Reflowing T4x video chipsets

Message

#1 Post by **richk** » Sat Jan 05, 2008 6:59 pm

The other positive about the R series is that the thicker, plastic case is more rigid.

I have had good luck fixing the problem, but I've always hesitated to offer it as a service. I repair motherboards on machines that I am building, because I can go thru the process of recovering from backup disks, applying updates slowly over several days, and occasionally move the machine around while it is running. At that point, I have a good idea that the fix will last. If it is someone else's machine I am repairing, I cannot spend that kind of time, especially on a repair that needs to be done quickly and costs under $100. I suspect that the places who do it for under $100 including shipping and have a 92% success rate define success as having the machine startup and getting paid. I can get that many to work initially, but I have found that maybe 15% of those fixes are not stable. Also, some of the machines with graphics problems actually have bad video memory chips that fail under stress or bad graphics chips that fail when hot. I guess I could let people use the machine for 30 days before paying, but even though I trust people, I'm not sure I trust people that much. I could also offer a 30 or 60 day warranty, but I'm not sure people would trust me that much.

joester · #2 Post by **joester** » Sat Jan 05, 2008 7:38 pm

In a previous life at my current employer, I programmed the rework machines used to reflow BGA's. In the original post, mention of pressing on the components during reflow was made. BAD MOVE. One sure way to bridge most connections together is to even touch the component during the reflow stage. Trust me, the weight of the component is sufficient.

Robbyrobot hit the bullseye. Reflowing or re-balling is the answer.

Reflowing is always the first step, and it's usually successful.

Reballing would be required if someone applied any sort of pressure during the reflow stage (component balls are liquidous and the component itself is floating on the balls). If two adjacent balls happen to bridge, then the entire component is removed, the solder balls removed, and new balls attached to the component.

Reballing takes much more effort. Avoid this at all costs, but it's your money.

Also, I feel it neccessary to mention that special high temperature fluxes are used for this application, as regular flux will burn off very early in the reflowing process. It's also crucial that all flux residues are removed prior to reassembly.

JB'ing the corners sounds good, but a professional underfill would be better if you feel it neccessary.

Joe

#3 Post by **carbon_unit** » Sat Jan 05, 2008 7:45 pm

I have a T42P board here to re-flow and a brand new heat gun. Any advice? I've seen the youtube videos. Should I get an infrared thermometer?

jamiphar · #4 Post by **jamiphar** » Sat Jan 05, 2008 7:59 pm

carbon_unit wrote:I have a T42P board here to re-flow and a brand new heat gun. Any advice? I've seen the youtube videos. Should I get an infrared thermometer?

The way I do it is to fold aluminum foil until you have about 6 layers, fold it once more and use a scissors to cut a hole that's half the size of the chip. Then unfold it once (back to 6-ply) and fit it around the chip. You should have a 6" (or larger) square of foil when you're finished.

I usually have my heat gun set on the lower of the two settings, which is 700 or 800 degrees.

I keep the tip of the heat gun one to two inches away from the chip and I move it around to keep the heat even. I heat it for a good 20-30 seconds, and I bring the heat gun a bit closer for the last few seconds just for good measure. I let it sit still and cool for several minutes and then I test it out.

Listen to joester when he tells you not to press on the chip, I learned the hard way.

#5 Post by **richk** » Sat Jan 05, 2008 8:06 pm

I agree with the earlier post that reballing is too much work for the value of the board. I also agree about the high-temperature flux, that you need to get under the video chips to the solder balls. Based on my experience, I think the heat gun method is not likely to get a lasting fix. It is really important to get the heat focused on the right area and to get it hot enough to reflow the solder without having adjacent pieces fall off or "cooking something". I've seen the video with the guy moving the board around and I think it is really unlikely that the fix will last. If the solder were liquid enough to reattach, pieces would come off. I remember one of my earliest attempts to fix the dreaded "T30 memory slot problem", where the memory slots become detached the same way the graphics chip does on T40. I sneezed and blew about 10 poppy-seed-sized resistors away. At a minimum, you need a temperature-controlled-hot-air rework station, where you can control the force and temperature of the air.

joester · #6 Post by **joester** » Sat Jan 05, 2008 9:52 pm

If you insist on doing this at home (nothing wrong with that decision) you really should have the right tools.

An IR thermometer is a really good thing to have. Look up the data sheet for the target chip, and find out the maximum temp. DO NOT EXCEED IT. As a general electronics assembly rule, three solders, and you're out. One was spent on assembling the chip to the board, so you have two shots remaining.

High temp flux can be purchased, but you'll get way more than you need for one chip.

We use Kester brand "tacky" flux. It's available in several flavors, we use the water soluble formula on commercial grade assemblies.

Joe

baisley · #7 Post by **baisley** » Sun Jan 06, 2008 1:16 am

I stand corrected on the fact of the different video chipset on the T43. As far as the base cover assembly, the T40 thru T43 is identical in material, weight and attached brackets. The embossing is the same as well. ASM P/T 26R7837 is used for the T40's and I think T41's ASM P/N 26R7837 is used for the T42 and T43's. I have a bunch of these on hand. They needed to have a piece of metal inserted in the entire bottom of these darn covers and some fibre reinforced material / embossing (IMHO) but that would add weight.!

Fully agree with what others have said about the R5* I have a few R60's and those things have an entire titanium frame assembly. They are stiff as hell! The base covers are more rigid as well.

Since we have some electronics / rework gurus here (I am envious- This is an area I want to learn.) Whats your opinion on spending the $50.00 to have it bench reflowed versus the heatgun. I am tempted to go the heatgun due to the cost savings. Other than obtaining the correct heat (critical) isn't flowing the solder the same with both ? (Please understand I dont know all of the intracacies, so please dont take offence!

Also, for those that have repaired the T4* boards, what has been your most common error? I have had multiple boards that wont post or show signs of life from either power or battery. The other forums I have posted to have mentioned that these boards are not that repairable and the most common thing they have done is to reheat the entire board to reflow any cold joints. I would be worried that componets would fall off the bottom of the board!

For those that have reflowed, have you heated the entire board like the you tube video or just the Video chip?

I sure am tempted to try one! I tried this on an XBox360 3 light error and ruined the unit, so I am very gun shy.

Thanks for everyone's input.

joester · #8 Post by **joester** » Sun Jan 06, 2008 10:13 am

I'll share my understanding of the process with hopes all can understand.

What we have is a blind solder joint and no real way to access it. Just like soldering two wires together, BOTH need to be at the right temperature for the connection to be made. In the same manner, BOTH connections on the BGA's balls need to be at the right temperature. Also, the component cannot be overheated, the board cannot be burned, and you want adjacent components to stay where they are regardless of whether or not they reflow as well. This is where the heat gun falls a bit short. You don't want copious amounts of air blasting the component. You also do not want the component to heat up too fast. This can cause thermal flexing within the component and damage it. The component and board must be heated together. The process we use heats the board and component up to about 120C, soaks for about a minute to ensure the heat is even, ramps up to about 50C below the maximum temperature of the component, soaks for a short bit, then bumps up the temperature up to about 10 degrees below the component maximum. This process is the starting point of all programs made for the component and board in question. Tweeking is done based on visual observation of the component connections and amount of time the balls are liquidous. Our target is 45 seconds of liquidous, then a gentle ramp back to 150C. Our machines are computer controlled, and thermocoupled to monitor the process. The machine creates a program that will exactly reproduce the heating curve without thermocouples attached for future rework.

There are also people on the rework line who reflow using heat guns. Their process is not preferred, but they get results. Two heat guns are used for the top and bottom to be heated evenly. They "hang" the board off the edge of the bench to expose the bottom of the target area. They use the tacky flux, and watch the component for movement. Yes movement.
When the component balls become liquidous, the component will "settle" a bit. Normally, it's a very slight shift you would miss if you look away for only a second. They also use two layers of high temperature masking tape to surround the component and hold the adjacent components in place. When the tape starts burning (browning and slight smoke), the component is close to relfow.

Now. If you attempt this with regular masking tape, be advised that the glue on the tape does not like heat, and it will become a permanent part of the board. Without exotic solvents, it's a real bear to remove.
Heat must be carefully applied to the component. Too much air will cause the component to shift in it's place risking the dreaded shorting of the balls. You can also blow the component completely off the board.
The process can be done, but I personally would not try it unless I had a scrap assembly to try first that I didn't mind damaging (I had access to the machines, so no reason to try this process). I have witnessed it on several occasions though.

I'm thrifty to the point of being cheap, but I would shell out the $50 if I were in your shoes.

It's your board, time, and money. That makes it your call.

small_thing_on_wax · #9 Post by **small_thing_on_wax** » Wed Jan 09, 2008 6:02 pm

Hello.

Speaking of Infrared Thermometer - witch emissivity do you use to measure the themperature on a ATI 7500 - around 0,93?

And a ATI 9000 ?

Might not make that important, since that doesn't make that big difference, and the "right" adjustment might also depend on the model you're using.

Tim M · #10 Post by **Tim M** » Wed Jan 09, 2008 6:47 pm

I believe ~0.95 is the recommended setting for most objects (at least from the Fluke documentation I've seen).

baisley · #11 Post by **baisley** » Sat Jan 12, 2008 11:52 pm

For those folks who had luck reflowing their BGA sockets, I congratulate you. I just tried on a nice t42 board I had and now its bricked.

Before I tried to reflow, it would at least post (indicator lights on lcd) as well as the fan.. now I have nothing! I followed the process highlighed on youtube. So its a hit or miss thing. I personally, will be sending the balance of my bad boards (2) to have them professionally reflowed.

Robbyrobot · #12 Post by **Robbyrobot** » Sun Jan 13, 2008 3:54 am

I personally, will be sending the balance of my bad boards (2) to have them professionally reflowed.

To where? And at what price? This is the kind of information that never makes it into the forum, and what I originally asked for when I started this thread (which now has been split).

briangig · #13 Post by **briangig** » Mon Jan 14, 2008 6:49 pm

I got my t40 mobo reballed for $75 by http://www.superiorreball.com/ been a few months and its still going strong.

Robbyrobot · #14 Post by **Robbyrobot** » Tue Jan 15, 2008 2:30 am

May I ask how long it took, from shipping to Superior Reball to receipt of the treated mainboard?

#15 Post by **carbon_unit** » Tue Jan 15, 2008 7:06 am

WOW, That is an overnight UPS shipment for me. About a 6 hour drive. I might send my T41P board there.
They reballed and reflowed it for $75.00?

Robbyrobot · #16 Post by **Robbyrobot** » Fri Jan 18, 2008 4:24 am

@briangig:

May I ask how long it took, from shipping to Superior Reball to receipt of the treated mainboard?

I have read some reports about long delays and unresponsiveness of this outfit, and would like to know your experience.

jaysnuh · #17 Post by **jaysnuh** » Fri Jan 18, 2008 8:54 pm

I've just submitted a mail fraud claim and am working on a small-claims case to try to get my iBook G3 logic board returned to me. Dale at Superior Reball has been sitting on it now for almost 6 months (Aug. '07). He is a really nice guy to talk to on the phone - especially if you're a new customer! - but once they've got your board you can pretty much rest assured that the majority of your emails and calls will go unanswered. I've been exceptionally polite and respectfull up until this point but I'm REALLY fed up. When I emailed Dale to let him know my intentions, I gave him until Friday to have a solution "in hand". When I spoke with him this morning, he said he had a replacement for me (wrong mHz!) and when I asked for the correct mHz he responded that he'd "try to get something working". Previously, I've been told that the logic board had been repaired and shipped, that it was completely dead, that they were going to replace the GPU, that they were replacing the entire motherboard with a used one from eBay (!). They are either lying or so completely disorganized as to be unable to provide the service they advertise. Avoid at all costs.

jimmy274 · #18 Post by **jimmy274** » Fri Jan 18, 2008 9:01 pm

jaysnuh wrote:They are either lying or so completely disorganized as to be unable to provide the service they advertise. Avoid at all costs.

Sounds like the most of the services you can get here - in Serbia...

briangig · #19 Post by **briangig** » Fri Jan 18, 2008 11:35 pm

My experience with them was ok. I sent just the mobo in, if you have them remove the mobo from the laptop its $125 I think. I told them I wanted it reballed, I assume that is what they did. they even send a set of x-rays showing the repair done! I got it back about 2 weeks after it was sent in. The only issues I had was they emailed me saying my ibook logic board was ready, I emailed them to make sure there wasnt a mix up, they confirmed they just mixed up the wording, but then I got another email later saying my ibook logic board was shipped...I did get my thinkpad motherboard in the mail, so it was just a mix up in the emails...I was pretty happy with the process, fixed my machine. I have another I plan on sending in to them. I did get the sense he was maybe real busy, or not 100% organized.

Robbyrobot · #20 Post by **Robbyrobot** » Sat Jan 19, 2008 4:51 am

Many thanks for the interesting feedback!

Now it would be interesting to hear from anyone with experience using FirstPhase Technologies in Arizona. Their website appears very professional, but there's no substitute for personal experience.

Robm · #21 Post by **Robm** » Sat Jan 19, 2008 11:13 pm

My T41 has been acting up on me lately and have been debating doing this or just getting a whole new machine since I have had it and it has worked flawless for some 3-4 years now...

jaysnuh · #22 Post by **jaysnuh** » Sat Jan 19, 2008 11:15 pm

It seems as though there are a handful of satisfied customers and a slew of unhappy ones. Just Google Superior Reball and you'll find some interesting threads. (I'll admit to having contributed to one or two of them...) Caveat Emptor.

#23 Post by **carbon_unit** » Sun Jan 20, 2008 7:57 am

OK, is there anywhere else to get it re-flowed that is not so "questionable"?

sjthinkpader · #24 Post by **sjthinkpader** » Wed Jan 23, 2008 1:18 am

Solder used for balls and attachment is called eutectic solder. It is kind of a solid/liquid material. The solder balls break eventually due to the heating and cooling cycles the machine goes thru when you startup and shutdown the computer. Likely ATi hadn't done a good job in verifying the solder joint reliability of their chips. The multi-chip versions are the worst (GPU+2RAM) on a single BGA.

For the motherboard versions before RoHS (removal of hazardous substances), the solder melting point should be about 215-220 deg C. For RoHS compliant boards, melting point should be about 240-245 deg C. This is due to silver and tin replaced lead in the solder composition.

In the JEDEC (Joint Electron Device Engineering Council) solder profiles, there is usually a preheat period at 20-30 deg lower, then raise up to the melting temperature for 5-10 seconds at the melting temperature. Obviously this is very hard to do with simple equipment.

ulrich.von.lich · #25 Post by **ulrich.von.lich** » Wed Jan 23, 2008 6:24 am

sjthinkpader wrote:For the motherboard versions before RoHS (removal of hazardous substances), the solder melting point should be about 215-220 deg C. For RoHS compliant boards, melting point should be about 240-245 deg C. This is due to silver and tin replaced lead in the solder composition.

I see the melting point of lead-free boards is much higher. Is it possible to reflow a RoHS compliant board without damaging other opponents? I hope it's feasible since all T6 series use lead-free boards.

sjthinkpader · #26 Post by **sjthinkpader** » Wed Jan 23, 2008 11:10 am

I think they do have rework procedures but it is now more critical. I think it was Motorola (Freescale today) that wanted the final higher temperature spec and it went to something like 245 +0/-5 deg C. So it is very critical.

Robbyrobot · #27 Post by **Robbyrobot** » Sat Jan 26, 2008 6:14 am

OK, is there anywhere else to get it re-flowed that is not so "questionable"?

Both of us are still waiting. I'm beginning to think there's a lot of talk and speculation, and very little real action and experience on this subject. That's surprising to me, as common as this "flexing" defect appears to occur, but it appears to be the case.

small_thing_on_wax · #28 Post by **small_thing_on_wax** » Tue Feb 12, 2008 7:27 pm

Until now I repaired 3 Boards.
I fired my gas oven up to 250°C. The temperature in the oven was measured with a bi-metall thermometer. Then I stuffed the board inside for about 4,5min. I measured the temperature on the chip surface with a radiative thermometer every min, then put them out to cool down to 50°C. The Videochip saged noticeable. Then i secured all bga chips on the board with a hot sticking pistol (like the newer r51 boards)
All three are stable since 3 weeks. I tested them quite hard... they even suceeded the 0°C test, in which all boards failed which i abused with my hotair. (they were shock resistent but failed to start after they were in some time outside in cold). I stressed them with some games and the good old glxgears and the chips seam to be ok, weven when they get rather hot.
I even "repaired" my broken board which i thought i had toasted with the hotair gun (tested the "oven method" with it, and now it works quite nice again).

So forget about the heat gun! But its not hard to fix them yourself in an oven. But I think a gas oven is better to controll than a toaster oven.

jimmy274 · #29 Post by **jimmy274** » Tue Feb 12, 2008 7:39 pm

small_thing_on_wax wrote:Until now I repaired 3 Boards.
I fired my gas oven up to 250°C. The temperature in the oven was measured with a bi-metall thermometer. Then I stuffed the board inside for about 4,5min. I measured the temperature on the chip surface with a radiative thermometer every min, then put them out to cool down to 50°C. The Videochip saged noticeable. Then i secured all bga chips on the board with a hot sticking pistol (like the newer r51 boards)

Ok, you put in the board after the temp goes to 250C, right?
Then you keep it on 250C, and after 4-5 min, you pull it out of the oven or put the oven temp to 50C?
How did the plastic connectors turn out? Did they warp?

#30 Post by **richk** » Tue Feb 12, 2008 8:05 pm

You mention a hot sticking pistol in your post about reflowing graphics chips. Can you tell me what kind you used and what kind of adhesive? Thanks

Reflowing T4x video chipsets

so far...

Re: so far...

reflowing graphics chips

Who is online