Pretty sure I have the dreaded mobo flex problem....

[JDSO]

...but just checking. I'm afraid I already know the answer....

I signed up here earlier today and found a few posts, including visionviper's excellent sticky'd post on GPU solder reflow.

Bought a certified used T42 from the IBM store Nov-2007 after quite a bit of research -- but apparently not enough. If extended warranties or service pacs were available I wasn't aware of them. All was hunky dory until ~2 weeks ago.

First symptom: After moving the unit a short distance with one hand, the display blanked, to be replaced by colorful pinstripes, not unlike bad 1970's wallpaper. Toggling display state with fn-F7 seemed to resolve the problem.

The problem recurred a few times after that, with fn-F7 apparently curing it each time.

The recurrences have grown more frequent, and also more varied. Sometimes the display "jiggles" back and forth on screen with a frozen mouse pointer. Other times it's a blank screen or various other forms of display garbage. fn-F7 no longer cures it; a hard reboot is necessary. Powering-up immediately after the hard shutdown typically gives a blank screen and unresponsive system; waiting several minutes before power-up has so far (except once) given me the IBM splash screen and a normal boot to the login prompt. The last time I tried that, though, I still got a blank screen. I haven't tried to power up since.

So, my questions...

1) Is my diagnosis likely correct?

2) Best solutions? visionviper's? Others? (Trying the GPU reflow is way above my pay grade but I can try to find someone who could do the work. I've seen postings for people who do the work, I think on this board, perhaps elsewhere.)

3) Any ideas what %age of T42's/T4x's experience the problem? Do new mobo's just grant a reprieve or are they more robust? (If the problem is chassis flex, then I guess a new mobo just postpones the inevitable...)

4) Has IBM been at all responsive? It's pretty clear that it's a widely known issue -- the postings here have pointed me to several places, viz:

http://forums.whirlpool.net.au/forum-re ... 59934.html

This problem appears to be related to a design fault. The T4x series tend to flex a lot and this breaks the solder joints (BGA type) of the graphics chip. Flexing the unit - simply carrying it around or moving it - causes the contacts between the Graphics chip and the motherboard to fail....

Your problem is a KNOWN ISSUE. There are details about it all over the NET (and this problem has occurred with models in the entire T4x series range inc. T40, T41, T42 and T43.) Point that out to IBM Service and firmly ask them to either replace your motherboard (at no cost or significantly discounted cost) or give you an updated notebook. They may not come to the party, but at least you will have the higher moral ground.

Many, many thanks in advance.

[richk]

I've used T4x machines for years and have not had a failure, but I take the precaution not to carry it around with one hand "like a dinner plate". The thin, metal case wasn't designed for that. I have 3 T42 boards coming in late this week. I need to test them out, but assuming they are good, I'll be selling them. I offer a trade-in on your old one. PM if you're interested

[chan_man]

The GPU problem has been an issue with the T4x series for sometime now. I would also agree that you should not carry the laptop with one hand, the unit will flex quite a bit enough to damage soldered components, especially small solder ball components!!

I actually broke a T22M one day doing just that, carrying the unit with my right hand while it was running a video, was in a hurry to get downstairs and by the time I got to the last step (just 6), the screen went black, it would not boot from then on. Ended up buying a used MB from the marketplace.

[JDSO]

Thanks for the replies.

I should be awake when I post.

I actually have no idea whether I was moving the unit with one hand or two. Now that I'm thinking it over, I probably wasn't. Because of where I often sit when using it, I would sometimes move it by grasping it at the front (grasping the wrist pads) and moving it to a table when I wanted to have it off the lap. The maneuver is still a bit like moving it "like a dinner plate" and naturally the center of mass is much further back because of the screen -- which means that if the chassis was not designed to withstand being held at the front, it would indeed flex. And it seems that the mobo takes the brunt of that.

I bought the ThinkPad in part because of their reputations for being pretty robust machines. As a poster in the whirlpool forum I linked to earlier sarcastically noted,

I reckon that if they're treated gingerly (ie not carried around too often, and when carried always carried like a book. Only used on a solid flat surface etc.) that minimises the chances of the affected models encountering this problem.

Unfortunately, I take the view that a NOTEBOOK PC should be designed to be portable.

I guess if I manage to get mine fixed I'll have to remember to carry it like a baby, and pray.

richk, thanks -- I'll PM you about the boards.

[sparta.rising]

You don't have to carry it like a baby, just don't carry it by the lip. Heck even before I heard of the GPU problem I could feel it flex when I picked it up like that. The best way to carry it should be the most obvious, by the back on the battery pack... you know where the grooves are that feel like finger grips.

Also, the cause of the problem is debated. While failure can be triggered by flexing, flexing may not be the cause of the solder points breaking. Instead, its thermal cycling. Unlike CPUs, GPUs are directly soldered to the board. Like the CPU, the GPU gets very hot, and then is cooled off. When metal gets hot, it expands, when it cools, it retracts. This repeated process of heating and cooling causes the solder points to crack.

While this problem has been identified in Thinkpad computers, it happens in all laptops. But most laptop brands don't have the kind of intelligent dedicated enthusiasts supporting them that Thinkpads do. The problem was infamous in iBooks (but most people just bought new Macs). I've heard of the problem in HP's and I have a Macbook Pro that I suspect has this problem (waiting on a power adapter from ebay before I try the reflow).

[Tim M]

^ And while a more robust design would reduce the occurrence of the famed GPU problem, a big contributor is the switch to lead-free solders for RoHS compliance. Such solder joints are known to be more prone to fatigue and to an interesting failure mode via "tin whiskers." Google will yield a plethora of information.

[DaveG11th]

As far as I can see, the situation results from converging vectors.

Decades ago, accountants took over the direction of design efforts by large companies, so costly components had to go. Plus, the entire cost of a component was neatly added up -- the original materials plus any follow-on operations.

One outcome of this was substitution of thermoplastic components for metal ones, as the plastic ones can be formed all at once, rather than requiring multiplle machining steps.

However, a growing collection of engineering plastics left most designers in the dust, so only the best and the brightest could actually come up with designs that met or exceeded the performance of metal. Particularly troublesome was prevention of flex and/or torque in flat designs, somethat that metal excels at.

Then there is solder assembly. RoHS compliance was needed to sell in the EU, so designers had to forgo lead solder for lead-free.

Plus, ball soldering came along, a godsend to people who didn't want the expense of threading tiny wires through holes (by which I mean the expense of high-precision machines whose placement granularity permitted the insertion of pinned assemblies into through-holes), or the bigger expense of sockets, which required not only costly sockets, but more expensive pins, strong enough to withstand socketing.

The brittle nature of balls of lead-free solder was an unknown when it first hit the RoHS big time, though you'd think that accelerated lifecycle testing might have revealed problems.

Add to this the fact that ATI engineers apparently were not kept apprised of the coefficient of thermal expansion of the IBM mainboard. Or they were inept and could not match it. Or IBM changed vendors and the supplier couldn't meet IBM's specs.

So, you get a brittle, breakable interface sitting between two planes that move differently during heating and cooling cycles. And those planes are mounted down on a substrate with minimal resistance to flex. Something's gotta give.

Unfortunately, that something turns out to be us....

I paid $300 for my T42, a fair market price, and I might actually cough up $200+ for add-on warranty after 12/6/08, when mine expires. Actually paying for extended warranty is a new behavior for me. Astonishing.

Still, otherwise, it's a great computer and a huge step up from my previous laptop.

Meanwhile, however, I wonder what percentage of each of the T4x models experienced the problem?