Fighting the myth of baking as a repair

[Ranayna]

Preface: I am not talking about baking components to dry them before soldering. I know that that can be required and is useful, and why.

I am active in various computer communities, that are talking about various aspects of PC hard- and software.
Of course, troubleshooting discussions are common in these communities.
And in many discussions, when simple re-seating of components does not help and there are no obvious incompatibilities, almost inevitably the topic of baking comes up.
The last time I stumbled across baking advice, was to bake a PC mainboard for 15 minutes at 130 decrees C, followed by 5 minutes at 150.

If i recall correctly, this baking myth comes from the early days of lead free BGA soldering. What spread this into the computer communities was a widespread issue with the first lead free nVidia GPUs, especially the mobile chips, that had cracking solderballs and/or interconnects. Baking apparently got the affected cards working again, and the cards apparently worked for several months afterwards. Also i think a number of X-Box consoles were affected by that as well.
What I do not know if there really *ever* was a successful repair with a home oven, or if people misunderstood proper reflow ovens as a normal baking oven.

My guts tell me that this cannot work and only has downsides, potentially ruining your board and more.
But I am not an expert, and there is so much anecdotal "evidence" flying around, that I would need well founded and researched arguments refuting this.
I want to be able to refute the myth without having to second guess myself all the time
And should it *not* be a myth, then I would like to know what an actually working baking entails.

So, my first Question: Is there any non anecdotal evidence that baking in a consumer oven can help?
If i am not totally missing something, even leaded solder does not melt at the temperatures typically recommended for baking. I have seen anything between 120 to sometimes even 180 degrees Celsius. So what can baking at these temperatures do, besides temporary effects due to thermal expansion?

Second question: I never really thought about it, but what are even the tolerances for a typical consumer oven? How accurate is the set temperature? Can there be hotspots reaching significantly higher temperatures?

Third Question: What are the arguments against trying this, even as a "last resort"? I can think of a few:
Stressing components, especially electrolytic capacitors, or maybe worse, soldered on batteries.
Professional repair afterwards becomes more expensive or even impossible.
Potentially melting plastics.
Potentially ruining your oven, or at least your next meal.

What other aspects am I missing here?
Or am I totally mistaken and baking really works?

[Psi]

A normal house oven will melt solder and reflow a PCB just fine. The temp control knob usually max out around 240-260C depending on make/model. Solder will melt around 180-215C depending on solder type. You want a little more temp than 180-215 to activate the flux but 240C is usually enough

There is the issue of temp accuracy and the thermostat tends to cycle over quite a wide range.
It's not ideal but not a deal breaker.
You're not getting anything like a proper thermal profile, but for home jobs it usually doesn't matter.

If the problem is a cracked BGA ball then you can often rejoin the ball by running lots of liquid flux under the IC and then putting it in the oven for 4min at 240C, but it's usually a temporary fix. The balls often don't reform correctly and are very prone to cracking again but sometimes you get a permanent fix from it.
You also have to be very careful with plastic parts on the PCB and heavy parts that might fall off under their own weight. It can be tricky.  You also have to rest the PCB on something without touching parts on the underside and without allowing the board to warp during heating.

The GPU+oven fix thing was not a reflow. The reason all those nVidia GPU chips died was due to cracks inside the chip substrate and not cracked balls. If you heated them, even to less than solder melting temp, the thermal expansion would move the atoms around enough for it to start working but it would always failed again soon after, sometimes minutes, sometimes hours, sometimes a week or 2. So it was not a fix, just a scam to charge money for doing a fix when it was not fixing anything.

[coromonadalix]

you would need all the tools to do reballing / stencils / heat soldder - desolder gun

correct paste ball size

etc ...  list goes on and on

[Ranayna]

A normal house oven will melt solder and reflow a PCB just fine.

I know it can. But anytime i see someone suggest baking a GPU, it is always at lower temperatures, as in my example 120 to 150 C
I suspect that this was enough for the nVidia issue, and it stuck.

[bdunham7]

What I do not know if there really *ever* was a successful repair with a home oven, or if people misunderstood proper reflow ovens as a normal baking oven.
So, my first Question: Is there any non anecdotal evidence that baking in a consumer oven can help?

Why would you expect or demand 'non-anecdotal' evidence in a case like this?  Can you think of any non-anecdotal evidence that jump starting cars can work successfully? 

Or am I totally mistaken and baking really works?

Baking can and does really work at least occasionally, but as for the subsequent life of the baked board I don't know.  Here's an anecdote: A local makerspace had a bunch of inop 3D printers, one of which was a fairly new CraftBot.  I determined that the MCU board was bad and the owner ordered one, but in the meantime I had a look and really wasn't able to nail down the issue or find any typical failures.  I lightly toasted it, it worked and many months later it was still working with the replacement kept as a spare.

You might say that its resurrection may have been random chance, due to handling, etc etc, but I think the most likely explanation is the obvious one--heating it up caused something to shift or change and it started working again. 

[DavidAlfa]

It's not a myth at all, at least in manufacturing, if the parts have been properly stored in a sealed bag with silica, probably will do ok, but if not, they might become popcorn.

Hand soldering is usually faster and less prone to this, still I've heard pops and cracks while soldering a part.
But the oven maintains high temperatures for longer time, if the vapour can't escape it will build pressure and you can figure out what happens next.
I'm not an expert, but you want to avoid reworking the boards even if the popping chances are only 1%.

10 years ago I fixed a friend's motherboard in a oven.
It had stopped booting, and it smelled like an old book stored in a humid place.
Literally cleaned it in the sink with some dishwashing detergent, rinsed and dried in the oven at 140°C for 30 minutes.
Worked a treat!

[Ranayna]

Can you think of any non-anecdotal evidence that jump starting cars can work successfully? 

I admit that it has been a very long time since i last checked the manual of my car. But i have at least once seen a car manual detailing the instructions how to jump start a car.
This was also taught and even demonstrated in driving school.

I very much doubt that any manufacturer ever will publish instructions on how to bake their product in a consumer oven.

Your printer board example is just that: anecdotal. Not that I do not trust you, but without even the details of temperature, time and equipment used, this is even more anecdotal than many.

[Ranayna]

It's notyth at all, if the parts have been properly stored in a sealed bag with silica, probably will do ok, but if not, they might become popcorn.

Hand soldering is usually faster and less prone to this, still I've heard pops and cracks while soldering a part.
But the oven maintains high temperatures for longer time, if the vapour can't escape it will build pressure and you can figure out what happens next.
I'm not an expert, but you want to avoid reworking the boards even if the popping chances are only 1%.

I am explicitly not talking about baking parts prior to soldering.
I clarified that in the very first line of my post, and explicitly mention repair in the title.

I know why components are baked and that this is important, but that has nothing to do with what i asked.

[DavidAlfa]

Yeah sorry, I noticed later.
For fixing GPUs, yeah it might help, but it won't last long if it's a bga issue (Cracked solder balls).
A reflow might help a bit more, I fixed several laptops with a heat gun, putting some flux and blowing compressed air so it came all under the bga chip, then preheating evenly for 3-4minutes, with a final blast over the bga until it moved when poking it with a thin wire.

[bdunham7]

Your printer board example is just that: anecdotal. Not that I do not trust you, but without even the details of temperature, time and equipment used, this is even more anecdotal than many.

Anecdotal or not, I think it answers your question as to whether it has ever been successful.  That was my point about the fallacy of dismissing anecdotes--one reliable anecdote can constitute proof for some questions.  People are serving prison terms based solely on anecdotes. Unless you think I'm lying, baking most likely did fix that one board in that case.  Why, how, temperatures, reasons, all that--I really don't know.  But I don't need to just to answer that question.

Now if you want to know if it is a good idea, how reliable the repair is, etc, then you'd need some research.  But I really wouldn't bother, I've heard enough anecdotes to conclude that it probably isn't worthwhile for the most part, except for a last-ditch 'nothing-to-lose' attempt to extend the life of completely non-critical part. 

[Ranayna]

Do you remember at what temperatures you baked this board?

[Fraser]

It is my understanding that the baking of GPU’s was a very specific, targeted response to the design of the “Flip Chip” construction. That is to say, the Flip Chip uses metal bumps to connect the die to the BGA platform that is, in turn, soldered to the host system PCB.

It is well understood that the temperatures used to “bake” a GPU are not high enough to actually melt the BGA solder balls. Ignoring cracked BGA solder balls, it is actually the die connection bumps and under-fill that is targeted by the baking process. The underfill will soften at the temperatures used for baking and MAY permit reconnection of umps that have become intermittent due to manufacturing process flaws, thermal cycling or oxidisation. The thermal effects of baking on surrounding components must always be considered if a melted mess is to be avoided. Electrolytic capacitors and batteries are not very fond of high temperatures used in baking a whole PCB. Localised heating with appropriate hot air tools is a better approach.

I cannot debunk the baking process as it does work in some cases. Is it a real, long term and reliable repair ? In my personal opinion, no. I would not carry out such a procedure on an important or mission critical system. Home users can do as they wish though.

To understand what baking does, just look at the physics of the situation.

The temperatures normally used do not create a molten state in normal leaded or lead free solder so reflow cannot occur

Thermal expansion and contraction can ‘wipe’ two conductive surface against eachother to re-establish contact through minor surface oxidisation

The high temperature can deform the underfill of a Flip Chip or even warp some BGA platforms a little and so reconnect intermittent Flip Chip Die bumps.


What baking cannot do is reflow solder a PCB unless the melting point of that solder is achieved and such a temperature can be unhealthy for the other PCB components if not reasonably controlled. The PCB can become a total loss.

In Australia, we would have called baking a PCB “Bush Engineering”. Whilst it may work, it is really an emergency measure and better options may exist.

Some people just elect to bake a PCB out of desperation and lack of knowledge, rather than taking a more methodical approach. That is NOT true in all cases, but sadly Social Media spouted it as a “magic” solution to GPU problems and I saw a lot of incinerated PCB’s as a result.

Controlled temperature HOT Air/Infrared reflow is a legitimate repair technique when dealing with the BGA to PCB solder ball interface. That said, removal, re-balling and refitting is the correct approach for a repair that you want to warranty.

Fraser

[bdunham7]

Do you remember at what temperatures you baked this board?

IIRC it was about 275F (135C) in a completely uncalibrated convection toaster oven for about 20 minutes with an extended cool-down (I left it in the oven).  This was an experiment and I was going to try progressively higher temperatures until something happened, but it worked after the first cycle.

I agree with what Fraser wrote, this is obviously way below reflow temperature.  IIRC this board had an STM32 of some sort, but I can't be 100% sure of that now because I worked on a few different 3D printer boards at the time and I could be mixing them up. 

[Ranayna]

There is one scenario that was given to me elsewhere, that i actually found somewhat plausible after a bit of research.

A PC mainboard did not work anymore after a cooling spacer was mounted to it with significantly too much torque. After baking at low-ish temperatures (around 120 C i think), the board worked again, for how long i do not know.

I looked up thermal expansion factors and was surprised that the factors of copper and epoxy used in FR4 are so different, with the epoxy having a much higher factor, especially in the z-axis once you take the fibre weave into account.
https://www.pcbdirectory.com/community/what-is-the-coefficient-of-thermal-expansion-cte-in-a-pcb

So i can very well imagine that a trace damaged by compression due to too much torque can get contact again after baking, when the epoxy essentially swells with heat.