291 points by woolie 2 months ago
I am a friend of a famous Apple products repair person and one of the things I've learned through our friendship is just how fragile repairs can be. As this individual quickly learned, putting a MacBook Air logic board into an oven is a terrible idea. Do not under any circumstances do it with your computers. It is one of the worst fallacies of DIY repair out there.
What you're trying to do (as he stated in the article) is that you're trying to reflow the poorly adherent solder beads in the hopes that you'll fix the faulty connection. But the way you're doing it introduces the risk of damaging the PCB and the ICs themselves through thermal stresses and gaseous expansion by any trapped ambient moisture. This will irreparably damage your board and it will lead to the phenomena he experienced here. ("popcorning")
The right way to do it involves a directed hot air gun and proper equipment so that you aren't imposing excessive thermal stresses onto the board as a whole. If you don't know how to do that, then please send it to someone who does. There are several good repair outfits out there right now who specialize in this and can help you out.
Do not repeat this person's mistake. This DIY method is like doing cardiac surgery with a chainsaw. It's a terrible idea for your data and wallet.
Edit: Here's a guide that outlines these issues and offers ways to do it properly with the right equipment. The "popcorning" is called board delamination. https://www.latticesemi.com/-/media/LatticeSemi/Documents/Ap...
Tell Louis that a lot of folks appreciate his content and it would be nice if he would stop getting hit while on his bicycle. Brains are fragile.
eBikes can be quite fun compared to other modes of transit, its Louis's choice whether he wants to risk biking (or doing anything for that matter).
Our streets should be safer for pedestrians and bicyclists, cars get free reign and little in the way of fines or charges when a pedestrian or cyclist are hit and injured.
I'm a cyclist.
I live in New York City.
I gave it up because after getting hit 3 times and the driver drove off, including an MTA Bus that ran me over from behind and an NYPD officer that refused to take a police report for it, I decided that it's not worth it.
Wow, that is awful! SPD is a pile, but they've never refused to take petty theft reports and have chased down hit and runs for me in the past.
Have the bike lanes gotten any better in NYC? I'm lucky enough to be near a separated two way bike lane that crosses a few other separated bike lanes here in Seattle, which has alleviated most of my fear of getting run over. Matted leaves covering the lane are a bigger threat than cars :P
They didn't take a police report from me because the hit and run was by a city employee.
No, they have not gotten better.
Ooof, city employees get special treatment here (eg: they are less likely to spend the night in jail or get ticketed), but SPD will still file a report and send it to our prosecutor.
I would recommend biking with a camera if I were stuck with what your dealing with :c
I appreciate the technical content but his personality often comes across as bitter and cynical which makes it difficult to watch more than few minutes of. He's absolutely correct that Apple has oppressive and user-hostile manufacturing and engineering, but if that's the source of his acerbic outlook he ought to find something else to do for a living.
When doing this I typically use a very small and controllable butane torch with tin foil as a heat shield over nearby components.
Even then I always communicate to the person that the repair is intended to facilitate RECOVERY ONLY and that all bets are off on the machine being long-term usable again. Usually this is only useful for large BGA chips. Unfortunately Apple laptops made the past 10 years use BGA chips almost exclusively through logic board design.
Save yourself some headaches and get a dedicated reflow station that you can control by temperature. Butane is way too hot for anything other than soldering in plumbing. Even a knock off 858D is going to perform leagues better than a torch.
FWIW, I did this with a 2011 MacBook pro 17" 7 times, and it worked every time. Each time it worked for about 3 months, and then the video card would glitch out again. Finally, apple did a warrantee extension on that logic board and replaced it.
Just my 2¢ to say that it does work sometimes.
I did this 3 times with a HP laptop from 2010-2011 and had the same experience, the graphics chip would last 3-4 more months.
My understanding was that the oven repair was/is mostly for when GPU’s had to move away from less based solder due to EU regulations but hadn’t gotten non-lead based solder right yet.
Also I wrapped most of the board in foil only exposing the GPU, and I put the board in hot and then killed heat after one or two minutes.
I thought it was great as a kid without any means to a new computer. Friends would give me their broken computers and I would fix them for myself.
The right repair I guess depends on what you can afford and what you’re willing to risk.
If you had to do it 7 times, that doesn’t really sound like it “worked”.
Well, it took a non-working laptop - video glitched and unusable, to one that was working for 3 months. Was it a permanent fix? No. But, it sure beat buying a new laptop on short notice.
Sounds like my home theater amplifier (Pioneer VSX-921). The audio DSP chip gets too hot during normal use and desolders itself ever so slightly, which gives the error UE22. Someone figured out that taking a hot air gun to the chip fixes it, but I had to re-apply the fix about every three months.
What angered me even more was that Pioneer USA acknowledged the problem and would either repair or give you a replacement model, but Pioneer EU pretends everything is fine and dandy, even when you point them to Pioneer USA its stance.
Usually it's not the chips actually desoldering so much as it is thermal cycling causing the solder joints to crack. Most ICs can't really survive being heated to solder melt temperature for an extended period.
Either way, it's pretty bad that they didn't do something about it. That's indicative of a pretty serious manufacturing or design flaw.
Would a fan help?
There's nothing that the folks who are willing to pay for home theater audio gear love more than adding an acoustical noise source.
No idea, I gave it away for free to a solder-handy guy who was unemployed and needed something to drive a 5.1 set he was gifted from someone else. I did try a heatsink, but that didn't help noticeably.
It sounds like it did work, the problem is faulty solder.
There's likely nothing wrong with the solder, but with the lack of flux.
Its Monday, I think there is a widespread lack of flux today.
this man speaks a good point
They replaced your logic board under warranty after you'd baked it 7 times?
There is a water damage sensor, but not a put in oven sensor ;)
If there was it likely wouldn't make it out of the factory haha
It was a special repair extension program that they only put in place 2 years after the original warrantee+Applecare timeframe ended.
This is not an issue that putting in the oven is going to resolve. It's an issue with the die of the GPU. By introducing heat stress to the die it may be "functional" for a short period, but it is not a fix. This fallacy has existed for decades and was most prevalent in the Xbox 360 days. This is why the "towel method" was said to work, but was never a permanent solution.
I have that same 2011, and had been using it headless as a third machine with VNC. A recent bad update screwed it up, and I was going to call it quits but then found two work arounds.
Temporary (actually fairly permanent, but if the laptop goes to sleep you run into the backlight problem where it won't turn on until you poweroff the laptop and then back on again)
Between 2011 and 2013 Apple had a repair extension program for video issues. This is how they fixed them: https://www.youtube.com/watch?v=XaGHcBZjmWA
I have a 2013 MacBook pro that just dies at random times. I could never get them to confirm that this specific year had issues.
But I have always suspected that it was something with the board and poor solder connections.
https://www.apple.com/support/exchange_repair/ - looks like a battery and SSD program currently on that laptop. Maybe check in?
Unfortunately those two related items are for 2016-2018 models
I did the same with my 2008 MBP 15" and it worked, for a while. I then lent it to a friend who had it when the fix stopped working - he figured out that another solution was to simply wrap i in towels and turn it on and leave it like that for a while - it got hot enough to reflow itself!
No it didnt, Solder melts at >200C. Its bad nvidia gpu die bonding, heating it up to ~100 softens the glue and relieves stresses temporarily.
Im amazed there are CS people on HN not knowing basic physics.
Not sure whether to upvote for useful information or downvote due to the unnecessary insult tacked on the end. :/
I believe this is the one that explains it. Or the one before it. The reflow over time goes back to how it was, if it works at all.
A toaster oven is a somewhat standard way for soldering SMT components in hobbyist circles.
I don't think I would like my first attempt to be on something expensive but its possible.
You mean a toaster oven with precise temperature control.
Appliances like this can swing +/- 20 degrees at minimum. Doing reflow in this is suicide without reworking the temp controls.
Exactly, you need a toaster oven (pref an IR one) with an accurate temp control that will run a temperature profile that matches the solder used on the board ... in particular it needs to bring the temp of the board to just below the solder melting point and soak the board so it's all at that temp, then quickly spike the temp over the melting point so it all melts at once, otherwise you put silly stress on everything as some stuff melts and some doesn't and you risk tombstoning and other evils
So don't just stick stuff in your oven
You probably should use a hot air gun or a reflow oven, but the toaster method should provide similar results if used properly. A non-nitrogen, non-convection, infrared radiant heating reflow oven is basically just a precise toaster oven.
You can achieve radiant heating of particular sections of a board by cutting aluminum foil to expose or cover particular board sections. You can insulate the board to prevent excess heat soak, you can take apart the oven to force it to ramp heat more slowly, and cooling is trivial. You can also remove moisture ahead of time by enclosing the board in a plastic bag with silica gel or calcium chloride, or a very long, slow heat ramp. (Or just leave the board in a sunny window in a dry environment)
The above is correct. Do not do it. In addition, even with hot air soldering equipment, "reflowing" the chips on the board is not a reliable method of fixing the problem. Even if you temporarily improve things, symptoms will likely re-appear.
If the problem is indeed in the connections below BGA chips, the proper solution is reballing, which involves desoldering and lifting the entire chip off the board, cleaning the entire area, then reballing, placing and soldering it in again. This is a non-trivial operation that requires specialized equipment.
> the proper solution is reballing, which involves [...] reballing
Sounds like that will take forever.
For an ordinary individual, perhaps, but not for a domain expert. There are people out there who do this for a living and are quite good at this. Just like we advise others to leverage our expertise in software. We should leverage theirs. Specialization, ftw!
edit: oops didn't see the recursion joke :(
(It's a recursion joke)
At the very least get some no clean flux under there.
But it worked for four people on the internet.
You can find more reports than that on HN alone (in the comments)
Worked just fine for me on an early 2007 Mac Book Pro that had a known Nvidia GPU issue. I did it in late 2012, when it was only worth about $200-250 on craigslist. After the bake, it worked fine for the next 2 years, but I was paranoid it would crap out again, stopped using it as my primary after a year.
I followed the instructions from : http://russell.heistuman.com/2010/04/27/cooking-the-books-or...
> The right way to do it involves a directed hot air gun and proper equipment so that you aren't imposing excessive thermal stresses onto the board as a whole. If you don't know how to do that, then please send it to someone who does. There are several good repair outfits out there right now who specialize in this and can help you out.
Even this method is fraught with danger, at least if one is doing it at home. I have decades of experience with soldering and repairing boards with SMT components and I am hesitant to use a heat gun on a board unless it's a last resort. A couple of years ago I had a Mac mini 2011 logic board with the infamous AMD Radeon BGA warping issue, and I attempted a heat gun repair following best practices. I got another six months out of the board before it started having video issues again, and another attempt to revive it failed. I didn't have anything but time invested in it (it was given to me as scrap when it failed the previous owner), but I'd never use a repaired board like that as a production machine because it will fail again and soon.
A hot air rework station, not a hot air gun. Hot air guns are too generalized an area and end up melting/moving the wrong components.
"This DIY method is like doing cardiac surgery with a chainsaw."
Wasn't there a funny game on Steam for that? Surgeon something.
Needed: PC surgeon simulator.
Reflowing ovens and home ovens are totally different animals. Reflow ovens use closed loop temperature controllers. A ramping profile gently heats the board to a preheat temp just below reflow (drives out moisture and other outgassing), holds it for a bit, then ramp to the final reflow temp and holds that for less than a minute (possibly seconds) before quickly backing down to preheat followed by a cooling cycle. Many manufacturers specify these times and temperatures in their datasheets for components as well.
You can turn an electric toaster oven into a reflow oven with a cheap programmable closed loop temperature controller. Though I don't know how even the heating would be or how effective depending on the model.
And this is why I spent $100 on a hot air gun for my video game console repairs & restorations instead of trying to do DIY reballing without proper equipment (which is $$$).
Consistent and controlled heat are absolutely vital when working with tiny SMD components.
In some cases there's nothing to lose. Got an old HP LaserJet free that had a similar issue and wouldn't boot. Used the oven to 'reflow', now have a working printer.
"cardiac surgery with a chainsaw" encapsulates this perfectly
The analogy breaks down completely once you introduce the concept of "not worth the cost of professional repair". We don't apply that to humans, but it's more often the norm than the exception with electronics.
That is exactly what a 'Lifetime Limit' is for health care benefits, or when patents are refused because the may become a 'bad outcome'.
What's with the apostrophe in that New York Times headline: "When Doctor's Slam The Door"?
That limit definitely exists in America.
Also, is your name a Doctor Who/Ice Warriors reference?
Hmm: So if I'm reading you right, you're saying: instead of baking, freezing?
off to put laptop into dry ice freezer
Not sure why you are so strongly saying people shouldn't do it. I've done it before and know people who have, and it's always been for a broken/out-of-warranty machine anyway. So its either trash or will be fixed so it's not really a risk.
A possible (probable?) issue is your oven. In my experience, they heat in cycles: if you set one to 180 it will heat to 240, then cool to 150, then .... Etc. Even a large electric range will cycle with a range of temperatures. In other words they are designed to maintain an average temperature rather than constant.
In addition there issue of calibration. Even expensive full size ovens can be significantly off.
How do I know this? I worked with Japanese laquer (Urushi) which was baked onto metal for bonding. But Urushi is the sap of a tree related to poison oak and over-heating leads to fumes that are not great. Experiments with small (toaster) ovens and large electric ranges produced bad results.
Most electric kilns for ceramics, especially those that are computer controlled, are highly recommended for sensitive work where temperature control is important.
I’m familiar with this from a couple of interests, too. One is trying to decarboxylate cannabis to prepare it to make edible goods. You want to hold it under 250° for a period of time, and the quality of the temperature controller and other factors about your oven or toaster oven do make a big difference. Overheating can ruin it fairly easily, and underheating fails to produce full potency.
Also we run into this in a glassblowing. One wants borosilicate items to rest in the kiln somewhere between 1040 and 1100 F (~600C). The simplest kilns have a infinite switch, same as an electric range. With attention and experience, it’s possible to make it steadily hold a temperature. More common these days is a digital controller. The simplest is something like a Fuji PXR3. However, just having a digital controller does not ensure that your kiln stays at the target temperature. The controller must be calibrated properly for the oven size, elements and insulation. In most kilns elements are either on all the way or off, so controllers use a relay to control power. The more sophisticated ones do this very quickly, like flickering. Typical controllers do cycles of a few seconds. If the controller is not calibrated properly, which is common in glass studios, the kiln will overshoot or undershoot the target temperature repeatedly by up to 50°. It’s complicated by the fact that these kilns sometimes have doors that remain partially open, and the amount of mass of glass inside affects the thermal characteristics, too.
For a very precise small oven, the ones that they make for enamel or firing dental porcelain are pretty good.
^RIP my intended formatting, sorry
As anyone who has tried to bake in an oven in a rental has found out...I finally bought an oven thermometer and my oven will easily go 100ºF over the indicated temperature. I literally either cut baking time in half or lower the oven dial by ~100º.
If you’re lowering the temperature by 100 degrees, your problem isn’t the temperature swing. It’s that your oven’s internal thermistor needs replacement.
Rental or not, the difference between almost any low end to high end cooking item is the temperature control.
Sous vide, big green egg grills, induction ranges, some of the air fryers, etc. It’s all about precision temperature.
>> Even expensive full size ovens can be
>> significantly off.
It's funny how often the cooks complain about the ovens not being accurate on the Bon Appetit YouTube channel. There are 3rd party thermometers installed in each one.
The ovens are Viking or Fisher and Pytel or something, high-cost ovens that the average home chef would dream of. And they stink!
> It's funny how often the cooks complain about the ovens not being accurate on the Bon Appetit YouTube channel.
I wonder if it's more the cooks than it is the oven.
My wife once worked at a shop where she had to prepare fresh-baked cookies from scratch, and their oven was often "on the glitch". The owner of the shop also owned a pizza place next door.
So when the oven went on the fritz, she'd do what any self-sufficient cook would do, and borrow the commercial gas pizza oven next door.
Now mind you - this was a commercial full-sized pizza oven, running at an insane temperature - well over 700 degrees. But it was a consistent temperature...
...so my wife would pop 'em in, wait an appropriate amount of time while also watching the cookies closely, then pull them out when they were properly cooked.
They'd turn out perfectly every time. That isn't to say you could cook anything at that increased temperature - you can't. But there are many things you can do, if you know what the heck you're doing.
Sometimes you gotta make due with the tools at your disposal, because you may not have a choice otherwise. Or at least try to do it. If things hadn't gone right with the cookies, in the trash they would've went - but it went so well, my wife started to learn how to use that pizza oven for all it was worth, because it was usually always at temperature and consistent.
So let me get this straight:
- The chefs on said youtube channel complain about the temperature being non-consistent on their ovens.
- Your wife also ran into this problem with her oven.
- Your wife was able to cook using a pizza oven, which did NOT display this problem.
- Therefore, the chefs from the channel are bad chefs, worse than your wife.
I have not watched this youtube channel and cannot comment on the skill of the chefs, but your comment's conclusion does not logically follow.
I perhaps wasn't as clear as I wanted to be.
The point I was trying to make was to use the tools you have at your disposal. My wife had access to a "flaky oven" (half the time it wouldn't turn on) - or she could use an oven that worked properly, albeit at a very high temperature that for many people, would burn the heck out of anything you put in it.
In the case of the youtube channel - which I don't know about anything either - maybe, if the ovens didn't hold a proper temperature, they should have just cranked it to max temperature, not worried about what the temperature actually was - and just cooked with that?
It would of course depend on what was being cooked; there's a world of difference between say, roasting some potatoes at high temperature, and baking a souffle or something delicate like that. You probably couldn't do the latter in a high temperature oven properly.
But ultimately, most cooking is essentially controlled burning. If you can do that, you can cook many things. Not everything, but enough.
Actual baking on the other hand - a lot of it involves accuracy and chemistry, and if you don't have control over the whole process, then the recipe will fail every time. If that were the case with that youtube channel (quite likely), then doing any kind of a "hack" with an oven probably won't work out well.
In praise of American build quality (at least back in the 1960s): My GE electric oven, installed when my house was built in 1967 — 52 years ago — is dead-on accurate when tested with a third-party Taylor oven thermometer. Full disclosure: one of the four burners on top is non-functional.
Likely. Kitchen ovens and small ovens in general do not have the accuracy required to reflow. As you say, they will cycle between full power and no power and rely on the large oven hysteresis to get an average temperature which is OK for cooking, but can be easily off by 10+C at times, even if ventilated.
One solution is to wrap the item in something which acts as a dampener. A couple of layers of crumpled tinfoil will help.
Note that you definitely can reflow in a low-end kitchen oven with the appropriate changes ;). I built two small ovens for reflowing. Search for "arduino reflow" for some pointers.
I found this to be the case for mid-level espresso machines as well. Luckily it is fairly simple to add a PID controller and a more accurate platinum temperature sensor to hold the brew temp to within a few degrees Celsius for a much more consistent extraction.
> In my experience, they heat in cycles: if you set one to 180 it will heat to 240, then cool to 150
Get a better oven. A 90 degree Celsius swing is insane. Ovens do generally have some amount of swing but it should’t be anything close to that.
Not to mention the picture appears to be a toaster oven; not paragons of heat stability.
Would a heat gun and an infrared thermometer perhaps be better - don't know if you risk blowing the components off the board though. Maybe on a lower setting.
I think a heat gun would + tweezers would probably be sufficient enough, however, he'll probably need to reference ifixit images to figure out where things need to go.
I think it's a bit too difficult to hold a heat gun, an infrared thermometer, a set of tweezers and flux, so I would just get a feel by testing it out with some other busted PCB.
Has anyone watched Louis Rossman's video series on this topic? It's pretty funny. He thinks oven reflows are a bad idea.
Reballing flip chip GPUs is BULLSHIT - the truth about dead laptop GPUs & repairing them. - https://www.youtube.com/watch?v=1AcEt073Uds
Luckily, he also shows the proper way to do it:
Linus Attempts BGA Graphics Chip Repair! - https://www.youtube.com/watch?v=Shn7LdIrViQ
I got an X1 Carbon rather than a Macbook Pro in part because if his Mac repair videos.
Bad soldering is bad soldering. Microsoft's XBox 360 had huge issues with it.
The problem is that it works from a QA perspective: They can test the thing for hours and it'll run flawlessly. It's only when it goes through hundreds of heating/cooling cycles that it starts to break down and cause problems.
You really never know which product or vendor is going to suffer from it next. Every GPU vendor has had some issues.
I got red-ringed and I didn't swear off of buying Microsoft products forever.
I was gonna chime in on 360's and saw your comment. The best solution that didn't involve replacing solder was a reflow combined with case mods to add screws that better stabilized the board directly around the problem chips which limited board warping so the inevitable solder melting heat cycles did not result in connection failures upon cooling, IME. There was rumblings that lead-free solder was a requirement for 'toy' classification but I can't confirm.
The whole industry was moving lead-free but the first companies to do it were the ones that had the least experience with how to manage the reflow temperature profiles.
It's not just toys. Lead solder is being phased out for many applications.
I mean even if the oven were perfectly calibrated for air temperature the radiation from the exposed elements would almost always cause this to happen. He’s basically grilled his motherboard not baked it, try baking a cake in that oven and I bet it’s also burned on the top and soggy in the middle.
The obvious solution if you must use that oven is to wrap the board in silver foil first and it might have had a chance.
That might also keep the vapor from the solder condensing on the oven. I would be worried about using that oven for food in the future. Is apple lead free for it's solder?
Nobody really uses lead in electronics anymore unless it's absolutely necessary. RoHS (regulation concerning lead use in electronics in the EU) really worked.
Lead-free solder has it's peculiarities but the industry long ago got used to it and leaded solder is now mostly seen by hobbyists. I've made the move to lead-free and while it's not quite as easy to work with, the peace of mind knowing I've eliminated at least one nasty heavy metal from my hobby is worth it.
The boiling point of lead is 1749°C. You're not going to get lead vapor from the solder at the temperatures of a toaster oven. The smoke is probably from some plastic component or some remaining flux on the board. But yeah, that oven shouldn't be used for food anymore.
Liquid lead has a certain amount of vapor pressure even before it boils. Just like you can have water evaporate even when you're at temperatures below 100 C you can have lead enter the air even when you're below 1749 C.
There is no leaded solder any more in consumer electronics. Even for leaded solder, the smoke comes from the flux core. You'll notice melting unfluxed leaded solder produces essentially no smoke.
Solids also have a vapor pressure. Yes you've increased the vapor pressure, but you're way below the temperature where significant vapors are released.
They might have antimony, bismuth, or cadmium. All are toxic. Also, many other components on the board could include God-knows-what. I'm not sure what Apple has or hasn't cut out of their stuff, but there's bound to be plenty of bad stuff left in there. I certainly wouldn't want to use that oven again.
Most likely. Everyone uses lead free for consumer goods that are also sold in Europe. It’s not worth the supply chain woes to be different.
Despite what everyone says lead free isn’t all that bad.
I learned on lead solder but quickly moved to lead-free and while there was a learning curve, I am confident that anyone capable of soldering leaded can be just as skilled with lead-free.
Everything these days should be lead free, but unfortunately that's just one of many toxic ingredients. I'd throw that oven out if I were him.
There's no lead in the solder firstly, and secondly even if it was all leaded solder, it's not going to vaporize and deposit on the oven.
That's a myth. Leaded solder is completely safe for those soldering with it. You're never exposed to any lead fumes. Lead in solder is only a concern for the environment when the product is thrown away and that lead gets exposed to animal life.
Apple products have been lead-free for more than a decade.
> Leaded solder is completely safe for those soldering with it.
True, As long as you wash your hands after handling it.
The lack of lead in modern solder is what causes these issues in electronics in the first place.
I remember it was a common discussion topic for the Nvidia 8800GT.
Is there significant lead vapour pressure at a mere 200ish C?
The vapour pressure of an absent element is approximately zero pascals.
Yeah, he definitely should use a sous vide circulator the next time.
Too hot. Water would be violently boiling and fat too much contact on the parts in a bag under vacuum.
Reflows ovens work because they allow nothing touching the parts on a level board, the solder paste melts into solder pools and the parts center themselves by surface tension over each pad.
Most of the time if you are hot air soldering and the solder, flux, pad shape, and part are good, you need no positioning tool at all. The solder melts and the part just wiggles itself into place.
I'm not sure why you'd say "too hot". Most sous vide circulators cannot even boil water, so it definitely wouldn't be "violently boiling". The general use is to hold a low temp (eg, 130F) for an extended period of time with little variation (ideally <1 degree F of variation).
Getting the temperature stable without putting pressure on the parts, and also not getting them wet would be challenging, but probably not that much more challenging than doing the same thing in a toaster oven or other regular oven.
As someone that has done a LOT of reflow work, I'm a little confused why you are arguing such a ridiculous notion.
First off, I think you confused C and F. The author wanted 170ºC in his oven, would would be outside the range of any existing sous vide and would be violent boil.
Second, reflow only works because of the perfectly level and no contact of parts, they need to be free to FLOAT to their destination guided by surface tension.
Oh, sorry, I was joking initially and misinterpreted your remark about it being too hot. You meant that the required temps are too hot for the circulator, which is obvious in hindsight. Thanks for setting me straight. :)
I'd still argue that a sous vide type circulator with a slight modification, a liquid with a higher boiling point, and a container for the board that doesn't involve vacuum sealing, would be better than the raw, unreliable oven in the OP. I bet it would end up costing about as much as a better solution, though.
130 degrees freedom is about 55 °C which is way too cold for any reflowing to occur. Consider that the failed joints in question will routinely see 70-90 °C in normal operation.
Doh, of course. He actually needs >100 °C which would be impossible with the average sous vide circulator. Oops.
The oven has to be pre-heated before putting the board in too. And silver foil balls on the corners to keep it elevated and heat evenly.
So you mean baked potato style? This whole thing is tickling me just the right way.
I've done stupider things during last ditch repairs. Once time I replaced the tantalum caps on a board blindly hoping that it would cure a power rail short. I'd never used SMD tants before and didn't know that the band was a + band and not a - band.
Within 3 seconds of applying power, there was a fireworks display which destroyed the board entirely :)
Oh yeah, tantalum caps... I stuck one into a prototyping breadboard but messed up polarity.
It's surprisingly hurtful if one of those little fuckers explodes between your thumb and index finger.
Ah, the good ol' tantalum caps. Violent ones for sure :)
Tantalum caps are great fun. I remember we would booby-trap each other's breadboards with them in EE lab, to this day I'm still extra vigilant about polarity of all components. Measure twice, apply power once.
Regular electrolytic caps also make a nice pop and ball of fuzz, often the shell will fly across the room at good speed. Also they have that unforgettable odor of shame that will follow whatever you were working on for a few hours.
I love blowing electrolytic capacitors up. I still do that today.
I don't know if this has been said or not already - but I wouldn't use that oven again for cooking food, especially since something somewhere off-gassed.
I don't know how dangerous it would be to eat food from an oven that was previously used for reflow soldering; if I had to guess, the danger would be minimal - maybe an increased risk of cancer or something like that. Likely, the fumes or whatnot was simply rosin flux vaporizing.
Rosin flux is usually something like refined pine resin with a solvent; it won't kill you to eat it (or breath in the smoke while soldering - though that can be a major irritant), but I am almost certain it might cause a flavoring issue for future baked foods.
Then again - maybe it would lend those holiday dinners a festive flavor?
You're over-estimating the amount of heating/cooking involved. Ignore the article where he's doing everything all wrong, the plastics etc shouldn't be burning, only the solder is warming up and reflowing when done right. Fumes, if any would clear out anyways.
I agree that when done right (ie, following a proper SMD reflow temperature "curve"), all that should happen is the solder should flow and not much else.
Even so, just as I wouldn't reuse a pan or pot I had created sugar candy rocket fuel in to cook dinner (no matter how clean I tried to make it afterward), I also wouldn't reuse an oven for cooking purposes, that had been used for reflowing a PCB previously.
It just seems like a "best practice" to me to not reuse equipment in that manner for food cooking and consumption.
That's just me; I don't find the potential harm, no matter how objectively slight it might be, to be worth it in the long run.
I heated a Nvidia 780 graphics card a few times using a heat gun and a laser thermometer (max 200 degree c). I did not want to use my oven as I don't want to risk it. That revived the gpu 2 times, going from vga quality and weird screen artifacts, to no issues at all, 2-3 months with each bake. In the end i bought a new one as it really is temporary.
I repaired a 2009 iMac gpu the same way with a heat gun but that was good for two years each time the first two times and didn’t work the third time. It’s worth a shot if you strongly suspect broken solder ball.
Same here, for the same equipement. Did not use heatgun but oven, wrapped in tinfoil. Same results as you reported. Finally sourced a compatible GPU from eBay, an ATI 5xxx from the next model year iMac. It was much much cheaper than sourcing an ATI 4xxx.
Opening the iMac the first time was very stressful.
We used to "fix" our old XBox consoles by wrapping them in towels and purposely overheating them to get the ?GPU? solder to reflow correctly. With some extra cooling methods added, still works to this day even though it got a red ring of death.
I also had the overheating RROD. The problem with the 360 was the CPU heatsink mounting hardware was insufficient to maintain contact and the board would warp. It was easy enough to fix with a set of screws from the hardware store to replace the crappy 'X' clip, plus some PC thermal grease like Arctic Silver.
Relatively speaking, the CPU heatsink was massive compared to the GPU. I would assume the engineers designing it knew that the CPU was going to run very hot, especially when it launched on a 90nm architecture.
I fixed the first RROD on an Elite (Falcon) by re-balling the solder joints using a hair dryer. The next fix was an identical process to yours and it still works.
I successfully fixed an Xbox 360 with the dreaded RROD with Percussive Maintenance: https://tvtropes.org/pmwiki/pmwiki.php/Main/PercussiveMainte...
I remember doing this in a bar at 2am to one of those electronic paper towel dispensers. They work great, except when they don’t (which is most of the time).
A swift elbow to the side cracks the plastic locking mechanism, dropping the cover and freeing the roll.
I think that one’s called vandalism.
I call it bringing them back into compliance with local bylaws on bathrooms.
I don't understand America's fascination with these. I first visited the US over 15 years ago and they were everywhere, but since then I don't think I've even seen a single one in the UK.
It’s often in the building code or local bylaws. They don’t have to be automatic, but there’s probably evidence showing that the automatic ones reduce overutilization.
Or do you mean the UK just uses air dryers? The newer vortex ones work ok, but the previous ones were useless, which encourages people to not wash their hands at all.
Only having 1.5kw to work with probably didn’t help.
> Only having 1.5kw to work with probably didn’t help.
Hand dryers get dedicated circuits. They aren't limited to residential outlet specs.
Sure, but copper is expensive and you can pull double the current for the same wire at 240 vs 120v.
Difference here is that overheating the GPU will always apply heat directly to the specific solder area in question. It’s actually a pretty good trick.
After doing the towel trick I laid my 360 on top of a big fan so it always had cool air being forced into it. It was RROD-free until eventually a thunderstorm took it out for good along with my router.
Used to do the 'blanket fix' with the first few Intel MacBooks that had failing NVIDIA graphics chips.
Wrap them up in blankets, powered on, and wait a few hours.
Yea, the original run of the Xbox 360 had issues with the then-novel lead-free solder breaking down under high thermal loads.
If anything lead-free solder would hold up to higher temperatures. The problem was just that the entire thing got too hot in general, there was a whole spell with the PCBs warping because of the heat.
Yeah, both the PS3 and Xbox 360 had terrible thermals when they first released.
Normal operating loads. My 360 was never in an enclosed space, and it still fell prey to this. I was do the temp fix with a heat gun, but disassembling and reassembling was a pain. I eventually left the board sitting in the open without the case.
They didn't fix RROD until the slim in 2010. Every other 360 is a time bomb.
After flickering unhappily for a while, part of the screen backlight on my trusty Lenovo laptop recently died, rendering screen usable but hightly annoying. Short of actually buying a spare part and swapping out, I did absolutely everything imaginable, including liberal amounts of percussive maintenance: Opened the thing up, had the whole screen out, checked connections as best I could, squeezed and flexed, had the screen out again, squeezed some more - you know how these things go.
So, the machine is stranded on my desk, hooked up to a stationary screen, lid kept slightly ajar. The most laidback member of the household comes by, a fully grown male of species felis catus. Apparently slams the lid shut and goes to sleep on laptop for the duration of the night.
Machine has been fine ever since.
To add to our set of anecdotes:
I baked a dead graphics card in the oven and it hasn't failed for half a year or so until I replaced it with something better. It probably wasn't the smartest thing I ever did and I wouldn't recommend it unless to anyone unless they have an oven they aren't going to use for food anymore in a well ventilated are (which was not the case for me).
The card had bad graphics glitches in whatever mode the BIOS puts it in and would cause boot to fail at an early stage.
First I disassembled the card and ensures that all remaining components were heat resistant and that there was nothing that seemed likely to fall off on the bottom side. I carefully mounted it in the oven on some balls of tinfoil. I also added a thermometer that came with a cheap multimeter because I didn't feel inclined to trust the oven not to overshoot.
I ramped the temperature up slowly, using hot air and keeping it under 100°C for a while in hopes of getting out any water and keeping it from going popcorn. When raising the temperature to the point where I expected the solder to go soft (can't remember the exact temperature) the thing suddenly started to smell rather badly (but not burnt), forcing me to open the windows and leaving the room most of the time. At that point I got worried about the fumes and where they might condensate but decided to go through with it, heating it up a bit further and then cooling it down slowly.
In the aftermath I got a working graphics card and a smelly oven. The smell went away after a few hours at max temperature, a lot of ventilation and some cleaning. I hope there wasn't a health issue with any substaces remaining in the oven as the smell was gone and food doesn't really get in direct contact with the inside of the oven. Anyway, it's the reason why I wouldn't recommend it or do it again.
Could have just mailed it to Louis Rossmann. 
As strange as this may sound, this technique isn‘t uncommon at all. Many XBox owners used this as a last resort to fix the infamous Red Ring of Death.
I actually once (kind of) fixed a permanent startup blue screen on my years-old LG G3 by putting the motherboard in the oven.
It worked again for ~2 months, which gave me plenty of time to backup all local data. Then the blue screen returned and after trying to bake it again, it would not power on at all anymore.
The above technique is a myth . See https://youtu.be/1AcEt073Uds .
Trying to randomly unsolder and put back resistors or other components when you don’t know what the error is is another bad idea.
The author should have tried to actually attempt to fix the logic board first. But before all that he should have inspected the board and use a voltmeter to see what the issue is. But even doing that he might need a replacement part.
> With confidence high and the bake nearly finished, for the final 60 seconds I thought I’d go off-piste and crank up the temperature to 180 °C - I wanted to make sure things were cooked through. Curiously peering through the oven window, all hell broke loose within 30 seconds: The room filled with sounds of popcorn being made as resistors and components desoldered themselves from the logic board and dropped onto the oven floor. The previously clear air was replaced with an acrid haze. Then the bake reached it’s finale as the logic board bowed up in the middle, accompanied by the screeching sound of the CPU being wrenched off its socket. I lunged for the power switch and yanked open the oven door, hoping to limit damage. Then, as quickly as the bowing started, everything calmed down and the board returned to its original shape. With the board still hot, a wooden spoon was employed to desperately poke the CPU back onto its mount - with little success.
I'm sorry about your MacBook, and I just wanted to thank you for this wonderfully evocative bit of writing which made my day.
For those that didn't read the article the error he made was using a small toaster oven. And then cranking up the heat at the last moment. The oven's thermostat is reading the current air temperature. But the heating coils are radiative. A small oven has wild temperature fluctuations. And when the coils are active they put out a lot more heat than whatever is dailed into the thermostat.
I recently repaired our iMac GPU after it started acting strange. Baked it in a convection oven for 6 minutes at 160° and not a second more. I also used a heavy baking sheet as a heat sink. I did not need a reflow station and if it failed I would have simply ordered a replacement GPU.
I think the error he made was baking the motherboard, period. As multiple people have pointed out this is a bad idea.
I actually managed to temporarily fix a laptop using a torch-like lighter directed at the GPU. This particular batch of laptops had faulty solder connecting the GPU to the board. The heat and cooling would eventually crack the solder and disconnect the GPU.
I had nothing to loose and saw a video about it on YouTube. So I cracked it open and ran the flame around the GPU for several minutes. I eventually got it boot up and it lasted for another 9 months and then I needed to do the process again. I did it a few more times and probably got another year and half out of the laptop.
> I figured, if it’s good enough for those four people on the internet - it’s good enough for me.
Words to live by. :)
Great writing, I'll make sure to suggest "percussive maintenance" from now on. Still leaves us wondering what would've happened with a less confident chef !
You may find that Debian with LXDE+Xfce installed (and the non-free repository enabled, for firmware components) will work quite a bit better than Lubuntu for the laptop you picked as your MacBook Air replacement. Recent versions of *buntu seem to be a bit bloated compared to available alternatives, and this especially matters on older hardware.
This article he's just being funny and deliberately doing everything wrong and ignoring instructions.
In 2012, I successfully baked my 2007 MBP logic board and it worked for 2 years after that. It was a known issue w/ the Nvidia GPU connection.
Obviously you have to pre-heat the oven, you don't want the heating elements to be on continuously as that would be too hot. A trick to reduce variance is to pre-heat to say 200c, then put the logic board in. The heat that escapes during that time would bring it close to your target 170c.
There should barely be any smell. In my case it was just the smell of burnt dust, slight metallic smell, probably the solder.
Followed the instructions from: http://russell.heistuman.com/2010/04/27/cooking-the-books-or...
If I recall correctly, Louis Rossmann claimed official Apple repair center baked some motherboards as well. I don't have a link to exact video, but it is probably covered here: https://m.youtube.com/watch?v=AUaJ8pDlxi8
And yes, I use Dell :)
> Weighing up the consequences of covering our main oven in molten MacBook, I also chose to use our standalone oven
Anyone who ever tried to bake a frozen pizza in one of these ovens following the instructions on the box would've immediately recognized that this is a very bad idea.
Guys, you will never ever be able to do work of a $1000 programmable reflow oven on a kitchen appliance
Depends on what you're doing and how patient you are- I know several people that have done similarly with kitchen-grade toaster ovens, thermocouples, and relays / SSR's to manage temperature and get professional-quality results. It's within reason that someone with a laser thermometer or such to be able to manually do the same thing.
Can a toaster oven vent heat away on demand? No
Can I wire a servo to open the front of the toaster? You bet. That'll be another $15 for a servo, some wire and some code.
I had the same issue with my MacBook Air with the exact specs, though I never dropped mine. Just shut it off one night after watching Thrones for a few hours and it never turned back. Pleaded with Apple for quite a while to see if they would replace the logic board due to defect, but they did not. Mine was purchased in May 2013 and dead Nov 2014. Finally broke down a couple of years ago and bought a low end I3, with 4gb of ram logic board and it's working again. Guess I would have tried this as well had i known it was an option at the time :)
If you still have that logic board, sell it. Someone will buy it for repair or donor parts.
Unfortunately I sent the whole thing back to the repair shop and I'm guessing they probably sold it. In hindsight I should have taken it out prior to sending it to them...
Another thing that is commonly baked, by professionals: old tapes that have trapped too much moisture and start to deteriorate. They're placed in ovens for much longer, at lower temperatures.
dry boxing and pre baking older parts are standard practice in electronics manufacturing. You put your parts in https://www.manncorp.com/dry-cabinets between the runs.
I've baked components before, only once with limited success.
Several years ago I had a 9800GTX graphics card fan go out, and its core temp rose to 117 Celsius. Eventually it died, and I attempted to bake the card to re-flow broken solder. I'm not sure if that actually fix anything, but the card worked for a bit longer.
Eventually I had to replace it.
I sort of wonder if using a heat gun in the area they thought was broken would have worked better. Then it would have been more localized and easier to control temperature-wise.
A MacBook Air can't survive a 30 cm drop (less than 1 foot)? Most phones can't survive a 10 second dip in a toilet bowl? Is it just me or is technology today weak?
I mean seriously, anyone could build these things better than the big companies are. Yes, the technology in the chips and logic boards is astounding. But the engineering of the cases is.. either amateurish bordering on the inept, or companies design in such brittleness to sell more devices.
> A MacBook Air can't survive a 30 cm drop (less than 1 foot)? Most phones can't survive a 10 second dip in a toilet bowl? Is it just me or is technology today weak?
Was there ever a time where you could do this things to your electronics and guarantee that they would work afterwards?
All it takes is for it to land the exact wrong way one time.
Back in the early 2000s my iMac (Lime G3/400), which ran constantly, wouldn’t turn on after being shut off and cooling down. It definitely didn’t fit in the oven, but it’s ass went in anyhow, with the monitor hanging out the front of the oven.
It came back on and worked for a couple more years... the next cool down killed it for good though. No amount of baking would bring it back. Granted, I didn’t pull the logic board (I’d upgraded to a G4 by then)...
Oh, the G4. My favourite Mac ever. It's aesthetics are unprecedented. I still run one in my studio - with dozens of GB of 360p formatted old shows like classic Simpsons and the 1985 Twilight Zone, and tossed a SEGA, GBA and NES emulator on it with a little USB game controller for when I need to unwind.
I think I'm most disturbed by the author's comment that he has dropped it from a height of 30cm multiple times.
It's a 11' thin and light laptop, it is easy to abuse of the portability and ending up using it in places such as on bed, sofa, moving it around the house, etc...
Not saying that OP shouldn't be more careful with such an expensive piece of hardware, but accidents happen, and some people are more careless than others (I'm one of them and a smartphone or tablet of mine doesn't even get out of the box withouth a matching protective cover).
I can't remember ever dropping a laptop. I'm sure I have, but not so many times that I'd remember... multiple of them. And I'm not terrible agile to begin with. Just sounds like carelessness.
More than a decade ago I "successfully" fixed a thinkpad with a loose GPU using a heat gun. Prior to the fix, the screen would have artifacts unless I held the GPU in place with my hand. The repair only lasted a few months though.
These days I would imagine the parts are smaller and more delicate.
Does anyone ever try to address these issues with a soldering iron? I assume these machine soldered boards have connections too fine or too inaccessible to solder individual connections. But if you are trying something as crude as baking the board.. maybe there is some way?
Most likely, one of the bigger chips would have developed a cracked contact, simply because of the bigger surface would introduce more stress if the whole board was deformed/shocked. Now, those BGA's, you cannot solder them with a soldering iron, reflowing is the only option.
> dropping it on the corner like that might have shaken a few bytes of SDRAM around
Is this a plausible diagnosis?
Short answer, no. He’s just being cute.
Longer answer... well... two components I can specifically think of can be effected by shock and vibration:
1. Spinning disk hard drives are one, although you could probably debate this because there are protections in place. The idea being read, shock, corrupted read now in ram. Head of a hard drive should park itself under shock, and checksum bits should protect against bad reads.
2. The other is crystal isolators. XTALs are tuning forks. And if you physically hit them the right way, they vibrate. It’s absolutely possible to induce a pulse per minute variance with a crystal oscillator under vibration. A lot of Apple products are moving to MEMS clocks, which are silicon “solid-state”, do not suffer the same problem, smaller, cheaper, but have other issues (remember that story about iPhones dying in the hospital because someone vented the helium? MEMS can die in helium, but the story was nonsense, not even slightly enough ratios to actually affect phones). Don’t know if you could hit a crystal oscillator so hard and so sharp that you actually injected a full clock cycle into a running device, but, answer the question yeah it kind of sort of could be possible to corrupt ram with a shock to a crystal and probably ceramic clock.
I managed to revive a GeForce 8600M GT board using the oven method a few years ago. A much smaller board with fewer components. The bake time was only about 8 minutes though, and I pre-warmed it thoroughly to ~105C in a separate oven first.
I did something similar to this on an iBook G4, I believe?
It's still working to this day. I feel like I was a lot more careful. The author seems to describe their behaviour as rather reckless and haphazard.
Certainly putting it up to 180 wasn't a great idea. :3
I did this around 10 times with my MBA Mid 2011 in the last years now, it worked every single time.
Why did you put it upside down with the heavy components underneath? You should have read the guides.
When reflowing components, one definitely does not want to do so upside-down, no matter the heat source, excepting in very special circumstances, perhaps.
Why don't people simply use their consumer rights where they apply? In UK there's 6 years period for electronics good where the seller is responsible for things working.
https://news.ycombinator.com/item?id=7371908 The earlier referenced article.
You are an excellent writer, so funny and gripping. I suspect writing is your true calling in life. Lol @ what on Earth is that oven? :)
>> Although I am tempted to take it to the genius bar now, just for entertainment.
Do they serve baked MacBooks at the genius bar?
Out of all dumb ideas sticking a logic board into a non-calibrated oven to "bake" takes a cake.
I and many others had great success in "baking" our nvidia graphics cards a couple of years back to resurrect them.
In a calibrated oven.
As a teenager I used to wrap my bricked (RRoD) Xbox in a towel to intentionally overheat it. Fixed the issue and kept it going for a while.
It seems like it was working fine until he cranked up the temperature...
I don't think that there can be a very wide temperature band between "too cold to fix" and "components starting to fall off" for a two-sided board, the temperature band might even be non-existent. (Or is surface tension a function of temperature, enough to hold components at low temps, dropping when hotter?)
But professional repair shops are using a more sophisticated form of the same method. I wonder if the have some kind of adaptive support layer instead of a simple grille or flat surface?
"if it’s good enough for those four people" made me lol
But at least now it is 4 to, well at least, 1 no success
I have a friend who did this with a PC motherboard, and it worked. Personally I would never try it.
> However, since 4.18.0 everything seem to work well!
This weird mixmastering of the version (18.04)... Yes, I know it's based on a date, but it's a version number. Probably best to just leave it.
He should definitely mailed it to Louis Rossmann. Sendyourmacbook.com
Ok folks, i think we found the guy that started the fire in paris tonight.
Better Call Louis (Rossman)
The saddest part is that it's impossible to replace that Macbook Air now. The new ones have the garbage keyboards and even worse they discontinued the 11" model.
Best you could do is pick up another used one. Can't even buy refurbs from Apple.
> Can't even buy refurbs from Apple.
No 11 inch available
TLDR: "or the final 60 seconds I thought I’d go off-piste and crank up the temperature to 180 °C"
You have to wait until it gets crispy.
HP stream makes a nice linux laptop. I'm running opensuse Leap on one with no problems. It handles KDE just fine.