting a PCB on a rework station and pressing the start button is not usually a good idea. The problem with rework machines is that a profile may not match the PCB heating curve very well. Small PCBs can heat up very quickly, if not too quick. But then big boards with thick copper planes may not heat quickly enough.

Granted, when T put the chip back on, also with hot air, the board didn't POST, but it could've been broken beforehand, since he didn't check - he got it from one of our other guys who just had it laying around.

That's a shame.
Should have take a few moments to test the card. It would have been a good way to see if his methods of rework actually work.

After having a few tries with the T890 himself, T said he doesn't quite like the way it heats up the chip, which seemed too much to him, so he went on his own quest to find a better method. I however would stick with the T890, just tweak the method slightly: ditch the aluminum tape and maybe dial back the temperature a bit so it doesn't kill the chip. If I remember correctly, the maximum is set to 200c towards the end of the profile. Not having done this before, I just rolled with whatever G said worked for him all these years...

If it worked for G, I guess let it be. (And it rhymes too )
My guess is, G probably copied the profile from somewhere on the internet, same way as the aluminum foil "technique". Of course, the aluminum foil stuff is crap. But his temperature profiles may be OK. Don't tweak anything until you've tried and verified them yourself a few times. Get some cheap low-end GPUs, like Radeon HD2400/3450 that are sold by the bucket online. Test each one, then play around with it on the machine and see if it damages the card.

He claims there's no point bothering with the T890 because it's a plain heater, not an IR lamp, so it's not that much different from what he did with the hot air.

Well, both Quarts and IR are similar, in that they heat the PCB by radiating heat from their heating elements. Hot air, on the other hand, heats air which then has to heat the PCB - i.e. heating through conduction. Air has poor heat transfer characteristics - that's why the temperature on the dial always has to be much higher than the temperature you intend to heat to. But hot air also tends be a lot less problematic with uneven heating. With IR/Quartz, reflective surfaces won't heat very well, which can mean things heating up differently on the board. But it's easier to deal with when it comes to setting up thermocouples, because the thermocouples don't get too affected from the IR heating. With hot air, the moment you blast the temperature probe with hotter air, it will read much higher temperature. The moment you remove the hot air, it will read the PCB temperature. So it can be a bit more tricky. But if I have to be honest, once you get used to doing things with hot air, you'll probably feel a lot more comfortable doing stuff on the IR machine. So yeah, maybe do what "T" did and practice some with the hot air on smaller chips. Obviously when it comes to the big stuff, you might have to use the machine. But for small GPU chips and chipsets, hot air may be just as good.

[Dannyx]

Re: Reballing discussion and tips

This is the thing T recently got - exactly the same as the bottom part of G's machine if you look at it. The only major difference is that this one reads the temperature internally, whereas G's T890 has the two K probes sticking out the back which you position on the board yourself, so in that respect the T890 is probably more accurate in reading the actual board temperature, irrespective of the machine's ACTUAL temperature (which will naturally be higher). Because of this T uses the K-probe of his multimeter to stick on the board, close to the chip, or even directly ON it, to get a better reading. He then most likely controls the amount of heat by simply moving the nozzle up/down - crude, but it works I guess. No reason why it shouldn't...

Since I'll "inherit" the T890 from G at some point, I'll clean it up a bit and stick with that instead of getting a separate heater like T's. If I REALLY want to try T's hot air method, I can always just turn on just the plate heater of the T890 and go to work with my hot air nozzle at the top.

Another thing I'd like to try is solder PASTE instead of solder BALLZ Don't know how/IF that affects the outcome and the speed of the process...

I'd also like to get a proper clamp to align my stencil with the chip in one go *BAM* just like that, since G's methods and tools at the moment are VERY jank by my standards and he makes it WAY more difficult and stone age-y than it should be, so of course he passed all of this on to us youngsters as-is, when it could be heaps better ! He positions the stencil in-hand which is annoying and is hell on your eyes, you get flux all over your fingers and is prone to slipping - shouldn't have to do that when there ARE tools that help you !