I'm not sure we can assume that flex is the real culprit, it may just be improper solder that cracks with heating and cooling from normal use.

Whatever the reason for the problem, let's hope and push Lenovo to address on a worldwide basis.

Perhaps resellers like Bill can help push this issue, too?!

It is incredible that Lenovo would address a universal problem on a country-specific basis. Very strange. Imagine if Intel only recalled the Pentium I CPU's in the U.S. when the calcuation bug was discovered (about a decade ago)?!

I've had the exact same problem in my t41. Slight movements of the laptop, including adjusting the position of the LCD (raising or lowering it) would cause the screen to go black, and sometimes (but not always) automatically reboot. Other times the screen would just stay black. If something with audio was playing, sometimes that would continue to play for up to a minute or so before cutting off as well. Most of the time, upon rebooting after one of these hangs, I would get a message at the startup blaming the ATi driver for the 7500 mobility Radeon.

Edit: Tried fixing it by putting folded up index cards inside, but it still crashed. Afterward I tried stacking up post-it notes on top of the GPU, but it still crashes as well. However, I haven't received any errors about the ATi driver upon rebooting anymore, so maybe I've got other loose chips as well?

Jeff, is it possible for you to upload pictures of your fix? I dont quite understand what it is you have done. On my T42 the cable from the touchpad is more like a "band" than an actual cable and is not easily moved sideways.

Anyway, heres some results from yesterdays googling:

Apparantly also Xbox and Apple owners have problems with BGA packages on the main boards.

On the link
http://www.macworld.com/forums/ubbthreads/showflat.php?Cat=&Board=UBB4&Number=220598
a guy describes how he resoldered his GPU on an iBook.

Even better; on the link below is a video of the actual process of reflowing the solder on a Xbox:
http://video.google.com/videoplay?docid=-6910389051301515628&q=air%20fix&hl=en


For those daring ones out there (i havent found out if im one yet), i think a combination of the two methods to reflow the solder (the video method and the one on thinkwiki) is better, although NOT TESTED. This is purely speculation, and comments for this procedure are highly appreciated:

1) cut a chunk of metal that has a cross sectional area corresponding to the GPU surface area (something linke 30x34x50mm or so).
2) put it in the oven at some temperature (probably 300 degrees celsius?). (This should be tested first)
3) pre-heat up the mainboard with a hot air gun
4) take the metal chunk from the oven and put it on the GPU for "a while"
5) remove the chunk
6) gradually lower the temperature on the mainboard with the hot air gun, as described in the videolink.

Of course, the two main parameteres in the above are missing: Temperature of the metal chunk and contact time with the GPU. Apparantly, the guy on thinkwiki heated the GPU to 500degrees C for 10 seconds, but im not sure if i can trust his temperature of 500degrees. I think it could be lower in practice.


As before said, this is pure speculations, i havent made any attemps besides trying to melt solder in my oven at 250 degrees C and that worked. But the hope is that it might work, maybe with a modified angle of attack?

Great links - exactly what i had figured i want to try when i get around to it. think i have four or so bad t4x boards with the flexing problem lying around, so time to warm up the heater-gun and experiment

regards,
Jesper James

Here's some hits when I Googled the following terms; all are Adobe Acrobat files:

solder reflow temperature profile

http://www.skyworksinc.com/products_dis ... p?did=2071
http://optoelectronics.perkinelmer.com/ ... rofile.pdf
http://direct.xilinx.com/bvdocs/appnotes/xapp425.pdf

Each document lists the maximum recommended temperature as around 230C. I believe that this is based upon both the melting point of the solder AND the maximum recommended temperature for the electronic components on the board. Note that these temperatures are normally the air surrounding the board in the reflow process. The profiles are essentially how the boards are heated and cooled over TIME to achieve the dual purpose of getting the solder to melt and to not destory any of the components.

Applying a piece of metal heated to 500C or even 300C on top of the BGA package can permanently damage the part. Of course if your system is not functioning correctly to begin with, then maybe you can chance it. If at all possible, I would try to follow the temperature profiles as outlined in the documents I've referenced (i.e., heating the BGA part and some of the surrounding area for some time and then quickly ramping up the temp for a short period of time). This could be done with a good temperature probe, metal shielding and an industrial heat gun (as shown in the video). The critical piece here is the temperature probe. Using the hot air gun by itself, you have no idea what temperatures are being reached. A digital temperature meter using a thermocouple adjacent to whatever you are trying to re-attach to the board is what I would use.

guys i love this thread!!! gonna try it next week. i have a heat gun that you can adjust in temperature steps of 5°C.
any good ideas on where and how to get a cheap thermometer in this temp range?

it is my opinion also that the "heated metal plate" method is not recommendable. you would be putting it right on the silicon which is no good. it would have to sit on the silicone for a while too since the heat needs to be conducted all the way down to the balls.

all the way down to the balls, guys.

All images are clickable links to HUGE versions on another site. Enjoy.

Thank you for this information! I successfully reflowed my bga video chip last night and let me tell you, it was scary!

First off let me point out these two videos of people doing this to their xbox 360.
http://www.youtube.com/watch?v=T5nfmJTQJr8
http://video.google.com/videoplay?docid ... 0fix&hl=en
I’m assuming they did this on the high setting because I had a problem getting my temperature to go above 190 degrees without getting the heat gun right up on top of the chip.

Disassembly of your laptop. It’s the service manual if you don’t already have it. ftp://ftp.software.ibm.com/pc/pccbbs/mo ... 192_01.pdf
If you notice on page 110 of this PDF there is a processor heatsink on some models that covers the video chip as well! I think if we could all get ahold of these it would not only prevent the video chip from getting hot enough to loosen, but help put pressure on it to hold it in place!

The fundamentals of bga reflow http://www.research-intl.com/reflowhand ... attach.pdf

On page 2 of this www.altera.com/literature/an/an081.pdf , there is a fantastic chart that basically covers how you need to go about hitting the correct temperatures. Here is an excerpt from http://forums.macnn.com/66/ibook-and-ma ... usb-logic/
Everyone should read this thread!

Please thank MacInEnterprise for this.

My process:
Make sure you know a round about temperature of your hot air gun. Start by practiving on an old pcb, I used a modem and played around with the temperatures until I was fairly comfortable with removing components.

First I disassembles my laptop completely to make sure that it wasn’t something grounding out on the case. I was able to support the mobo firmly by some plastic on the right middle, and bring it to the edge of the table and press down on the plastic covered area to get it to mess up. Even a small amount of pressure on the top right corner of the chip alleviated this.

Notice that you can have every component off the board besides the ram, proc, and monitor and it will work fine. Just be sure to WEAR A STATIC BAND!

Be careful when you remove the connectors, I pried this one to hard and had to resolder it.

Now, go to a cooking store and buy a digital thermometer made for the stove with a probe that can read up to 450 degrees. Fix that near the chip, but make sure it’s not putting pressure on any nearby components because you don’t want them to move when it get’s hot.

I didn’t use a piece of solder on the top of the chip because when I tested this on the modem, as soon as it formed a ball it went flying! So I choose the ground to the bottom right of the chip. I put a little flux on it and soldered the base, then built up some solder into a cone shape, when the air got hot enough, the cone formed into a ball, but stayed stuck right in it’s spot on the soldered ground.

With my heat gun on low, after heating up the rest of the board. I could reach around 180ish doing small circles over the video card. If I switched it to high the blob would instantly melt and I would need to pull the heat gun about a foot away from the board to maintain 180 temp. I found I had more control close up, but was afraid of the hot air blowing components away.

Insulate everything with foil and peel off any surrounding plastic. I found that just curling some foil over the little plastic connector kept it plenty cool. This pic is from my first unsuccessful run, later, I cover everything but the video chip.

I cannot stress how important it is to have a digital thermometer in there! All in all, I probably had this thing up to 215 C for about 15-20 seconds and it works fine now.

When you are done, you can test without putting the board all the way back together, and try again if you need to. I left the temp gauge on it and let it cool to about 100F before I started stress tests again. Also, DON’T forget the white plastic lining on some of the chips if you took it off. This can short out your board or at least stop it from powering on.

I also put some support in the case under the keyboard by tearing up strips of shipping label paper and folding it, sticky side out. It took three stacks of these to make it the right height so that the spacebar would not stick. Be very careful with the positioning, you don’t want your ribbon cable to melt on the video card, and you don’t want it to be tight or twisted by the pressure, causing premature wear.





Sorry this is long and not very well structured, I’m tired from being up all night playing with my new toy

Note from Moderator: No warning in the subject line = no images. Links only.

[edit] sorry, new here. I just checked the faq and searched the forum. i saw nothing about this in the faq, but i was able to find you posting the same thing in other threads with some added info about 50k size and such. Is this better?

lol danage - you owe me a new t30 keyboard

Anyway i dived into the pile of bad t40 motherboards, found two that i with some confidence could say was suffering from the gpu issue, one real bad - will only start with MAJOR force applied to one corner, and then barely reaches the bios screen. Another board working somewhat, but the lightest touch/flex gives flicker on the screen and then frezes the machine.

I have a cheap bosch paint stripper heat gun, and have lent a watlow thermo regulator and a type 'j' thermocouple - a combo good for measuring up to 500 deg. C with great accuracy.

The nice thing is that i measure the heat gun temperature at the nozzle on medium setting is around 230 C. perfect for this operation...

Did as described in several posts/links above, insulated everything except the gpu with tinfoil, heated it slowly (30-40 secs) up to around 200C, hold it there for 10 secs and slowly down with the temprature.

Added the variation of putting a "D" cell battery on top of the GPU partly to fend off the air to the gpu/memory, but also to put a little bit of preassure on the chip - dunno if it makes sense - anyone??

The first board did show some effect of the operation, does now boot up more often than before but is not 100% stable, guess the problem might have been to large... I might try a second run on this board...

The second board - just boots up and works perfectly after the operation!! i will put it in a machine and see if it holds together in the long run... At least any amount of flex pressure to the chips can't prvoke the problem anymore

Great thread - keep the info comming guys!

regards,
jesper james

Great post willpower101!

Thanks for doing all the illustrated testing and congrats with the successful re-soldering. I finally got my hands on a hot air gun - so will give it a try.

~

Nice pic of bottom of BGA chip for anyone interested.


That sounds scary to me. Are we thinking of the same thing? Those giant heavy batteries that go in flashlights? It seems like that might be overkill and flatten out the solder. I'd also be afraid of the solder spilling over into the other channels. At least you'd definitely know when the solder melted as the chip sinks flat

I thought of using a weight, but the way BGA's work is when you get them hot enough (evenly!) they will re-align themselves. So i don't think the weight is necessary. If you DO want to use a weight, i probably wouldn't use anything that weights more than a few grams. Something like a large washer, nut, or maybe even as large as a watch or a sink aerator. It would also need to be perfectly centered to avoid the original problem.

Also, i'm going to try and purchase a "Long" version of the heatsink off ebay tonight. I think this, combined with added reinforcement, may be the key to preventing a relapse.


Clicking on this will take you to a ebay search with the operators:
(thinkpad) (fan, heatsink) long

thanks for a very informative article, willpower. i hereby no longer support my own idea of using a metal chunk as heat-source.

for your info, i have the long fan as you show in the picture. yes, it helps by making it press firmly onto the GPU, but still its not a permanent fix and chassis-flex still jams the pc. on the other hand, this should be balanced by the risk of doing some permanent damage to the system board using the heat-gun-trick.

my only worry now is that it seems as if everyone is successful in using the heat-gun technique, so i'm sure when i try it ill pull statistics in the other direction. isn't that how things work?!?

nevertheless, congratulations on a successful fix and thanks heaps for a thorough and useful article!

yep it was 140grams worth of battery - distributed over 696 pins (is it not?)

But firstly i dont think this method ever melted more than the outer two or three layers of pins, it would require the heating up of the whold package or enough air flowing in under the whole of the chip which i dont think happens - esp with this battery on top...

But at the sametime i also think the solder breaks from the flexing must be somewhere in one of the outer 'squares' of balls thus making fixing this way poosble.

I've had a long fan the entire time when my problems started. My fix that has worked for the past week and a half has been to put aluminum pieces (not tin-foil, pieces cut out from an aluminum seet). I chose this and nothing other filler because it has got good conductivity (being that my radeon chip needs that extra cooling still). This by itself fixed nothing though, sure the extra fan piece pushes down on the metal which pushes down on the chip..but it doesn't push down nearly hard enough. So on top of that I put about .5cm of velcro strip I had lying around that pushes up aganist the keyboard. It makes the space bar a little more shallow..but I am tired of it turning off and freezing so I can deal with it.

I don't think this is too risky, but who knows. I'm just upset that IBM/Lenovo had known about this problem when I sent my laptop in to get fixed last year and they put in a "faulty" motherboard still. Has anyone considering talking to their state's Attorney General about forcing IBM/Lenovo to fix the problem?

Which reminds me... i emailed the shop where i bought my lap, and they informed me that the product manager (Rene Vinord) from Lenovo Nordic was mis-quoted in the article. Thus, Lenovo/IBM does NOT repair machines beyond expired warranty.

I had a very similar problem with my laptop (T42) a few months ago. It was 2 years old. The screen would go black and it was unresponsive. The diagnosis was a bad motherboard, so I opted to replace it, even though it was out of warranty.
Three weeks after getting it back from IBM I started having screen static....

ok so i guess most people here are convinced that the problem with GPU loosening is due to general flex in the chassis.

how does one prevent this in the future? i dont think it is optimal to re-solder the GPU and keep your fingers crossed that it is "strong enough" to absorb stresses due to chassis flexing.

has any of you found a - or thought of - a clever way to make the chassis stiffer? or prevent the GPU part of the system board from flexing? is there a mechanical engineer outhere?!?!????

Well, it has been posted a few times that it is not prudent to pick up a T4x series ThinkPad by one corner only. Many of the users experiencing this problem had that habit.

Folks, that flexing of the keyboard is wrong. The problem is the too short well soldering of the BGA arrays during manufacturing. There are hundreds of small balls to be soldered underneath the graphics chip. During stress (i.e. heat when GPU intensive things happen) and different thermal properties of the components, the BGA chips can loose a few contacts to the mainboard. Therefore by pressing it or using a wedge you can reconnect again.
Usually this problem is quickly solved with an industrial heat gun that uses extremely hot air to re-solder the BGA. You don't need a new mainboard. the whole process takes a few seconds and gives lasting success.

Unfortunately these hot air resolders are industrial and expensive, so you would need to go to an electronics outfit for help.

If you google for Apple G4 iMacs or Sony GRX notebooks, all with disconnected GPU chips you would stumbel across a page where a hobby technician did some good repairs using a tea light with alcohol. He put it with good physical contact on the ATI chip, heated it up and the problem was permanently fixed.
The main problem is the very large area underneath the BGA chip which makes industrial soldering a problem. The Intel CPUs are housed in a socket, so generally no problems occur because there is enough space for the pins to thermally expand and dissipate heat. Most GPU problems occured with systems that don't use a thermal cooling system (T40, R40 with short fans).

Just an engineer's comment here...

Sorry for this engineer (memory test) to disagree with another engineer, but chassis flex (not keyboard) IS probably a major contributing factor to loose and/or broken solder balls underneath large BGA components.

Not sure what you mean by "too short well soldering" but if you are referring to an incomplete reflow of the solder due to, 1) lack of enough time or 2) lack of high enough temperature or 3) both, then the board should have failed initial inspection, which I hope IBM/Lenovo would be doing, or at some other point in the quality control process. Yes, the GPU gets hot. But hot enough to melt the solder (~185C)? I wouldn't hope so.

Yes, there are different thermal expansion characteristics involved here. A plastic or ceramic encapsulated integrated circuit has a different coefficient of expansion when compared to the printed circuit board (PCB) it is mounted on. Because almost all the components used in today's (b)leading-edge equipment are using surface mount technology (SMT), we are seeing more of these failures. However, because of this usage of SMT, there is now also the concern of mechanical stress (i.e., flexing) causing failures. There are many reports in the forum of certain components lifting off of the board which ARE NOT subject to, nor emitting large amount of heat.

There is also the problem of lead (as in the "leg of the part", not the chemical) contamination leading to the separation of said lead from the circuit board. In this case, flexing of the PCB breaks the loose bond that has occured.

Yes, hot air reflow can solve this type of connection issue. "vlyne" and I were discussing this back on 12/17-12/18 in this very thread. Go back to the second page in this thread and read what I said there.

I have found a good article which explains some of these points:

Article on reflow soldering

This has been a question bothering me for days since it's been fixed. I haven't taken it back apart, but i noticed there was room for the board to move in there. I think possibly putting rubber supports in key places would help prevent this. Mine seems to flex the most from bottom left to upper right, and is much stiffer the opposite way.

As a side note it's still working perfectly; and i'm a ROUGH user. I use this thing in the car on the passenger seat a lot. (not while driving) I tend to treat my [i]personal[i] laptops more like a book than a delicate machine, because that how i need them to be. I've always been like that and never had a problem. Every time i pick it up by a corner or something where i'm not fully supporting it, or rest it on one knee, i wince hoping this won't be the one that does it in, but so far so good!

I LOVE YOU GUYS!!!! I left my T42p for dead, and am about to go to Australia for 5 months. Read this thread, applied sticky notes under my keyboard, and everything works PERFECTLY!!! You guys saved me from having a $3,000 coaster. <3

I have the same problem, and it's easy to fix it, the problem you have are the hardisk airbag, and the only think you have to do is on the Active Protection System Software change to the lowest sensivility and the time remaining to hardisk stops, that fix the problem, temporally, if you unistall this software you'll never going to have this problem, try it, and let me now it works for me

While I didn't suggest the "post-it-note" trick, maybe a donation to the site is appropriate ?

well, I got the new used motherboard installed. Took about 4 hours going pretty slow.

I decided to reinstall XP since the USB ports didn't recognize when I plugged things in. so far so good.

never mind. It is back to its old self again .... oh well. I think I spent enough money on it to buy a used one.....

Sounds to me like a class action issue..........................

IBM / Lenovo is well aware of this problem and warannty or not, the main boards should be replaced at no cost to the ThinkPad owner.

My 2 cents worth

I've had my T42 since June '05, and only this last week I started getting the freezes and shutdowns. I crammed some aluminum foil above the GPU, (I didn't have any post-it notes) now it's back to fine, only temporarily until I can get some time to do the heat reflow... using aluminum should be okay right? Oh, and my T42 has that "long" heatsink (the one that goes over the GPU) that someone posted earlier, the aluminum is sitting on top of that.

Oh and I'm definitely in if people want to complain about this as a group.

stuckj and all,

I also have a 4.5-year old A31. I also went through 2 systemboards, the first failed after a screen issue, and the second I bought to replace it and was fine and just started having the same issue recently. The first systemboard had minor screen garbling issues (funny array of dots on screen but screen is readable - it's only a bit annoying for working). Flexing it by hand changed the pixel pattern, but the board went ok again (no screen issues). Then recently it failed in a worse way (the screen was black on turn on most of the time and wouldn't boot, but 1-2 times I got it to boot). Now that I read the previous posts in this thread, it seems similar, so maybe the board is still recoverable. But I replaced it with a second hand systemboard bought on ebay, which tested fine, and worked fine for a while. Now, the second systemboard has the screen garbling since recently, and flexing the board changes the pattern of dots on the LCD.

HOWEVER, when I put the primary screen on the external LCD monitor, I don't see any of these screen problems (I don't think -- or I can't tell). This is a little strange, since I would think the same video ram is used for either (although the monitor has 1280x1024 res, while my A31 LCD has 1400x1050).

Can anyone suggest if this is likely the GPU BGA contact problem discussed here? The external monitor data point makes it a bit confusing.

If so, it seems A31's may have this issue too. I'm pretty gentle with the laptop, but I carry it in my backpack every day, so it's possible it gets placed under stress once in a while.

stuckj, let us know here how your attempts work out.

Thanks,

Milos

Hi all,

I recently bought a T40 on ebay that was listed as being "fully tested" but was sold ''as is". Yeah, you know were this is heading.... At first it would lock up after a few hours. Quickly it turned into a paper weight and was lucky if it would boot up at all. I didn't have any luck with the post-it-notes. I did find though that the problem with mine was also the top right area of the gpu. I could press down on it there and it would work. I practiced using my wife's embossing gun (can be found cheap at a craft store)on some old boards l had. I also used an infrared temp reader to get chip temp readings. The embossing gun has a lower volume of air that comes out of it compared to a heat gun and it also has a narrow tip. I found discharge temp to be close to 500F at best. The embosser was able to melt the solder and I was able to remove small chips. On the gpu sized chips I could not get the chip over 350F even after 10 minutes of direct heat. I focused the hot air not at the gpu but so that it would blast under that small space under the gpu. I also only went around the top and r.h. sides only..... I did not heat around the whole chip. My T40 now works perfectly. I have to thank everyone that posted here, without this message board I would never have figured this out and would still have a paper weight.

Out of warranty or not