I've got another laptop needing a nVidia chip reflow. I'm refining the technique. What's the recommended air temperature? I have heat guns with open loop variable power input. And some with thermal feedback to control the air temperature.
I'm using thermocouples on top and bottom of the board and manually controlling heaters while watching temperatures. None of the heat sources are currently set up for computer control.
I'd like to try fixed heater positions and control the heater power for better repeatability.
First thing that comes to mind is to have the air temperature track the board temperature with the difference set to get the proper temperature ramp at the chip. Then there's the relationship between air flow rate and air temperature.
That should minimize overheating. This has the potential to bloom into a local microcontroller and a computer interface and graphs and and and...
Or, am I worrying too much and should just use fixed temperature air.
Nope. The web is full of instructions written by people who have no idea what they're doing with zero specificity for reproduction of their work. Just because ten people took a butane torch to their GPU doesn't mean I want to try it. There's one video that made a great production out of heating the WRONG chip.
The whole concept of reflowing without reballing is a crap shoot. I don't know a lot about it, that's why I'm asking questions, but, when your balls are fractured, you need ALL the balls melted at the same time on BOTH sides of the fracture and maybe some vibration to make the crack close so surface tension can mend the fracture.
Doing that without breaking something else is gonna require some precise temperature control.
The other end of the spectrum is videos showing jigging the board in a commercial reflow machine and pressing "go". Still no specificity on what temperatures are being measured or what the air temperature driving it might be.
I have a hot air gun with digital temperature readout with one degree resolution. They can't be measuring anything but the air temperature. What we want to know is the temperature of the workpiece. Given that air temperature is the only parameter it can control, the needed info is what AIR temperature should we be using as we measure the workpiece with other tools.
The thermal profiles are well known. It's the process of getting there that's not obvious.
I'm interested in the physics and thermodynamics of the process.
Formatting. Slashes around something put something in /italics/ Asterisks around something makes something *bold* Underlines around something _underlines_ it.
for nvidia chips (some ATI too) reflows on laptops (mainly mac laptops)
10 minutes letting the heat stabilize on the board, then 3 minutes with
7 on air flow control then stop everything and let cool the board. I don"t desoldier the chip, just reflow it (put some flux on the border of the chip)
I have so far some 95% success on some 20 laptops. I began this one year ago, the first ones are still working ...
Recommended by whom? Recommendations are all over the map. Here's mine.
Much depends on the type of solder used for the BGA chips. My guess(tm) is that most BGA's use solder balls which solder at 217C and starts to reflow at about 200C Meanwhile, the rest of the board is SnAgCu solder paste, which starts to melt at about 225C. These temps are from my measurements, not from the solder manufacturers charts and take into consideration my instrument inaccuracies. The trick is to target the oven temperature between these two temperatures so that the BGA balls reflow, but the components on the PCB are untouched.
Getting the temperature even across the board is a major problem. I've made measurements with an IR thermometer and thermocouple probes which convinced me that this was impossible. Therefore, protecting the PCB with an aluminum foil shield might be a good idea.
In the past, I would use my cheapo SMT rework hot air gun for reflowing BGA chips. It would work on the smaller BGA and flip chips, but fail miserably on the larger chips. I gave up using it for BGA repair.
Recently, I purchased a Black and Decker toaster oven at Kmart for about $25 to do some experiments. I reflowed some old PCB's to test the settings and thermal control. Things went well, so last week, I reflowed 14 HP JetDirect cards and 4 HP printer formatter cards. Some photos: All 14 JetDirect cards worked, although it's too soon to determine if the fix is permanent. Only one of the formatter cards didn't work. Not a bad percentage for essentially an uncontrolled reflow.
During the above reflow test, I make quite a few changes to the temperature, timing, protection, etc. What I settled on is:
10 minutes preheat to 215C without the board in the oven. When I open the door and insert the board, the temperature drops quickly. I then reset the temperature dial to 190C and bake for 8 to 10 minutes. At the end I turned off the oven power, but did NOT open the oven door. When the temperature drops to about 100C, I open the door, leaving the boards still in the oven. About 3 minutes later, they are cold enough to touch and are removed.
Not the best temperature curve possible, but it works. What makes it work is that the thermal mass of the cheap junk oven is very small compared to a proper oven with insulation, firebrick, and a proper controller.
Much of my inspiration came from: and various articles on:
Notice that I did not protect the PCB with aluminum foil. I did that with some early board tests and found that it didn't do much except protect some electrolytic capacitor jackets. (I did remove the plastic front panels from the JetDirect cards). As long as the temperature was kept below where the solder paste on the PCB reflows, there were no problems. None of the other key factors seemed to be particularly critical. The bake time was varied from 5 minutes to 15 minutes. Anything over 7 minutes at 190C seemed to work.
Also notice that I didn't use flux. I'm fairly sure some really fluid flux would have helped, but I didn't have any handy and didn't have time to make some. When I previously tested the use of flux, I found that the alcohol base had evaporated long before the solder balls had begun to reflow. In other words, the flux didn't do anything. I did as well using just alcohol and no flux which just cleaned the BGA from accumulated dust and grease.
Several authors have suggested putting a weight on the BGA to help push the broken connections together. That proved to be a problem. The added thermal mass of the weight caused the heat distribution across the BGA chip to change, which probably resulting in good reflow around the edges of the BGA chip, and lousy reflow near the middle. At least that's my guess why a weight made things worse.
Whether all this will work for motherboards has yet to be determined. I'm not even sure they'll fit in my toaster oven. I have about 5 candidate laptops waiting, but the PCB's are still in the laptops. It will take time to extract and retest them. Probably next month.
Good luck.
--
Jeff Liebermann jeffl@cruzio.com
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558
I suspected that one should use the lead-free profiles. ???
I've been afraid to put a whole motherboard in a toaster oven, assuming I could find one big enough. There's lots of plastic stuff/connectors etc on the board. Most toaster ovens have exposed elements. You get lots of infrared and a variable amount of convection heating. That scares me too. Thought about putting reflective foil everywhere but the chip and count on the IR to do it. But it's completely uncontrolled.
My current tooling consists of a bottom heater made from an old TIVO case. I cut a 2"x2" hole in the top for the BGA chip area to sit on and drilled a hole in the side for a power-controlled heat gun. Welded some baffles inside to guide the air from the heat gun to the square hole.
I sit the board over the hole with thermocouples on the underside and touching the BGA on the top. Aluminum foil protects the part of the top that's not the BGA. Flux applied.
Top gets heated with a commercial SMT rework heat gun with precise air temperature control. My air compressor is running wide open and I'm taking all the air flow I can get.
I know what package temperatures I want and have some idea from the thermocouples.
What I don't know is what air temperature I should be using from the heat guns.
Too low won't get it all hot enough. Too high will burn something.
I was hoping to avoid a bunch of experiments to determine it empirically. Would be nice to have a target temperature to start from.
This site is interesting:
formatting link
I hava a Maxtra 852D+ rework station. It's capable of 550W, but the air flow isn't high enough to keep the heater duty cycle high. I have very little actual experience with the unit. And my thermocouples read the air temperature 30C lower than the digital readout. Very small changes in position make radical differences in the measured temperatures.
Yes, if the oven temp controller had the ability to follow a temperature profile. I'm not even trying to do that. Buy coincidence, just shoving in the board into a pre-heated oven, and letting it cool down seems to follow somewhat of a reasonable ramp, soak, and cool profile. However, that's pure luck, not intent. My point is that for reflow (i.e. repair) it does not seem to be that critical. Once I understood what was happening, I soon discovered what was the causing most of the reflow failures: - Too cold during soak. - Too short a soak time. Everyone was too worried about destroying parts by over heating, so they went for the lowest possible temperature and the shortest possible soak time. That's not going to work as I proved to myself by trying some of those recommendations. What I found was that I could soak for quite a long time as long as I didn't get close to the melting point of the solder paste used to solder the non-BGA parts. If I could keep the temperature at about 220C, I could leave the board in the oven nearly forever and it would come out ok. However, there is a problem. Because I didn't have a proper temperature controller, doing that would cause the temperature to slowly climb. The result is that the soak temperature is not constant but rather an increasing ramp. I can live with that as long as I target the end of the ramp at less than about 220C. If you want to do it right, go thee unto eBay and spend about $70 on a programmable oven controller.
Presumably, you could find a sacrificial motherboard to test. Shove it into the oven at the highest expected temperature for a reasonable time and see what melts or burns. Thermoplastic parts will melt. Thermosetting plastic parts will not. You'll find most connectors made from thermosetting plastics because the original motherboards are soldered in a IR oven, which uniformly heats everything to the same temperature. However, if you want to be sure, LOOSLY wrap the connectors in aluminum foil. You want some air gap between the foil and the connector so that air acts as an insulator.
Really? It thought the UL banned that in the 1950's. My Black & Decker has two glass tubes with heating elements inside.
I think you'll find that it's almost all IR with a tiny amount of red in the visible spectrum from the red glow. Convection implied air flow. Unless you buy a convection oven, the IR (heat) is very unequally distributed around the oven. I proved that to myself when I ran my thermocouple probe around the oven to see how bad it was. My guess(tm) is about 40C from the hottest near the center to the coldest near the lower corners.
Note my domain name: LearnByDestroying.com You haven't learned anything until you've destroyed it and then had to make it work.
Yep, but as I found, it works well without the proper controls. Sure, I would like an overpriced oven with the proper profiles, but lacking the financial justification, I used what I had and it worked. Foil works if leave an air gap.
I'll make it simple for you. Hot air does not transfer enough heat by convection. Close in, radiation is a far more efficient heat transfer mechanism. Loose the hot air gun, hair dryer, or flame thrower, and switch to something that resembles an IR source, such as a heater.
Again, lose the hot air gun. I have one of those. It will work with a nozzle that is made for the BGA chip. I only have one size (forgot which one) and that works because *ALL* the hot air lands on the BGA top. That's not the case with just blowing hot air in the direction of the BGA. When I tired that, I found that it took me about 30 minutes to heat the BGA top to 220C. I could do the same with the proper nozzle in about 20 minutes, or with an IR heater in about 10 minutes. (The times are from memory and may be off somewhat).
Your air compressor is cooling the chip. Try the same wattage heater with slow or fast air. You'll find that the final temperature of the slow moving air is much higher than the fast air.
It doesn't matter. Monitor the BGA case temperature and do whatever it takes with the hot air gun to make it happen.
Man did not discover how to use fire without someone getting their fingers burned. I could work out the thermodynamic concepts, but with your uncontrolled setup, it's unlikely to be achievable or reproducible. It think you could forget about putting your finger in the fire, but a thermometer or thermocouple would work nicely. One measurement is worth a whole bunch of guessing.
I gave you that in my previous rant. Somewhere between where the BGA solder reflows, and where the solder paste starts to soften. My guess(tm) is between 200C and 225C.
Stencils and solder paste work very will with BGA chips. Good video, but I wouldn't try that with a 400+ pad BGA chip.
I have exactly the same soldering station, except it says SAIKE instead of Maxtra. Not great, but good enough and the price was right.
Yep. The thermocouple running the display is near the nozzle end in the hand piece. By the time the air has gone to the nozzle, it has cooled down quite a bit.
There is no one answer because everyone's setup is radically different. I just want to fix my backlog of dead boards and move on to something more interesting. Good luck.
--
Jeff Liebermann jeffl@cruzio.com
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.