• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

TsunamiJuan's Hot Air Rework Shananigans(56k warning)

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

TsunamiJuan

Member
Joined
Mar 27, 2011
Location
Soviet Mexico
So as some of you might know I recently bought a Hot Air Rework Station. For doing reflow soldering of SMD's and Reballing of BGA's. so heres my adventures and mistakes with it. So everyone can point, laugh and learn along with me if they are so incline (probably alot of laughing and the release of alot of magical smoke).


MIVE Repair
First Project On The table is my Asus Maximus IV Extreme, with a destroyed decoupling capacitor. Which has turned into a nightmare to remove.

The top of this cap was broken off accidently due to insulation/defect. As I have said before These things are really really fragile. They are a plastic case with a slight bottom metal backplane and several small plates inside that have an air gap between them. So if you break the top off and moisture, dirt, water hotglue gets in there your screwed and it will probably kick the bucket. If your me you get all 4 of those in there and then give it a shake and claim win (okay not win but fail). As i have learned today the case on these melt at around 170c. Which is kinda stupid cause you have to heat them to around 260-300c to get the solder to melt

The trick to remove one of these is to heat the underside of the board at the same time as your hitting the top with the hot air rework gun from the top. I don't have a Under board heater so I used the extreme overclockers companion, the hair dryer. Throw inside a Mac g4 power cube upside down (wow another use for a mac, and I though this thing was only useful as a trash can, which is actually what I use it for. Make sure to apply plenty of flux around the Capacitor itself. This helps keep the solder on the pads from bridging each other. Aswell as protect the surface from the heat your throwing at it some. Your gonna have to get the temperature in the 240c+ range to remove it. Its alot easier if you can heat the under side of the board close to the same temperature range as to much heat to the top of the cap just makes the plastic melt and crumble to the board. Also helps to throw some tinfoil over the VRM's and other stuff on the surface where you are blowing the hottest air. To prevent desoldering them or moveing them in the process.

A few shots of the underboard heater setup I rigged.

underboard_heater_1.jpg

underboard_heater_2.jpg

After much work and the need to crank the hot air rework station up to 500c to get enough heat to transfer (this board has alot of mass so it acts like a giant heatsink, this component also is in the middle of the power planes) I removed remains of the decoupling cap. Which I had pretty much Mutilated during the process. Also ended up tearing part of one of the pads by being hasty and aggressive, not enough that it should be a problem though.


Before (shot of the damage cap under the microscope)
Decouple cap_s.jpg

After - but before cleanup
pads6_s.jpg

Torn Pad
torn pad_s_small.jpg

Now I just need to finish cleaning up the pads and attempt to put the new capacitor in place which I sure is gonna end up being difficult, if i dont kill one or two in the process (these cost about 4$ a Cap, not to bad but not great either)
 
Last edited:
So it took me about an hour to clean the pads, a combination of not enough heat from the iron, being extremely careful and bad thermal conduction due to the pads from the Capacitor still being stuck onto the board. It took alot of work to get the small pads that where still stuck to the board off, due to a combination of problems.

Gonna have to order some larger tips for my iron the .2mm pencil tip will not work for this on such a heavily copper clad board. With a contact area that small its only getting a thermal transfer of 240c on intial contact with around 270c after some solder gets in there, The setting for the iron's temp probe is 300, if you check the temperature further up the tip its accurately maintained. I checked the temperatures with a K type thermocouple and a calibrated thermometer.

Also decided not to try to solder a new cap on there this weekend. Decided to do it the right and easy way. Ordered 10grams of low temperature solder paste for reflow soldering from newark. This stuff has a low melting temp of 170c. This should make resoldering this to the board an easy process. Sadly it wont get here till the end of next week. Stuffs really exspensive to :\ Cost 25$ for the tube, and another 10$ for tax and shipping. Stuffs suppose to be pretty nasty as well and your advised to not breathe it or let it contact your skin. It has to be kept in the fridge or it degrade and cure or maybe the activators just break down. The stuff has a 6month life span.

Heres a picture of the pads after after the cleaning, theres still a bit of one pad stuck to it. I am not sure if i want to try to remove that one, since its in close proximity to the pad I tore, and there is the posibility of catching the torn pad in the process.

pad_clean1_s.jpg

Also heres a picture of the underside of a new decoupling cap. The dimensions according to the picture are 12mm high, 16.5mm wide. I think this was around 80x magnification.

oe907 bottom_s.jpg

As you can see, not alot of room for error, and with the temperature sensitivity of the plastic case. Low temp solder paste is really the best option. Plus its EASY.

Maybe I will tackle Desoldering some Ram tommarow, and Reballing it in preperation to reinstall it on a video card :D.
 
This is going to be a very cool thread. I'm looking forward to further updates. As a side note, Lowes has some solder paste that I have used before. It turned out to be pretty good stuff. Melting point of 140 or there there-abouts. Like 5-6 bucks a tube, I think. It's called Solder-it. But to be honest, it may stink I don't really know alot about the different types of solder needed, and what have you. Good luck on the RAM
Edit: I used the Solder-it to fix a resistor on a GTX 580 and it worked pretty well. Held strong and had nice silver finish to it.
 
So I did some work with the ram over the last two days. Pulled the heatspreaders off one of my spare sticks of OCZ reaper DDR2 1066. had no problem removing the heatspreaders off the reapers, not sure if the PCB survived the task I didnt bother checking it afterwards. The Fiberglass certainly flexxed something stupid while warm, a deflection of almost two inches. I tried a different technique on the other stick of ram i had, since it was already pretty much dead. I just grabbed the heatspreaderand used a pad scraper on the adhesive. This worked fine on one side, but ended up pulling two chips off the otherside.

Ripped_off_ram_pny.jpg

Ever wonder how good the thermal tape they us is under your heat spreaders? Well on these pqi sticks aswell as on the OCZ reapers the tape seems to be like 80% adhesive. Not to mention look they left the normal sticker manufacturing sticker on on under the spreaders. Probably charged more for that cosmetic spreader in the case of the pqi.

Ripped_off_ram_pny_with oem tag under spreader.jpg

oemtag.jpg

You can tell that I tore some of the traces off the pcb in the process of the two chips that got ripped off.

ddr2 pads 20x dirty_s.jpg

Heres a few shots of one of the pqi chips pads, after cleanup, you can tell that I was a little to agressive and might of tore the pads off if i was any more aggressive.

ddr2 pads 20x_2_s.jpg

ddr2 pads 190x_s.jpg

After that madness I actually used the rework station to pull two chips from different sticks, one from the pqi, and one from the OCZ sticks. Starting to get the hang of this, its also alot easier with the smaller pcb's. Took me roughly a minute and a half to pull each chip off. Could probably of pulled more than one chip off the pcb's at a time but there was no reason to at the time. You can hear the flux under the chips pop some as it releases. Than I just picked up the chip with a suction pen, and it came right off the card.

heres a few underside shots of them after removal probably got them a little hotter than they need to be, but not hot enough to cause any damage, roughly 240c which is still safe. Probably only need about 210c. What can I say i am learning still.

pny chip which is 17mm x 10mm
pny_ddr2_desoldered_dirty_s.jpg

ocz chip 11mm x 11.5mm
ocz_ddr2_desoldered_dirty_s.jpg

The chips on the video card that i am trying to replace are samsung and 13mm x 9mm. The pins appear like they should all have the same pin out
Heres a look at the ddr2 stencil made for larger pinout of some chips but should work fine with these 11 row chips.

ddr2_stencil_s.jpg

I ordered some more tools today aswell, got some suction based vises coming, and some helping hands coming. Tired of chasing small stuff across the table and burning my fingers ontop of it. Still need to order a bga stencil holder. Aswell as some extra small tipped tools, for cleanup of the balls on the chip surface. So might be about as far as I can for the time being.
 
So in essence, you use the stencils to make your balls and place your chip or what have on the stencil, add some heat, and you have a re-balled chip?
 
Yeah theres two types of stencils, theres direct heat stencils like this one, and just placement stencils. Direct heat stencils get clamped over the chip while being reballed. Placement stencils have a small gap between them of a super small amount, so that they dont get flux on them, and enough that the balls down escape between them and the chip.

Once you have the stencils in place you just pour some of the balls on, make sure all the holes get covered. For placement stencils you have to reposition the balls a little after removal, but you save a bunch of time cause you dont have to clean the flux off the stencils between uses. Once you've got the balls in place you apply some heat with the hot air gun. it takes a little bit of time for the them to heat up to a point that they will let enough to bond to the pads. After thats if its a direct heat you pull the stencil, and its ready to be positioned on the pcb and heated again so that it bonds to he pcb.

Its really not that complex of a procedure, just alot of really really fine movements involved. The ddr2 uses .45mm balls aswell, Which are just about as small as they get, there is some stuff out there with a .35mm ball i think. I dont want to even think about how much of a pain that stuff is to work with.

I'll try to post some scale pictures at some point of the different ball sizes in contrast to a dime or something.
 
So its almost the weekend and about half the stuff I ordered got here today. The stuff from Hong Kong, should show up tomorrow :D. Got the solder paste I had been waiting for so it was time to crack into the MIVE fix again.

solderpaste.jpg

This is surface mount solderpaste btw, For manufacturing and RMA rework. Its 62 tin /36 lead /2 silver , which is what most high precision surface mount work seems to be done in (from what I can tell). Probably would of been a ton easier to work with if i had a needle with it, as you only need a very very small amount, and in very very small areas.

One of the hardest parts was getting the component down without spreading solder all over the place and since there are 3 pads we need to solder in 4 places its alot more difficult to get into position properly. Not to mention Cap Fail (aka me). Had damaged the solder mask quite a bit during cleanup. Live and learn right. So I put down a really thin layer of solder. guessing at about how much i needed. I then heated it so that some of the surface stickyness disappered from the solder and it had dried a bit(this particular solder has no problem going through multiple heat cycles/liquid phases, some solder paste will not do this apparently). At that point I hard marked the side of the package with a sharpee to help me know where the insulated parts where between the pads on the decoupling cap, So i could align it correctly on the pads. The marking it helped a ton otherwise it would of been a horrible guessing game.

Once I hit this point I once again hooked up the ghetto underboard heater. Then prepared to heat the top with the rework station. About half way through the heating process my underboard heater (aka hair dryer) died. I had enough heat in the board at that point though that I was able to finish the process. Sadly I made some big mistakes in the process. I got the top to hot. Resulting in the top of decoupling cap melting some (had a figure this might happen). Also because of the precision nature of this work I decided not to tap/wiggle the chip, to get air pockets out from under it and help it settle. But with the top melted its probably a good thing I didn't or I would of probably killed Another cap.

From 2-4x manification it looked like I had a good bond. So I decided to let everything cool and check it out.

new_Tokin_1.jpg

new_Tokin_2.jpg

After it cooled I took a look at it again mostly by eye and figured it was on and good to go. So I pulled out the volt meter, fixed the couple pins I had knocked partly out of whack by the socket cover. Then plugged the board into a power supply and gave it a test. To which I got a resound FAIL. Though in an acceptable form. No magic smoke was released here so that is good. I still thought that cap as soldered cleanly at this point so I hooked up my laptop to the board and give it another check. (I had suspend power to the board and ROG connect functionality, but just no power on state).

At that point I figured that I had killed something else around the area of work. Bobnova suggest a way to test the stuff around the socket and I put that into play. Luckily everything in that area seemed to still be functional. :confused: so it was back to the drawing board to try to figure out what was going on.

At that point I decided I would throw the board under the microscope to get some pictures of my proud solder job. A herp derp later and OOH looky, one of the pads is floating free, how wondifferus. Turns out my solder job was FAIL. While it looked good to the naked eye. It was ineffectively done, and quite dirty once magnified.

new_tokin_pads4_s.jpg
Thats our culprit right there atleast the main one. The same pad that I tore partly, and destroyed a good section of the solder mask around during the cleanup.

Some shots of the other joints
new_tokin_pads3_s.jpg
new_tokin_pads1_s.jpg

So what have I learned today.
- A good smd solder job at this level is reliant on good presoldering cleanup.
- Allways check your cleanup job under the scope before you resolder.
- Solder paste looses about 1/3 of its mass or more as it melts. Just cause it looks like you have just enough to do what you want if it was already a liquid probably means you don't have quite enough.
- if you ruin the solder mask, use some conformal coating, solder mask, or Nail polish to repair it. This will take care of those Worrys about bridging contact points.
-Precision tools like picks, and ESD rated tweezers and other specilized smd tools make all the differance when it comes to pad cleanup and mounting.

i'll take another crack at this tommarow, gonna try to just use the soldering iron to get some solder under that pad first without melting the package. If that doesn't work or I get ambitious i might just pull the cap, fix the masking underneath which I didn't do today. Since I didn't figure out how to repair it while working on it. Also the rest of my tools will be here, New tips for the iron, small diameter desolder wick, more flux, and some ESD surface mount tools. All of which would really help me out in doing this.
 
So quick update, Turns out that you cannot start a p8p67 board without a processor in it. It will not go into an active power state till it has checked the microcode on the chip. So After updateing the bios remotely over ROG connect so that it would be compatiable with the celeron g530 i bought as a sacrifice. I gave it a go, and shes alive :D.

I am pretty stoked, Nice to have my other MIVE back, Gonna burn it in overnight to make sure it keeps running and is stable before i throw an i7 in it.
 
So After last weeks victorys over destroyed and broken electronics. IE repaired the MIVE, then fixxed the amd e350 mobo/cpu, and last but not least finally ordered parts to fix my old lcd monitor, and did the work for that. I have been feeling semi triumphant. Thats all changed since my reballing supplys have shown up.

When I first though about doing this, It honestly didnt look that difficult to me. It was like most things, a matter of having the right tools and knowing how to do the work correctly. Well I pretty much have all the right tools at this point. Even the majority of the technique isn't to bad. However even if I didn't have arthritis, this would still be a PITA.

Here is why, the 0.45mm solder balls tend to hold a charge. They stick to everything. They stick to the container they come in. They stick to the outsides of the tweezers. They stick to the stencils. They stick to each other. If thats not enough slight amounts of airflow including the hot air rework gun blow them around. Even out of the stencil pockets :\. heres a great example of just how small they are without using the microscope to show you. Heres a dime, and a vial of 25,000 solder balls 0.45mm in size.

reball2_1.jpg

So despite the static fun, I gave things a shot (really the static fun is the frustrated point i am left at right now).

I clamped the first chip into the 90mm reballing station and gave it a clean up. These chips are just so small that there just isn't a good method of holding them or mounting them on something to hold them still. Unless you want to leave them stuck to the heat spreaders. To which they are almost impossible to get back off again, without cracking the chip in half. With the new tips for the iron, it takes me a whopping 30 seconds to clean the pads on these chips. Takes me longer to get them mounted so i am not chasing em around the table.

reball2_4.jpg

Once I finished the cleanup i gave it a light brushing of flush, and decided to use the direct heat stencil clamp/reball station. So that i could hold the stencil in place while dropping the balls on it. Which was fairly painless.

reball2_3.jpg

reball2_2.jpg

I threw the little thing in the vise. Just so I wouldn't be chasing it around the table aswell. From there I started pouring the balls onto the stencil, this gets a decent amount of the holes filled, but there are allways those stubborn ones that just wont fill. So you have to go back through with the tweezers under a magnifying glass. You then get to spend a bunch of time chasing the balls around. Making sure they either get into there assigned holes or getting ones that are stacked two high or stuck to each other off the stencil. Touching them with anything besides the end of the tweezers tends to pickup over half of them right out of there slots in the stencil. They even stick to the ends of the tweezers. But after not to long I was ready to throw some heat at it and hopefully get them to set.

Well after the first try i only got about 7 balls to bond to the chip. So I added more flux thinking that was the problem cleaned the stencil up. Filled in the balls again, and gave it another shot. This time i spent more time giving it the heat and watched it closer under magnification. In the process the air from the hot air rework gun actually blew several out of the holes, The flux also ended up floating them out of the holes. (Insert angry noises here)

So try 3, Less flux, more of the same. Got all accept a few here and there to stick this time. So at that point i figured i would just drop them on without the stencil.

chip_1_s.jpg

So it was back into the 90mm reballing station. To place the missing balls one at a time. Originally i though i would be able to put them all down in one shot and be good to go. Not that case the air would blow them out of position. (the slight breeze that there is :'( ) . So it ended up taking me a bit to get them in position and then hold the hotair rework gun and help poke them to the right spot with the tweezers a little while doing it. But finally we had success. One fully reballed ram chip, with seven more to go :(.

with flux still on it
chip_1_1_s.jpg

Cleaned of flux
chip_1_3_s.jpg

This brings me pretty much up to my current position. I removed the other seven chips from the dimm. Cleaned the pads and have started on chip number two for reballing. With still a 1/5 fail chance for ball placement. So I decided to pretty much take a break for a bit. I did a little research and tried some of the things i found. But they actually turned out worse than what i was already doing.

So i think what it comes down to at this point is that I need to figure out how to get the slight static charge (if it is static and not mild magnetism) under control. So maybe some anti static surface sprays, and some grounding of the stencil and stuff so that things aren't attracting each other.

This is definitely turning out to be much more difficult and time consuming that I though it would be. Mostly due to the chip size and solder ball size. Like I said before, if the solder balls where larger .5mm even. I probably wouldn't be having this problem.
 
Wow! Those tiny solder balls don't look like any fun at all. I would probably never be able to even consider doing that type of re-working because my hands shake.
 
I am glad you are doing this. Between the electronics classes I took years ago, xbox isssues, and dead motherboards, I had been contemplating diving into this stuff for fun.
:rofl:

Good work though! There is quite a few videos on youtube. I don't know if any of those may give you some clues as to your troubles. Thanks for the pics and write!
 
Yeah I have been watching alot of the youtube reballing videos. Also been looking at alot of the xbox sites that have people reballing. The thing is that they are all working with a much large ball size which makes things alot easier. There does not seem to be alot of people doing .45mm balling.

However My success rate has gotten up. I have 5 chips reballed now. Usually only missing a few balls after the first attempt. Plus its taking me less and less time. So i am getting there. Might try to finish the last 3 up today. Then I can take a look at putting them on the video card.
 
Quick update, On things.

I finished reballing all the ram today. Still took a good chunk of time but i am getting alot faster and more efficient at it. So tommarow when the good camera is done charging i will take some pictures of them before I mount them on the card :D.

Also I broke down and ordered a IR board heater. For reflow soldering work on motherboards and other high mass smd devices. It should be here in the next few days. Along with parts for the next SMD rework adventure. Replacement sockets for motherboards :D. Also have some evbot's and an evga untouchable coming for continued 3d abuse.
 
TJ streamed some of his work last night and what he is doing is very impressive. Patience is a word you can use when you think of him. hehe
 
That stencil shot make it look like the stencil has tapered holes. If that is the case you can probably fill the stencil on the bench and then slap the ram chip onto that, at which point the balls wouldn't float/sink/etc.
Might just be an optical illusion though.

EDIT:
Looks like slapping the ram on top won't work orientation wise.
 
Alright so gonna start out with the promised picture of the reballed chips. All 8 of the ocz reaper ddr2 1066 chips, that I previously removed from a stick.

reballed_ram.jpg

So anyways yesterday I took the next step and got the radeon out, and put it on the chopping block. Gave it a good bath in Isoprophyl alcohol to remove left over excess flux and dust around the working areas. There was a good deal of fine dust on the card so I figured it was worth it. Plus Ed had previously Zombied this card, so theres some large pools of solder and flux, aswell as some hot glue on it from the volt mod.

Radeon_front pre.jpg

After that was done I put the card in the vice to hold it. Then removed one of the ram chips. To my surprise it was an 84pin ddr2 chip, not a 60pin chip like I had on hand. DOH!!!. Since I already had it off the card I figured i would Just slap a new chip on there anyways and see how it fared.

It wasn't to much trouble to get the new chip on, I lined up the front of the chip with the pin out (front of the chip has a staggered missing ball pattern on it). Made sure I had a really really light amount of flux down aswell. Gave it so some heat and watched it wiggle itself into place. It gave a little shake and aligned itself perfectly.

radeon_with_ocz-reaper.jpg

After that point I decided to give it run after i cleaned up the flux. I wanted to see if it still worked and how unhappy or happy it might be. It powered up first try went into display mode no problem. However you can tell it wasn't the happiest camper due to the ram mismatch.

Unhappy_ram.jpg

On the hole I am counting this a victory, just need to track down the right ram, But totally doable, And the card still works :D
 
Back