Switch OLED issues with modchip, blue screen without

[RubyRoid]

I'm having issues with my freshly modded Switch OLED.

It was working and booting fine as soon as I installed the modchip but it was late so I shut it down and put it on charger for the night. The next day I went to install a larger SD card and got an error message. I've tried the old card but that didn't help. I've tried several other known good cards with no success. I've opened an issue in the hekate github repo and got a response that it's either fake or corrupted SD card or the SD connector itself (Update: contacted CTCaer directly and we came to conclusion that the issue is with one of the RAM modules). I don't think it's either one of those since I've been using tested good cards and opened the Switch and carefully inspected the SD connector and found no issues, all the pins look fine, nothing is bent or damaged.

Since I couldn't make the modchip work I decided to disconnect it. I desoldered and isolated all the wires from the chip and tried to boot the console. Got a blue screen. Tried to boot it holding the VOL+ and VOL- buttons but still getting the blue screen.

I'm not sure if simply desoldering the wires from the chip is enough to properly unmod the Switch but I couldn't find any proper instructions online.

Is there a way to make my Switch work again?

 

[RubyRoid]

All of them. But don't simply rely on diode voltage. You really want to check resistance to ground. And since you took off all the wires, look at the clk pad very carefully, and make sure you don't see a hole in the middle of the pad. Did you at any time use solder wick? If by chance you have a squeeze bottle with a needle applicator cap, shoot some IPA under the cpu from the opposite side of the clk point (pouring it won't have the desired effect... need that squeeze bottle with the small needle applicator). watch very carefully for a bit of solder or wick. If a solder ball comes out, then there's your issue, and would explain why squeezing the cpu into the board did not get rid of your blue screen.

Yeah, I've got this bottle which I loaded with 98% IPA and a proper needle applicator cap:

I blasted some IPA from the opposite side of the CPU a couple of times but nothing came out except for clear IPA. No solder balls or wick particles. I did it with EMMC and RAM chips as well with the exact same result.

I left the CLK point soldered to the board though. I've resoldered it once already and don't really want to go through the process again. I didn't see any holes on the pad last time I soldered to it.

This is how the CLK point area looks at the current state:

 

[Rodel]

Not yet. I've got a soldering station with hot air blower but I didn't use it much. What is the recommended temperature and duration to reflow the chip? How to confirm that it reflowed successfully?

just add flux and heat temperature about 280° for 1 minute.

 

[RubyRoid]

A, C, D, 3v against ground.

I've tested those multiple times, found no shorts.

 

[Rodel]

if you just try all, like reflow the apu,emmc etc. nothing works..means the apu or emmc has been damage.

 

[RubyRoid]

if you just try all, like reflow the apu,emmc etc. nothing works..means the apu or emmc has been damage.

I've just tried to reflow the EMMC, but that didn't help. Are you sure that 280°C is enough?

 

[doom95]

No, 280°C is not enough to reflow the eMMC. But that's not (currently) your issue.

First I would use a thicker wire, only for the 3v3 point and GND. The others carry signals so the very thin wire you used is excellent.
Next you should shorten the length of the CLK and DAT0 wires as much as possible. Leave next to no slack. These carry very high frequency signals and having them too long negatively affects that.
Finally if neither of this solves it, reflow the FPGA on the modchip. I've initially only read about, but recently also experienced myself, that poor soldering on the FPGA can cause very strange issues.

 

[doom95]

Short wires example. CLK runs underneath the shield along the RAM, and surfaces right in front of the solder point on the modchip. DAT0 here is without the adapter since it was a EMMC upgrade as well, so that wire is here soldered directly to the ball underneath.

 

[urherenow]

What kind of paste is that? Some can be capacitive, and others conductive. Either of those can cause trouble when drowning the caps like that,,, and you used too much. Use a card of some kind to sort of squeegee a thin, even layer on, and it won't do what I'm seeing here, and still do its job well. It only needs to fill microscopic holes/dents in the metal, to maximize surface-contact area with the shielding/heat sink.

Doubtful that has anything to do with your problem, just an observation. Trouble with that would likely give you a dead switch without even a green screen. I simply can't zoom in with enough clarity because you're not using a microscope. From what I see, you could have either taken too long or used too much heat, and almost flowed the cmd resister off... it almost looks crooked. Of course a problem with that resister would also likely give you a purple screen instead of blue, so we're still back to a problem with an IC (emmc, ram, or cpu).

@doom95, he already disconnected the modchip and is still stuck with a blue screen.

@RubyRoid, Doom95 is the man with this stuff. I'd listen to him before listening to me...

 

[doom95]

Disconnecting the wires from the modchip still leaves the wires soldered to the switch itself as antennae on the high frequency signal lines, so that's still important!

 

[RubyRoid]

@doom95 Thanks for your feedback! Currently I'm trying to at least make my Switch work again without the modchip. If it works I will try to mod it again and run shorter wires as you've suggested. Also great job, your install looks very nice and clean! What kind of wires did you use?

@urherenow It's the stock thermal paste, I don't think it is conductive as it covers the capacitors like this straight out of the box. Regarding the resistor, it's fine, not crooked or anything. Sorry I can't provide a clearer picture, I've already ordered the microscope you suggested, will be in my hands in a month or two I hope

 

[doom95]

Wires underneath BGA chips should always be the thinnest you can get your hands on, in my case it's AWG40 (solid) magnet wire.
The other stuff is AWG36 (stranded) telfon. Looks pretty, bends nicely. Can be had very cheaply on aliexpress.
Good luck getting the switch to work again. They're usually resilient.

 

[RubyRoid]

Run extended tests on the RAM, they show errors with one of the modules. Does anybody know how to reflow the RAM chips?

 

[zakwarrior]

Hi, thanks for sharing your experience. I agree, it is pretty similar to the issue I'm having with my Switch.

I have a few questions:

1. Do I have to desolder the wires from qsb and 3.3V point before probing?
2. In case my EMMC failed, am I screwed if I did not backup any keys?

Again, i don't see why it would of damaged anything with the RAM, RAM errors happen and they are build to have spare zone in case some places of the ram are dead, it will automaticaly move data to the working spare zones, so i wouldn't worry about the RAMS, also how on earth would the RAM be damaged ? because a reflow is to repair the balls and as i understood the OLED didn't fall or take any hits right ? so for me playing with the rams is a waste of time, this is not the problem IMO

Also reflow at 280° that wont work, never except if you use like me a small preheaterthat you put to 230-250° then yes using 280° on top would work, in fact this is the best to NOT damage anything 250 max under and 280 on top else you would need to use like 400 on top if you heat only one way, and doing this you can damage things around

1. Well normaly you prob when you install the qsb and you prob in order to know if the qsb is installed correctly, so yeah of course you prob before soldering the wires or you can prob now but take wires out first
I never recommanded that qsb from the v4 oled because you can really damage your emmc

2. No there are solutions, you could buy another EMMC same one 64GB but you could even also change to a 32GB from the v1 or v2, as for the keys you are lucky they are stored outside the emmc so even with a blank emmc you could get them passing by hekate, now it kinda hard work because you need a donor nand, you need the keys from that donor nand to decrypt it, then you need to generate a fake prodinfo using a donor prodinfo, inject all this in the rawnand decrypted and then encrypt the nand using your keys from your switch

Now the problem with that since you don't have your prodinfo is that after this the switch can only work hacked and you will never be able to run as official and go online

In my opinion you don't even need to reflow or reball the emmc because you didn't break any balls from the emmc they are soooooooooooo soooooo small that they don't break, forcing with the qsb would not damage the dat0 ball, even if you force the qsb it would damage the qsb connection not the ball, for me you clearly made a short in between 2 balls and burned your emmc

If you check the data sheet you you probably shorted dat0 & dat1, it does work for a short time but then leaving this short for few hours with switch on would for sure kill the emmc,

Hopefully i'm wrong....

So first thing i would do is take out the qsb, check what happens but i think you will still have blue screen

You can try reflow but in my opinion the balls are not damaged so i would be a waste of time (but again this is just my opinion respect to the person proposing this, and you can still try)

Second would be to get a EMMC and replace the old one, i HIGHLY recommande you to get this stencil and system to block your emmc else its just almost impossible to reball and you cannot reball just using solder on each connection)

Once replaced just solder back everything up, like someone said you should always use thicker wires for VDD and VSS (GND and 3.3v) at the LEAST for GND

I would NOT use hwfly 0.7.1 as i got my self wierd problems with the firmware sometimes, i would install the lasted ORIGINAL firmware from this chip, which is the firmware r0.31 chipnx

If everything soldered correctly, you will be able to load hekate and then lockpick, then after that it's a lot of software stuff

But first you need to get the hardware done

Oh last thing for the QSB i recommande highly this one (not the first version with on it written ALLIGN ME!)

This one you can almost not install it wrong, since the balls from the emmc blocks it in the exact place, then you just prob and verify the diode value to GND and to 3,3v, if values are good solder ir where you cut the shield out and it wont move, don't need to glue it

Once you got the hardware on control and hekate loading, send me a mp and i can guide you with the software stuff, but i have a repair business and i'm pretty busy and i'm not always checking gbatemp so give me one day to reply

Voilà i hope it helps

P.S: One thing i just thought about, you said it worked once right ? well you could be lucky! If you loaded atmosphere and didn't touch or erase the SD card after that, you have your prodinfo !! Because it does a backup normally and you can find it in atmosphere/automatic backup, you will have there your keys and the podinfo
With this original podinfo you will be able to load the switch again in stock and use it normally, but you must change the EMMC for a 64GB; it will not work with a 32 (only with hack)

 

[RubyRoid]

Again, i don't see why it would of damaged anything with the RAM, RAM errors happen and they are build to have spare zone in case some places of the ram are dead, it will automaticaly move data to the working spare zones, so i wouldn't worry about the RAMS, also how on earth would the RAM be damaged ? because a reflow is to repair the balls and as i understood the OLED didn't fall or take any hits right ? so for me playing with the rams is a waste of time, this is not the problem IMO

Also reflow at 280° that wont work, never except if you use like me a small preheaterthat you put to 230-250° then yes using 280° on top would work, in fact this is the best to NOT damage anything 250 max under and 280 on top else you would need to use like 400 on top if you heat only one way, and doing this you can damage things around

1. Well normaly you prob when you install the qsb and you prob in order to know if the qsb is installed correctly, so yeah of course you prob before soldering the wires or you can prob now but take wires out first
I never recommanded that qsb from the v4 oled because you can really damage your emmc

2. No there are solutions, you could buy another EMMC same one 64GB but you could even also change to a 32GB from the v1 or v2, as for the keys you are lucky they are stored outside the emmc so even with a blank emmc you could get them passing by hekate, now it kinda hard work because you need a donor nand, you need the keys from that donor nand to decrypt it, then you need to generate a fake prodinfo using a donor prodinfo, inject all this in the rawnand decrypted and then encrypt the nand using your keys from your switch

Now the problem with that since you don't have your prodinfo is that after this the switch can only work hacked and you will never be able to run as official and go online

In my opinion you don't even need to reflow or reball the emmc because you didn't break any balls from the emmc they are soooooooooooo soooooo small that they don't break, forcing with the qsb would not damage the dat0 ball, even if you force the qsb it would damage the qsb connection not the ball, for me you clearly made a short in between 2 balls and burned your emmc


View attachment 324332
If you check the data sheet you you probably shorted dat0 & dat1, it does work for a short time but then leaving this short for few hours with switch on would for sure kill the emmc,

Hopefully i'm wrong....

So first thing i would do is take out the qsb, check what happens but i think you will still have blue screen

You can try reflow but in my opinion the balls are not damaged so i would be a waste of time (but again this is just my opinion respect to the person proposing this, and you can still try)

Second would be to get a EMMC and replace the old one, i HIGHLY recommande you to get this stencil and system to block your emmc else its just almost impossible to reball and you cannot reball just using solder on each connection)

View attachment 324337
Once replaced just solder back everything up, like someone said you should always use thicker wires for VDD and VSS (GND and 3.3v) at the LEAST for GND

I would NOT use hwfly 0.7.1 as i got my self wierd problems with the firmware sometimes, i would install the lasted ORIGINAL firmware from this chip, which is the firmware r0.31 chipnx

If everything soldered correctly, you will be able to load hekate and then lockpick, then after that it's a lot of software stuff

But first you need to get the hardware done

Oh last thing for the QSB i recommande highly this one (not the first version with on it written ALLIGN ME!)
View attachment 324339

This one you can almost not install it wrong, since the balls from the emmc blocks it in the exact place, then you just prob and verify the diode value to GND and to 3,3v, if values are good solder ir where you cut the shield out and it wont move, don't need to glue it

Once you got the hardware on control and hekate loading, send me a mp and i can guide you with the software stuff, but i have a repair business and i'm pretty busy and i'm not always checking gbatemp so give me one day to reply

Voilà i hope it helps

P.S: One thing i just thought about, you said it worked once right ? well you could be lucky! If you loaded atmosphere and didn't touch or erase the SD card after that, you have your prodinfo !! Because it does a backup normally and you can find it in atmosphere/automatic backup, you will have there your keys and the podinfo
With this original podinfo you will be able to load the switch again in stock and use it normally, but you must change the EMMC for a 64GB; it will not work with a 32 (only with hack)

WOW! Thanks for such an extensive post! This is really helpful!

It would be difficult for me to go over everything you said but I'll try:

1. RAM. I got a special payload from CTCaer to test RAM and it shows lots of errors. We're unsure if one or both modules are bad, but this made us to believe that there is something wrong with at least one of the chips.

No, I didn't drop the console and handled if very carefully, I also didn't do anything funny around the RAM chips during the modchip installation. The closest I got to them is while soldering to CMD point.

2. I agree, I might have easily shorted the DAT0 and DAT1 points during QSB installation as I didn't have much experience. I only verified the diode value to GND, not 3,3v.

Yesterday I removed the QSB but the console still showed the blue screen. Later I had to reinstall it back to run the test payload another time and had an opportunity to probe the diode values again. It shows 404 to GND (or 605 if I swap the multimeter leads around), but zero to 3.3v point. Anyway, the modchip training went fine and I was able to run payloads again.

3. I also ordered a new QSB, very similar to one you suggested, hope it will do the trick next time:

4. I found a way to enter hekate menu on my Switch. On the minerva error screen I eject the SD card first and then press a button, it opens the text menu. There I was able to enter console info and see the eMMC info. Does it look right?

5. In case I need to replace the eMMC chip, where can I find a replacement? I searched eBay and AliExpress for "switch emmc" but only found 32GB boards for original Switch. Also what other equipment to I need to do it? I currently only have a soldering station with a heat gun.

6. Regarding automatic backups on the SD. I didn't know it's a thing so I removed the files on the card and overwritten everything a couple of times in attempt to make hekate boot. Anyway, I tried to see if it's possible to recover the deleted files and I think I succeeded! I got both BISKEYS and PRODINFO files, compared them to files from other Switch and they look legit

Once again, thanks for such a detailed answer and willingness to help!

 

[RubyRoid]

Update: I was able to get an access to eMMC partitions with NxNandManager and an experimental payload I was asked not to share, and my eMMC looks completely fine. I was able to mount it and browse the folders and see the contents with no issue.

So it is most certainly not the eMMC issue after all and everything again points toward the failed RAM chips.

I feel sort of relieved that one of the potential issues can be crossed out but on the other hand seems like RAM chips issues are not that common and I have no idea how to diagnose it further.

That would be awesome if anybody has any suggestions on what steps should I take next.

 

[Rodel]

looks like your APU is the problem.BSOD common issue is the APU.if you did reflow the chip make sure you move it and proper heat to flow in ball.

 

[RubyRoid]

looks like your APU is the problem.BSOD common issue is the APU.if you did reflow the chip make sure you move it and proper heat to flow in ball.

I haven't tried reflowing APU or RAM chips. Do I need any other tools except for hot air gun on my soldering station? What temps/duration should I use?

Also I'm not sure the flux I use for soldering would work for this task. What flux would you recommend?

 

[zakwarrior]

On question can you dump you emmc or not ? Does it dump without error or do you can an error

Your case is really strange... I mean blue screen don't just come like that, and also missing a part of data in emmc can also give a blue screen

 

[Rodel]

I haven't tried reflowing APU or RAM chips. Do I need any other tools except for hot air gun on my soldering station? What temps/duration should I use?

Also I'm not sure the flux I use for soldering would work for this task. What flux would you recommend?

Need to reflow the APU,the temperature must be in 480° and fan is 5 or 6 in about 5-8minutes,use Amtech 559 flux.need to use pcb holder and a tweezer to move a little bit of the APU make sure it move to flow the solder under the the APU.

 

[Rodel]

On question can you dump you emmc or not ? Does it dump without error or do you can an error

Your case is really strange... I mean blue screen don't just come like that, and also missing a part of data in emmc can also give a blue screen

Yeah i also experience repairing a corrupted emmc but the common issue i repaired is reflowing the APU to solve the problem.