Another one done - Dell XPS M1330 GPU replacement

[Th3_uN1Qu3]

Replaced a GPU today in a Dell XPS M1330 with a brand new datecode '10 chip. This puppy is known to run hot so a new, fixed revision chip was the only way to go.

Factory fresh chip, came with lead-free solder balls of course. Took that crap off and reballed it with leaded solder before installing it on the motherboard. Laptop working great now and will be back to the owner tomorrow.

Enjoy the pictures.

[Th3_uN1Qu3]

Re: Another one done - Dell XPS M1330 GPU replacement

And the last ones.

[sparker1]

Re: Another one done - Dell XPS M1330 GPU replacement

Well done. It would be a good feeling to see the monitor light up with all the graphics. I am interested in doing re-balls too, but the cost is hard to justify for a hobbyist. Your set up obviously does the job. How do you control the temperature (I can see the thermocouple) but the experts go on about preheat, thermal soak, reflow, and cooling times. Do you have a processor to control it or do you do it manually?

[Th3_uN1Qu3]

Re: Another one done - Dell XPS M1330 GPU replacement

Well done. It would be a good feeling to see the monitor light up with all the graphics. I am interested in doing re-balls too, but the cost is hard to justify for a hobbyist. Your set up obviously does the job.

It isn't that expensive in the long run. Just start low and work your way up. A lot of people, me included, get good results with just a cheap hot air station fitted with a BGA nozzle of the appropriate size, a stand for the hot air wand, and some sort of bottom heat source. I use a 41x41mm nozzle, which is adequate for all laptops and the Xbox 360, but you will need larger ones for desktop cards or the PS3.

The only requirements are that you have control over both heat sources, come up with a way to support the board so it doesn't warp, and the nozzle you use for your hot air station must have one or more nets inside for uniform air distribution. Mine only has one net and works fine but i've heard about some which have two or three. Beware that there are also nozzles which don't have nets, those will not do the job.

If you don't want to cook your own bottom heat source, a lot of people use and recommend the Puhui T-8280 preheater. Their IR stations are crap, don't touch them, but that preheater will be fine to use and it's cheap.

How do you control the temperature (I can see the thermocouple) but the experts go on about preheat, thermal soak, reflow, and cooling times. Do you have a processor to control it or do you do it manually?

I've thought about making a controller but i can't be bothered... I'm the one turning the knobs. The airflow control stays at a fixed position all the time, the bottom heater needs maximum two or three adjustments through the whole process, it's only the temperature control of the hot air station which needs a bit more attention.

The bottom heater is controlled by a simple circuit, with a 555 timer driving a triac through a zero crossing optocoupler. A potentiometer controls the amount of time the output of the 555 is "high" and thus the power of the heating elements. The zero crossing bit is important - that way the triac has very little loss since it always fires at the beginning of the mains waveform, and you can get away without a heatsink on it. It's the same driving method they use in variable power vacuum cleaners and stuff like that.

As for thermocouples, i highly recommend using Omega TCs. Unless you're an idiot like me and break them. As i broke mine and they ain't cheap, i reverted to using the thermocouples that came with my $80 multimeters. Well, at least they were $80 when i bought them, they may be cheaper now. I have the Uni-T UT60E and UT70A multimeters which are my main meters, along with 4 or 5 cheapies.

Anyway, the thermocouples are probably the most annoying part of my setup. They're accurate, but it's a pain in the ass placing them and making them keep their position without disturbing something. I place one thermocouple a few mm away from the chip, to one side, and another on the top of the die. I aim to keep both temperatures as close as possible to one another through the whole process. +/-5C deviation between the two is okay, but you can achieve tighter tracking easily if the top and bottom heating are well balanced.

The thermocouples that come with really cheap meters like the DT838 or the DT9205/9208 are going to be fine too if used with a quality multimeter, but cheap meters are going to be way off. So if you have a cheap DMM that came with a thermocouple, don't use it for this purpose. It's better to buy a good multimeter, or a dedicated temperature meter.

You're also going to need a stopwatch/timer of some sort if you plan to do this manually. I use my phone for that. As for the process itself, i've detailed it before, but here it goes again.

Unsoldering a chip:

• I start heating from the bottom (preheat phase). 3 to 4 minutes to reach 150C on both thermocouples. I do not time this part. Anywhere from 2 to 10 minutes is going to be fine, the ramp rate of this part of the process is not critical. I usually apply flux before i start heating, but if using longer preheat times the flux may evaporate and you will need to add more.
  
• When 150C is reached, i turn on the hot air station. Air is always around 3'o clock, temperature starts on minimum. At this time i turn on the stopwatch.
  
• I then turn up the hot air station at a rate of about 2C/s, to reach the solder melting point in 30 to 45 seconds. This is 183C for leaded solder, or 217C for lead-free. I hold at this temperature for another 30 seconds. (soaking phase).
  
• Then i turn up the heat faster, and continuously poke the chip with my tweezers to see if it moves. When it does move, i take the hot air wand away and lift the chip using the vacuum pen. Don't stay above 220C for more than 15-20 seconds or you risk killing the chip, and never go higher than 245C absolute maximum. This is the reflow phase.

If it doesn't move after being above 220C for more than 10 seconds, most likely there's still some epoxy holding the chip down. Hopefully, you did remove some of that before you started! There is a simple trick i use to lift epoxied chips without damaging them or the board, but i'll get to that later.

The resoldering process is largely the same, maybe the times are a little shorter, and holding temperature will be 183C, with peak temperatures of 210-215C as you will always use leaded solder when reworking. After the reflow phase is complete, i turn off the hot air wand first, then the preheater. I leave the hot air wand in place, as when i turn it off, it will turn off the heater but turn on the pump to maximum airflow and keep it that way until the heating element has cooled down. I find this useful. Once the hot air station has completed its cooling process, i take the wand away, as temps will once again rise a bit if you leave it in place, due to the nozzle trapping the heat that is still coming from the board.

Let the board cool down and don't touch it until it gets down to 80C, then you can take it off, remove the aluminum foil/tape, and clean the flux residue with isopropyl alcohol. The flux comes off easier when it is still warm.

The desoldering/resoldering process isn't a big deal really once you have set up your tools. It's cleaning the board and the chip from the old solder which is the difficult part of this type of rework.

[sparker1]

Re: Another one done - Dell XPS M1330 GPU replacement

Thank you for the detailed response. There is a lot of "hands on" information there. I've printed your post out. There's a lot I don't understand (eg. triac?) so I'll be doing more research too. Thanks again,

[mockingbird]

Re: Another one done - Dell XPS M1330 GPU replacement

Excellent job the unique. I always enjoy seeing your work.

[dellxps15]

Re: Another one done - Dell XPS M1330 GPU replacement

good work unique really like your way to do things clean and easy

[Th3_uN1Qu3]

Re: Another one done - Dell XPS M1330 GPU replacement

Thank you for the detailed response. There is a lot of "hands on" information there. I've printed your post out. There's a lot I don't understand (eg. triac?) so I'll be doing more research too. Thanks again,

The triac is an electronic component that can be set to conduct over a certain period of an AC cycle. Think of it as a MOSFET for AC. http://en.wikipedia.org/wiki/TRIAC

You'll see a lot of people using SSRs (solid state relays) to control the heaters, never understood why. SSRs are basically a pair of MOSFETs back to back. Just like mechanical relays, they can switch both AC and DC and can handle high power. But SSRs are quite expensive, and the humble old triac does the job just fine.

In my case, by using a zero crossing optocoupler, the triac always fires at the beginning of the AC waveform, so, to decrease power, cycles are skipped entirely. This does not matter as the heating elements have very high thermal inertia.

[Dimitrios]

Re: Another one done - Dell XPS M1330 GPU replacement

Nice Uniqu3! Thanks for sharing. I have also some experience with reballing and using hot air rework with 3 netted nozzle which is nice but needs more airflow to pump the air! I have a problem of bending motherboards from heating and that's why my reballs on these type pf motherboards are not successful, I suppose. I have also a Puhuit T-8280 preheater.
When you find proper profiles and technique it is somewhat easier to do succesfull reballs but even the smallest detail matters when it comes for success or failure and you must not give up.
I also agree that cleaning solder and apply new solder balls is the most difficult part of the process!

[Th3_uN1Qu3]

Re: Another one done - Dell XPS M1330 GPU replacement

If the boards bend, it means you aren't supporting them properly.

And here's the ultimate stress test on this XPS M1330. FurMark, then IntelBurnTest on top of that.

It does reach some scary temperatures, and the fact that the BIOS reads the temperature of the GPU 10C lower than the real temp, doesn't help either. (For some reason, the GPU temp is reported by the EC as "Chipset"). The highest speed of the fan is at 85C - so that makes it 95C for the GPU - yikes!

Also i fired up RMClock to get the CPU to maximum speed, as even with an official Dell power adapter, it limits to 1.6GHz when it detects maximum load. The CPU in this thing has been upgraded to a T9500, which is quite a lot for this small laptop. If i remove the power adapter under this load, the highest i've seen HWINFO report was a maximum of 77.8W drawn from the battery - that's more than most 15" laptops! Crazy.

But, given that this is a brand new, fixed revision nvidia chip, it should last for a good while. Hopefully.

[Dimitrios]

Re: Another one done - Dell XPS M1330 GPU replacement

I don't have a jig to support laptop motherboards on my Puhui T-8280 preheater. I am trying to find one.

Have you thought putting a thin and small copper shim between heatsink and gpu? It works wonders dropping temperature even by 30C! It also depends on the design of heatsink...

[Th3_uN1Qu3]

Re: Another one done - Dell XPS M1330 GPU replacement

Those temperatures are with a shim in place... It's an aluminum shim, but there's 2-3C maximum difference between aluminum and copper. This little laptop is just crazy hot.