My Experiences Reflowing Videocards (so far 100% success rate)

[SeanB]

Having destroyed a fair number of BGA packages I will say there are never bond wires inside, just the flip chip on the PCB, so the join must be either a solder ball onto the silicon or a conductive epoxy.

I just today took apart a DECT phone, and gently desoldered the RF pcb, using a gentle application of heat from a MAPP torch. The 2 sided SRBP base board ( 1.6mm thick with conductive ink through hole vias and tenting top and bottom) did survive the gentle heating with only minor charring, though I did desolder a lot of the SMD components inside the screening cans of the RF board, it rattled after it fell to the floor, there were loose chips and capacitors inside the steel screening can. I can understand doing this to a BGA will fix it temporarily, though perhaps taking it to 1000C was not the best method.

[wraper]

Having destroyed a fair number of BGA packages I will say there are never bond wires inside, just the flip chip on the PCB, so the join must be either a solder ball onto the silicon .

If BGA have not extremely high terminal count, very likely it will be with bond wires. Especially those with thicker package type.

[janoc]

Just wanted to post that video but crispy_tofu was faster than me
here is Louis Rossmanns response to that video (headphone users be warned audio is kinda crappy and loud):

That's one pissed dude! But agreed 100%.

[iamdarkyoshi]

@all_repair,

My comment relates to flip chip designs. These are what Louis is talking about. The fully encapsulated ceramic and plastic case formats do not use the same bump technology (afaik).

As for repairing a BGA chip through re-balling it....... at hat is only the solution if it has in fact got some form of solder failure situation and not just an internal fault or associated component failure.

I think Louis has explained this in his previous videos on the topic. From what I have seen, Louis is ticked off with people automatically reaching for some form of hot air gun, IR heater, or oven, before investigating the most likely cause of the failure, and taking appropriate action from a position of knowledge rather than assumption. I can relate to that.

I stated, I have no issue with someone cooking their own PCB but I do take a dim view of someone calling them self a component level repair technician blasting chips with heat in the hope of a quick fix and fast buck. As Louis says, it does the industry no good to have such cowboy practices prevalent within it.

Fraser

I usually use "reflows" as a last resort, I first perform the golden rule: "Thou Shall Check Voltages"

I test the VRMs first, then I see if I can reflash the BIOS. If the VRMs show normal voltages and the bios could be flashed and verified, then the next thing to try is a "reflow"

[Rasz]

Having destroyed a fair number of BGA packages I will say there are never bond wires inside, just the flip chip on the PCB, so the join must be either a solder ball onto the silicon .

If BGA have not extremely high terminal count, very likely it will be with bond wires. Especially those with thicker package type.

good picture, now for all the "lets cook it fans":
 Nvidia was caught, and forced to pay for recalls several times, using bad underfill material resulting in cracked balls between die and package due to thermal cycle stress.

http://www.theinquirer.net/inquirer/news/1013947/why-nvidia-duff-chips-shoddy-engineering
http://semiaccurate.com/2010/07/12/nvidia-backpedals-gf100gtx480-underfill/
and so on

Cooking gpu works by softening underfill and relieving stress accumulated from all the hot-cold 20-90"C cycles. What you end up with is a GPU with CRACKED balls, but reformed underfill allowing contact/conducting. It will work for some time if not stressed thermally.

visual explanation:


This is not a fix, this is plumbing with duct tape, or fixing engine by smashing at it with a hammer.


Why does it sometimes work on non gpu stuff? cooking can temporarily heal dead capacitors and make contact in cracked solder joints (thermal expansion) - 'problems starting, works when warm'.
Extreme cooking (450"C heatgun lol) could (if you are lucky not to destroy whatever you are working on) remove tin whiskers and reflow cracked rohs joints, and I could do ballet. There are even cases of "fixing" 8051 controller chips with cooking:
https://www.eevblog.com/forum/chat/how-could-a-mems-gyroscope-get-fixed-by-heating-it-up/msg705547/?topicseen#msg705547
somehow heating this chip to 450"C heals semiconductor structure??

[iamdarkyoshi]

Having destroyed a fair number of BGA packages I will say there are never bond wires inside, just the flip chip on the PCB, so the join must be either a solder ball onto the silicon .

If BGA have not extremely high terminal count, very likely it will be with bond wires. Especially those with thicker package type.

good picture, now for all the "lets cook it fans":
 Nvidia was caught, and forced to pay for recalls several times, using bad underfill material resulting in cracked balls between die and package due to thermal cycle stress.

http://www.theinquirer.net/inquirer/news/1013947/why-nvidia-duff-chips-shoddy-engineering
http://semiaccurate.com/2010/07/12/nvidia-backpedals-gf100gtx480-underfill/
and so on

Cooking gpu works by softening underfill and relieving stress accumulated from all the hot-cold 20-90"C cycles. What you end up with is a GPU with CRACKED balls, but reformed underfill allowing contact/conducting. It will work for some time if not stressed thermally.

visual explanation:


This is not a fix, this is plumbing with duct tape, or fixing engine by smashing at it with a hammer.


Why does it sometimes work on non gpu stuff? cooking can temporarily heal dead capacitors and make contact in cracked solder joints (thermal expansion) - 'problems starting, works when warm'.
Extreme cooking (450"C heatgun lol) could (if you are lucky not to destroy whatever you are working on) remove tin whiskers and reflow cracked rohs joints, and I could do ballet. There are even cases of "fixing" 8051 controller chips with cooking:
https://www.eevblog.com/forum/chat/how-could-a-mems-gyroscope-get-fixed-by-heating-it-up/msg705547/?topicseen#msg705547
somehow heating this chip to 450"C heals semiconductor structure??

Man this is why I love the EEVBlog forum. Such an expansive amount of neat info to learn!

[all_repair]

Having destroyed a fair number of BGA packages I will say there are never bond wires inside, just the flip chip on the PCB, so the join must be either a solder ball onto the silicon .

If BGA have not extremely high terminal count, very likely it will be with bond wires. Especially those with thicker package type.

good picture, now for all the "lets cook it fans":
 Nvidia was caught, and forced to pay for recalls several times, using bad underfill material resulting in cracked balls between die and package due to thermal cycle stress.

http://www.theinquirer.net/inquirer/news/1013947/why-nvidia-duff-chips-shoddy-engineering
http://semiaccurate.com/2010/07/12/nvidia-backpedals-gf100gtx480-underfill/
and so on

Cooking gpu works by softening underfill and relieving stress accumulated from all the hot-cold 20-90"C cycles. What you end up with is a GPU with CRACKED balls, but reformed underfill allowing contact/conducting. It will work for some time if not stressed thermally.

visual explanation:


This is not a fix, this is plumbing with duct tape, or fixing engine by smashing at it with a hammer.


Why does it sometimes work on non gpu stuff? cooking can temporarily heal dead capacitors and make contact in cracked solder joints (thermal expansion) - 'problems starting, works when warm'.
Extreme cooking (450"C heatgun lol) could (if you are lucky not to destroy whatever you are working on) remove tin whiskers and reflow cracked rohs joints, and I could do ballet. There are even cases of "fixing" 8051 controller chips with cooking:
https://www.eevblog.com/forum/chat/how-could-a-mems-gyroscope-get-fixed-by-heating-it-up/msg705547/?topicseen#msg705547
somehow heating this chip to 450"C heals semiconductor structure??

I know Rasz that you have a lot to share.  Thanks.
I am getting ready to do BGA rework but so far there is no immediate need.  Checking the type of bonding that the BGA packaging uses shall be crucial in determine course of actions

[all_repair]

I have been keeping an iMac Intel 17" 1.83GHz that has  a GPU problem for a while.  I have fixed the supply problem, and used it for another a year ore moew.  My plan is to use this to practise on BGA reflow and reball.    I don't have the GPU part number, and could not find it online so far, to check on the type of its BGA packaging.

Just wondering is this GPU worth trying?

[MyHeadHz]

Before an old GPU died I noticed the temp was always high, but no matter what I did it stayed high... even maxing out the fan speed... turning off the fan... nothing at all changed it.  I had chalked it up to a failed temp sensor till the GPU finally failed and I found my culprit! (see attached)

I cleaned out the fan first thing.  I used a Radio Shack heat gun to reflow it, and I never had any more problems from it.  My first reflow! <3

As far as GPU's go, when I was finally ready to upgrade I remembered the reflow experience.  I went on craig's list and found a dead r9 280x for $30 and picked it up (at the time they went for ~$400 USD).  After a quick reflow and some new upgraded thermal paste it's been doing great ever since!  Approaching 2 years now.  That saved me hundreds!  In case you are wondering, it was modded for bitcoin mining and the warranty was void (hence the cheap price).

For both I used an infrared thermometer and heat gun to heat everything to my desired goal temps.  Then, I went through with a soldering iron and hit up a few key points that I thought were especially important.

Sure it isn't ideal, but it works just as well as a new one.  It overclocks very well and holds it with no artifacting.  If it fails again, I'll just reflow.  A lot of people look down on reflowing, but for stuff like this I honestly see no downsides.

I've also reflowed and thermally optimized (internal ducting, thermal compound upgrades, shielding, etc) lots of things... xbox's, playstations, and lots of other electronics over the years (for me and friends).  I originally bought several broken xbox's to test out my skills.  I got all of them working, and to date nothing that I've ever reflowed has failed.  The only down side is the extra space the spare xbox's take up in storage :p

I even reflowed the fuel pump relays on my F-150.  I put 50k miles on it after that with no issues whatsoever.  That alone saved me thousands. It has since been retired to live out the rest of its days on the farm :p  Actually, resoldering would probably be a more accurate term for what I did there.  I sucked the flux, put new proper leaded solder and flux.  The lead-free crap was probably the reason it failed in the first place.

[crispy_tofu]

Apparently LinusTechTips followed up with Louis 

[Fraser]

Good to see

Fraser

[Stori]

Wow nice, I never would have thought that 

[Bud]

Folks, prepair to vote for Motherboard Repair Party in the next elections.

[iamdarkyoshi]

Folks, prepair to vote for Motherboard Repair Party in the next elections.

They would get my vote

[Bud]

I'd like to share my observation about poor thermal paste application that I found on two totally different products, one was a may be 4 years old Dell laptop with Nvidia chip, the other was a PCi-E EVGA card that I revived, also with Nvidia GPU. Both had insanely thick layer of dried grey thermal paste, I guess it is of a popular Silver Arctic brand. It was more than 1 mm thick. I think such improper use of thermal paste may have contributed to the malfunction of the Nvidia GPUs. Once the paste dries out it may become an isolator and the chip just slowly fries itself. Perhaps it may make sense after a year or two pull the card out and change the thermal paste, if the card is out of warranty.

[MyHeadHz]

yeah, too much thermal paste is an insulator.  And yes, if it dries it will also lose thermal conductivity.  However, not all thermal compound "dries" like that.

It is also worth noting that changing the thermal compound only voids the warranty of some brands of graphics cards.  That is also an important factor to check when buying one, especially considering how poor the thermal compound is of some graphics cards.

[Rasz]

It was more than 1 mm thick. I think such improper use of thermal paste

what you saw was quite proper 3m thermal pad, maybe 2-4C difference between that and thin fresh thermal compound.

[Bud]

Could have been a pad on the Dell but certainly paste on the EVGA, squeezed from under the heatsink and hanging from all four sides of the chip.

[Russ314]

First off - my apologies for dredging this up, but I really want to add some pertinent info to an otherwise very informative thread.

Oh, and as for the comments about "cooking" chips with a 450*C heat gun - those guns are barely hot enough to cook dinner, let alone a GPU. I regularly assemble SMD boards using standard Sn63/Pb37 solder, and 450*C is almost the exact temperature I use. I can set the gun to 800*F and blast the fresh SnPb paste until the cows come home, but it's not going to melt. 850*F is where I like to be. I even go as high as 900*F for smaller parts (0805, SOD-123, etc.). I use a large expansion (NOT reduction!) nozzle btw. Without the nozzle, the temp would probably have to be >1,000*F for lead-free solder, and that's approaching the limit of the gun.

If anyone wants to figure out the exact heat-gun conversion ratio, just pour a small amount of distilled water into a test tube, and try to boil it. Common sense will have you start with temps near ~220*F, but you'll quickly realize that 220*F is not even hot enough to burn your hand. In order to make the water boil, you'll have to get the gun to around 400*F. Since water boils at 207*F in my area (Los Angeles), that means the conversion factor is about 2:1. In other words, to create a 420*F reflow profile, you'll have to set the gun to 420*2 = 840*F.


So that's the first thing. More to the point of the thread, I want to explain that I've figured out the (actual) permanent fix for GPU issues. I've been using my Vaio laptop all day everyday for the past 10 years, and during that time, I've had to reflow the GPU about 10 times. It was after that tenth time that I finally discovered the real fix. That was 18 months ago, and since then, my laptop has worked better than it did when it was new. Here's what I did:

1. Reflow the GPU. I won't tell you how, since you already know. I will say that I didn't have my "good" Steinel heat gun at the time, so I had to use a cheap Harbor Freight gun. After a very, very long thermal soak period, I put the heat on "high" and held it point-blank against the GPU for 28 seconds. I was going for 30 secs, but I started to hear sizzling/popping noises... I backed the gun off and held it there on "low" for less than a minute, then put away the gun and wrapped the laptop in several layers of thick blankets so that it could cool to room temp as slow as possible. It's also worth noting that I drowned the board with Kester 951 before the reflow, but I'm not sure if that helped. It hadn't helped in the past. After reading this thread, I guess I know why.

2. This is critical - BEND THE SPRINGS on top of the thermal cover/pad that rests on top of the GPU. Bend the springs upward, so that whatever you bolt on top will press downward with maximum force. On my laptop, there is a large cast metal heatsink (for the GPU RAM) that bolts on top. By bending the springs upward, I made the cast metal part push the GPU pad aggressively against the GPU. This is a big part of the permanency of the fix. I don't know whether pushing on the GPU forces out more thermal paste, which closes the gap, which increases heat transfer from the GPU, or whether the force squeezes the GPU package and closes any internal gaps/cracks. Those are both ideas I hadn't thought of until reading this thread. To be honest, my reason for doing it was that I wanted the GPU to be held in place for whenever the solder melted again. After all, the reason for the GPU failure was gravity. When the GPU overheated, it would DE-solder if the laptop was right-side up, and RE-solder if the laptop was upside-down. So I figured the added spring pressure would counter the force of gravity and hold the chip against the board.

3. Since the cast metal heatsink/GPU cover is dangerously close to the bottom of the laptop, I wanted to make sure it never experienced any mechanical shock when the laptop was moved/handled/etc. This fix was easy - I just bought a $20 padded plastic thing that sits on my lap, and my laptop rests on top of it. This proved to be one of my best purchases ever, because now I can bounce my laptop around all day without ever having to worry about the GPU (or other internals) being stressed or compressed in any way. The pad also provides a constant air gap for the fan intake. It also protects my sperm .


So that's it. Reflow the GPU, bend the springs, and get a base/pad thing. Even a portable kitchen cutting-board would work. I don't know whether it was the springs or the pad that helped the most, but I suspect it was a combination of both. One thing I know for sure is that my previous reflow attempt used the same heat gun, on the same setting, in the same manner (but for only 20 seconds, not 28), and with the same flux, and the repair only worked for maybe a month or two. Btw, I also toyed with all the GPU settings via nVidia's software (e.g. clock speed, frame rate, etc.), but nothing I did affected the GPU temp, so I wouldn't bother with any of that.

Since the fix, the GPU temp has been rock solid. It used to constantly go up and down 30 - 40 degrees, but now it only moves 10 degrees (F). And the fan almost never comes on. Before the fix, the fan would constantly wind up and scream for mercy, but now I never hear it. The only thing that raises GPU temps is watching Youtube vids in full-screen (which I never do). Watching small-sized Youtube vids won't even cause the fan to increase from its idle state.

Hope this info helps someone!
Sincerely,
Your friendly local embedded Microchip engineer turned organic chemist turned couch bum

[Russ314]

I just tested the heat gun in a formal experiment. I'm too tired to post a full write-up, so here's a summary of my lab notes:

Heat Gun: Steinel HL2010E (not recommended)
Temperature Setting: 800*F
Airflow setting: LOW
Nozzle: 75mm "Spreader Nozzle"
Distance from nozzle to work surface: 1.0"
Method of temperature measurement: thermocouple (i.e. a kitchen thermometer, with the probe laid flat on the work surface directly below the nozzle output)

TIME     TEMP (*C)
T=0         24.9
T+30       86.6   <-- The gun reached 800*F at exactly 30 secs
T+60      142.8
T+90      169.1
T+2min   178.6
T+2.5     187.3
T+3min   193.2
T+3.5      205.9
T+4min   210.8
T+4.5     212.6
T+5min   213.9
T+5.5      215.1
T+6min   216.7
T+6.5     218.3
T+7min   218.5
T+7.5     222.1
T+8min   220.3
T+8.5     219.6
T+9min   220.4
T+10min 219.1

As you can see, it's a pretty good reflow profile. Starts to reflow after 2 to 2.5 minutes, and peaks at ~210*C after 4 minutes. Leaving it on for a full 10 mins only adds a few more degrees beyond that. At the 10 min mark, I switched the gun to "cool down" mode, and it took 2 mins 10 secs to cool all the way back to 120*F.

I would like to amend my previous post btw; 850*F is a bit too high. I tried it and it melted the white plastic SMD connectors on the board before it melted the solder. Dropping the temp to 800*F worked perfectly, and the plastic never melted/singed/discolored again.

Hopefully this clears up once for all the myth about cooking/scorching/incinerating/vaporizing IC's with 450*C heat guns.


P.s., I know the pics are convoluted, so I'll elucidate. The heat-gun is attached to the chemistry lab ring stand with two clamps, while the third (articulating) clamp holds a laser. The laser is CRUCIAL for the success of the setup. You really need to know where the center-line of the airflow is, because being off by even a few millimeters will drastically reduce the effective reflow temperature, and it'll also unbalance the air forces that keep the components pinned down. If you move even slightly out from the center-line, the tiny parts will blow around. That's how I settled on the 1.0" nozzle distance btw - I found that the limit for my setup was 0.920", and moving any closer caused my smallest components (0805 resistors) to blow around.

[KL27x]

So that's the first thing. More to the point of the thread, I want to explain that I've figured out the (actual) permanent fix for GPU issues. I've been using my Vaio laptop all day everyday for the past 10 years, and during that time, I've had to reflow the GPU about 10 times. It was after that tenth time that I finally discovered the real fix. That was 18 months ago, and since then, my laptop has worked better than it did when it was new. Here's what I did:

Basically, unless you discover an actual fix like you have, using improvements to the mechanical fixtures, fixing a board simply thru reflow isn't going to fix the underlying problem. Problem with Louis take/rant on it... if the flip chip is failing because it uses a bad underfill..... what is the point of insisting you have to replace it with a genuine replacement part which also has a bad underfill in it and will also fail... If you want to disassemble/reassemble your laptop more than once in a lifetime to keep it operating, you have way too much time on your hands, lol. This makes top 10 list of things I don't want to do with my time. By third time, I will be ready to professionally disassemble laptop with sledgehammer and landfill solution. This is why Louis is so angry, no doubt. If I can't charge $500.00 for just opening a laptop, I'm not gonna enjoy doing it, at all.

[rx8pilot]

It is hard to imagine that it is financially productive to do these repairs as a business, although I guess some people have figured it out. I do a lot of 'in-house' repairs on various pieces of commercial gear that are very difficult or impossible to replace outright. I do a LOT of design, prototyping, and small lot production so I have all the tools and experience needed to do BGA re-work and I would be hesitant to dive that deep into a laptop or desktop computer. Just disassembly and re-assembly takes a long time and the repair can always end up being a rabbit hole adventure.

[Russ314]

This makes top 10 list of things I don't want to do with my time.

I hear that. For me, it ranks just slightly below going to the dentist.

By third time, I will be ready to professionally disassemble laptop with sledgehammer

That reminds me of my first laptop... It was a 1997 Dell Latitude, and the screen broke after only 2 years. I wasn't able to fix it, so I ended up removing the CD-ROM drive and stuffing the hole with 18 grams of perchlorate explosives. I set it off in my backyard and the damn thing nearly vaporized. 30 minutes later, I was walking around on my roof, picking up coin-sized fragments of plastic and metal, and putting them in my shirt (which I had folded upwards and was holding in my teeth). Out of nowhere, a cop car came FLYING up to the front of my house, slammed the brakes, and turned on a powerful spotlight (it was dusk by then), and started combing back and forth across the front of my house. All I could do was freeze in place. Incredibly, the cop (who was focused on my kitchen windows and the bushes around them), somehow failed to notice me standing there, like a deer in the headlights, holding an armful of evidence.

Ah, yes... the good ol' days


By the way, I spent all day yesterday reflowing small surface-mount boards, using the setup I previously posted. One thing I've learned is that attempting to map a definitive correlation between the temperature setting of the gun, the temperature reading of the thermocouple, and the actual temperature transmitted to the board/solder is a pointless endeavor. At the exact point when the solder reaches 173*C (reflows), the heat gun says 427*C, and the themocouple says 105*C. So the readings are all pointless. All I can say for sure is that the 800*F setting seems to work best if you want to reflow with a heat gun.

[rx8pilot]

I use temperature indicating crayons to help me heat a pcb with a reasonable profile. It takes a lot of guessing and complexity out of the task.

Sent from my horrible mobile....

[Russ314]

I use temperature indicating crayons to help me heat a pcb with a reasonable profile.

That's a good idea. What temperature(s) do you use? I just checked Amazon... those damn things are expensive! Some of them are listed for over $125?! For a crayon??

The "Thermomelt" sticks seem cheaper than the "Tempilstik" equivalent. Are those the only two brands? I'm thinking I should get one for 400*F (205*C) to let me know when I'm at peak temp.


EDIT:
Here is the Amazon price for a 463*F Thermomelt stick...

List Price: $181.04
Price: $9.85
You Save: $171.19

I call shenanigans.