Nvidia GPU chip problem

[Trion]

@katar83,

Thanks for the update on NVIDIA GPUs. The temp for lead free solder is 215 degrees, but as you say 217 degrees is ok. Beyond 225 degrees can permanently damage the GPU. I've been lucky with mostly needing only to reflow the boards and only occasionally change the GPU. What i do after the reflow is to put a copper piece between the GPu and heatsink for proer heat transfer along with arctic silver or similar high quality thermal paste. Also we update the BIOS to keep the fans on. Sometimes you may need to make a larger opening for the air intake. This i neeeded only for one laptop. I seem to be getting around 6-8 DV2000/6000/9000 series laptops a month now. I charge around 90USD to the customer. Its good business ;-) But i am happy that i give them a good solution.

What i do to remove the epoxy glue is heat the board to around 160 degrees and use a scalpel (knife used by medical surgeons) and remove the glue. Then i do the reflow of the GPU.

B.R.

Trion

[bryanba]

I am looking for someone that may be able to help me reflow or reball this chip.. Any help would be appreciated,

Thanks

[Dan Collins]

The G6150 +09 should be ok at 230 max. I see that the 09 date code has more ground planing thus more heat dissipation. But what I do is follow the IPC-7095B for my profile. For those unfamiliar with the IPC, it is the industry standards set by the top people in the field of soldering and reflow and is followed by every major manufacturer in the world. I suggest everyone get a copy, I think it is $50 on the website IPC - Association Connecting Electronics Industries. It is a wealth of info covering every step of the process and how to correct any problem. The IPC guideline for lead free reflow is as follows. (I added some steps to compensate for the RE7500 heating)

Preheat 1 till temp reaches 50C - LH/PH, UH/R/parked
Preheat 2 for 30-60 seconds to 140C - LH/FR, UH/FR/normal
Soak for 90-120 seconds to 217C - LH/FR, UH/R
Reflow for 60-90 seconds to 230C - LH/R, UH?R
Cooling 30-60 seconds to 20C - Fan on, LH/off, UH/off

The best way to cool this quick is to park the upper heater, fan on, and put a shield between the board and the lower heater when temp hits 217C on the downward cooling slope. Slow cooling will make the solder grainy.

Preheat 1 stage is important to minimize board burning. Heating too quickly causes the board to burn.

The most important thing on the reflow stage is to keep the solder at a liquid state for 60 to 90 seconds.

A correct profile should take 4 1/2 to 6 minutes from start to finish. Patience is mandatory when reflowing.

Another useful tip is to use the new Jovy upper heater reflectors (JV-URS). They direct the heat only to the BGA and not to the surrounding components. Verify the centering of the laser. It is not real accurate.

For the above example, raise the upper heater to 3-3.5 or 2-2.5 if using the reflectors. Moving the upper heater up or down will adjust your timing. I noticed that not all RE7500's are created equal with the temps so play with the upper heater height to get the timings just right.

See if this doesn't double your yield.

[Dan Collins]

Slight correction to previous post. Cooling time was not clear enough.

The G6150 +09 should be ok at 230 max. I see that the 09 date code has more ground planing thus more heat dissipation. But what I do is follow the IPC-7095B for my profile. For those unfamiliar with the IPC, it is the industry standards set by the top people in the field of soldering and reflow and is followed by every major manufacturer in the world. I suggest everyone get a copy, I think it is $50 on the website IPC - Association Connecting Electronics Industries. It is a wealth of info covering every step of the process and how to correct any problem. The IPC guideline for lead free is as follows. (I added some steps to compensate for the RE7500 heating)

Preheat 1 till temp reaches 50C - LH/PH, UH/R/parked
Preheat 2 for 60-90 seconds to 140C - LH/FR, UH/R/normal
Soak for 90-120 seconds to 217C - LH/FR, UH/FR
Reflow for 60-90 seconds to 225C - LH/R, UH?R
Cooling 30-60 seconds to 20C - LH/off, UH/off, Fan on and park UH when temp drops to 217C

The best way to cool this quick is to park the upper heater, fan on, and put a shield between the board and the lower heater when temp hits 217C on the downward cooling slope. Slow cooling will make the solder grainy.

Preheat 1 stage is important to minimize board burning. Heating too quickly causes the board to burn.

The most important thing on the reflow stage is to keep the solder at a liquid state for 60 to 90 seconds.

A correct profile should take 4 1/2 to 6 minutes from start to finish. Patience is mandatory when reflowing.

Another useful tip is to use the new Jovy upper heater reflectors (JV-URS). They direct the heat only to the BGA and not to the surrounding components. Verify the centering of the laser. It is not real accurate.

For the above example, raise the upper heater to 3-3.5 or 2-2.5 if using the reflectors. Moving the upper heater up or down will adjust your timing. I noticed that not all RE7500's are created equal with the temps so play with the upper heater height to get the timings just right.

See if this doesn't double your yield.

[linears4]

Originally Posted by Dan Collins

Slight correction to previous post. Cooling time was not clear enough.

The G6150 +09 should be ok at 230 max. I see that the 09 date code has more ground planing thus more heat dissipation. But what I do is follow the IPC-7095B for my profile. For those unfamiliar with the IPC, it is the industry standards set by the top people in the field of soldering and reflow and is followed by every major manufacturer in the world. I suggest everyone get a copy, I think it is $50 on the website IPC - Association Connecting Electronics Industries. It is a wealth of info covering every step of the process and how to correct any problem. The IPC guideline for lead free is as follows. (I added some steps to compensate for the RE7500 heating)

Preheat 1 till temp reaches 50C - LH/PH, UH/R/parked
Preheat 2 for 60-90 seconds to 140C - LH/FR, UH/R/normal
Soak for 90-120 seconds to 217C - LH/FR, UH/FR
Reflow for 60-90 seconds to 225C - LH/R, UH?R
Cooling 30-60 seconds to 20C - LH/off, UH/off, Fan on and park UH when temp drops to 217C

The best way to cool this quick is to park the upper heater, fan on, and put a shield between the board and the lower heater when temp hits 217C on the downward cooling slope. Slow cooling will make the solder grainy.

Preheat 1 stage is important to minimize board burning. Heating too quickly causes the board to burn.

The most important thing on the reflow stage is to keep the solder at a liquid state for 60 to 90 seconds.

A correct profile should take 4 1/2 to 6 minutes from start to finish. Patience is mandatory when reflowing.

Another useful tip is to use the new Jovy upper heater reflectors (JV-URS). They direct the heat only to the BGA and not to the surrounding components. Verify the centering of the laser. It is not real accurate.

For the above example, raise the upper heater to 3-3.5 or 2-2.5 if using the reflectors. Moving the upper heater up or down will adjust your timing. I noticed that not all RE7500's are created equal with the temps so play with the upper heater height to get the timings just right.

See if this doesn't double your yield.

thats my suggestion:

swap new bga to leaded solder first... soldering unleaded type will give a too narrow process window and the chance to burn the chip or the motherboard is too high if the thermal control isn't too much accurate.



the distance between the lower heater and the board is too much, so try to elevate the entire machine with something, about 2,5 cm, or one inch if you prefer...

another important suggestion, and i think, the real key for good resuls :

don't turn on, or put the upper heater in normal mode if the temp of the board si not yet 130/140 degrees... if you turn on the upper heater from the begin, or after a 50 degrees preheat, the upper heater action will heat the thermocouple directly, and when you reach 140 degrees you cannot be shure that the pads are at that temperature, because thermocouple is thiny and exposed and quickly reachs high temp, more that the surface under the bga.

[Dan Collins]

If you can control the temp in the range 217 to 230, then it is best to leave the new BGA as is. The extra 2 thermal cycles of reballing with leaded is too risky in my opinion. And, to get the unleaded balls off, you have to apply over 217C to the chip anyway. The PC board is much more forgiving. Lets hear some more opinions on this!

I have no problem with the lower heater and upper heater spacing achieving the desired profiles. Last thing we want is to ramp up too quickly or ramp down too slowly.

I do admit, the TC can be affected by the upper heater unless you use the reflectors, then you can control that much better. But that is why I use two TC's or just do visual liquidus. There is no way around some heater influence to the TC unless you use reflective tape on it and that is usually space prohibitive or doesn't hold. It is best to check when the solder becomes liquid and then look at the TC reading. That lets you know how far your TC is off.

One more suggestion is to just use the TC as a reference and watch the chip. When the chip starts to drops and aligns, you know it has reached liquidus. This is when it is important to note the drop temp and try not to go more than 10C above that. It should stay in that range for 45-60 seconds total. So turn off both heaters about 20-30 seconds after liquidus to allow a slow cool down from the peak temp. Then wait for the TC temp when that chip drop occurred and.then park and fan on. Then I put a metal shield in between the lower heater and the board to get a quick steady cooling. This way you are just using the TC as a reference and not for accuracy.

That reflow time, above 217C, is the most crucial time for the heater settings. The board too close to the heaters will slow down the cool down from the peak reflow temp. This is the part that is the most frustrating and takes the most patience to control. But once you got it, you have mastered the machine and repairs start working. At least it did for me.

This is not an easy machine to use. It requires a clear mind and multitasking abilities to control it properly. Now if we had better temperature control, it would be easier. What I mean by that is control the heaters like in a reflow oven. Up and down.

[Dan Collins]

One more note. I have mentioned it before, but a video microscope is not a luxury, but a necessity with this work. Especially with a manual machine such as the RE7500. You must have a way to visually inspect the solder joints to insure your technique is correct. Without it you are working in the dark and have no idea if you are doing the reflow correctly.

[linears4]

i agree... this is a machine that needs some experimenting to take the best of of it.. i choose to work on leaded solder cause lead free is general known as a beast...of course it will add more trouble on the solder quality. i think it needs a large casistic to evaluate the results on the long run. If you repair several boards at week, long therm failing (but not too long) can be very painful for your pocket.
on the last, a general methode can't be wrote down, cause even changing the board form factor will add new rules in the game.

i'm looking for a solution to give to this baby a real thermal control. i will see

[Jovy_User]

I have owned a Jovy RE-7500 for about a month now. I have managed to do many reballs on many different chip types (nVidia, ATi, Intel, etc) many with success and quite a few failures (popcorning, temperature issues, etc). With regards to nVidia GPU issues, I have managed to successfully reflow the chipsets from the HP DV6000 (6150) and DV9000 (8400GS & 8600GS) series. I do have some concerns when it comes to simply reflowing BGA chips as I have read about so many cases where the problem resurfaces within a 1 - 3 moth period. I do not use lead free solder as this is usually the cause of the faults in the first place, im sticking to Sn63Pb37. I still face many issues as I have only owned the RE-7500 for a short period of time, but I have become much more confident the more I experiment, it is simply 'trial and error'.If you have any concerns, get in touch with me so I can share my experiences and hopefully help!

[maz99]

please look profiles Strefa Klienta

[Dan Collins]

Reballing with leaded is the ONLY way to go if you are not replacing the chip. A simple reflow is not effective and has high failure rate. If you have a Jovy, you can do it right. Don't cut corners or the product will be back on your bench in no time.

The main issue I have is that damn red epoxy that sometimes likes to permanently bond the chip to the board causing us to use higher temperatures to remove the BGA. I have destroyed too many chips or removed pads because of this problem. I try to remove as much of the epoxy as possible, but sometimes it gets under the chip where I cannot get to and it has a higher melting point than the solder.

Anyone got any suggestions other than dental picks and lots of patience? Maybe a chemical that dissolves the stuff.

[Dan Collins]

One other note on reballing. Use preforms to save time. And put a rigid screen made of aluminum underneath the preform so the preform stays on the BGA while baking. You will get 100% yield this way. Stencils are only useful if you use paste and not balls. And then there is always the problem with keeping the chip on the stencil and the stencil warping. Been there, done that. Which is why I switched to preforms.

[BGA7410]

Has anyone used the solderquick preforms to reball the G6150-N-A2 GPU,If so can you provide the part number for the preform,and any other information you used when orderinng from solderquick i am having a hard time figuring out which preforms on their website to use. Thanks for any assistance.

[geniusmiki]

Originally Posted by Dan Collins

One other note on reballing. Use preforms to save time. And put a rigid screen made of aluminum underneath the preform so the preform stays on the BGA while baking. You will get 100% yield this way. Stencils are only useful if you use paste and not balls. And then there is always the problem with keeping the chip on the stencil and the stencil warping. Been there, done that. Which is why I switched to preforms.

hi, i can't understand this concept.
what you meand for preforms? and "put a rigid screen" how you can do this?

[Dan Collins]

the screen is to prevent the preform from warping or pulling away from BGA while heating.