Hacking Hardware Picofly - a HWFLY switch modchip

[TankedThomas]

Just been up all night finally installing a chip in my Switch Lite. It half-works, and this has probably been answered before, but I haven't yet found anyone with the exact same error so I was wondering if anyone can give me some advice please.

No photos right now but the soldering is relatively clean.
I used two bare mosfets (would use a flex cable if I did it again though because they're hard to secure).
I had some problems getting solder to stick to the little square (GND) on the processor's substrate but I seemed to pull it off in the end.

So the Switch still boots into OFW. The RP2040-Tiny is flashed properly. When I turn it on, I get the blue light, then what appears to be the "CMD not connected" flash, short-long (three times), then once I shut the Switch down, it turns blue again (I assume that last one is normal.

I forgot to check continuity before I put the IHS back on but everything else seems normal.
The only potential problem I've found is from doing a continuity test while the battery was plugged in (probably not the smartest idea but I didn't think it'd be a major).
Bridging CMD and GND (I used the microSD shielding) made the lights flash on the RP2040-Tiny.
Then I tried it a second time, but then it wouldn't power on with the button. Maybe something to do with me checking continuity between everything else and GND the go around?
Maybe pulling RST to ground made it boot? I don't know. I unplugged the battery, plugged it back in, and it boots normally again. I think this is probably all intended behaviour though.

If need be, I'll upload photos when I wake up, but it's taken me several hours, and the ultraviolet light they include with the solder mask stuff is completely useless (and I can't find any proper consumer-grade ultraviolet lights in New Zealand), so for now, I'm tidying up and going to bed.
If anyone can help though, I'd really appreciate it.

 

[karatefeet]

Yep, its weird because I used low melt on Samsung EMMCs and they have been 2 years going strong. One would think the same method would translate to any other EMMC but these SKHynix ones seem to have...real problems compared to other EMMC manufacturers. Were your recalls SKHynix ones too?

My thing now is too wait for the aerosol flux remover to arrive and wait for a recall to come back in lol because I have fixed it with a re-flow so maybe another 3 months!? I won't do anything to it except heat it up a little to soften the flux and then spray the aerosol under neath the EMMC too rule out if its the flux thats the cause or not. My money is on the low melt but I want to cancel out variables where we can.

~Meanwhile, last week I bought an OLED tablet for one purpose only....attempt the Kamikaze! And luckily it was also an SKHynix EMMC so it worked out well in terms of testing this method with consistency within our variables.
I've heard people on here asking how to practise the kamikaze method and its an easy answer imo. I have some Switch Lite donor boards and practised on random spots till I was comfortable. I got lucky and found a spot where there were lines running a layer or two underneath. Exactly what I needed to practise on because I was planning to tap into the Dat0 line on the reverse of the motherboard like @CarlosCruz did. If I didn't have a donor board, I would find some scrap pcb around and try it on that. Even if it was a RP2040 board! You just want to get to grips with the pressure and accuracy of the grinding pen and your hand. I have to give it to @CarlosCruz for giving such clear steps on how he did the method and here were my results:

View attachment 446116View attachment 446117View attachment 446122

For the life of me, I couldn't get solder paste to form in the well-like hole. It was frustrating. I tried using some solder on the tip of my iron to dip inside to the point but the line couldn't hold the ball. Too loose it seemed. Solder paste just clumped and didn't form a ball. Might be the paste itself being old as I hardly used it for a year and the casing of it was damaged so maybe oxidation ruined its chemical properties!? Luckily, I saw @FXDX used a repair pad (ones used for cpu line repairs) to make the connection. I had a bunch of Relife ones in abundance and it was super easy. No heat station needed. Just tinned the exposed line (carefully), put the Relife repiar pad on that line, quick touch or two with the iron, removed the excess tail of the new line I put in and then checked the connection on my DMM. Solid strong 0.710v. Solder masked it all and then was pretty much done with the Kamikaze. First time is always rough, same with the first time re-balling, but now I know what works for me. Next time will try to make a more even 'drill'...maybe wider and more symmetrical but I think I can stick with the Kamikaze in future.

The results were great too. The EMMC backup of the NAND had time of 36min 17s. Never had that on a SK EMMC before. Those are usually Samsung backup times. Benchmarks all passed too and glitch timing were quick (2-5 seconds, although I use mosfets and pull down resistors so they were glitch timings generally are a non issue for me, but still quicker than before regardless.)

I have ordered some smaller tip bits for the grinding pen so I can have a bit more accuracy but thank God it was good result. (kapton tape around so debris doesn't go too far away and its honestly doable.) I think Kamikaze is the most reliable post mod method so, my thought process is, by tools for it (like I did for the re-ball method) and get good and better at it. But not everyone will have the same thoughts on this but to each their own.

Yes I know one was a SKHynik for sure and believe the others were as well.

 

[Takezo-San]

Yes I know one was a SKHynik for sure and believe the others were as well.

That is no way a damn coincidence

Post automatically merged: Jul 9, 2024

Just been up all night finally installing a chip in my Switch Lite. It half-works, and this has probably been answered before, but I haven't yet found anyone with the exact same error so I was wondering if anyone can give me some advice please.

No photos right now but the soldering is relatively clean.
I used two bare mosfets (would use a flex cable if I did it again though because they're hard to secure).
I had some problems getting solder to stick to the little square (GND) on the processor's substrate but I seemed to pull it off in the end.

So the Switch still boots into OFW. The RP2040-Tiny is flashed properly. When I turn it on, I get the blue light, then what appears to be the "CMD not connected" flash, short-long (three times), then once I shut the Switch down, it turns blue again (I assume that last one is normal.

I forgot to check continuity before I put the IHS back on but everything else seems normal.
The only potential problem I've found is from doing a continuity test while the battery was plugged in (probably not the smartest idea but I didn't think it'd be a major).
Bridging CMD and GND (I used the microSD shielding) made the lights flash on the RP2040-Tiny.
Then I tried it a second time, but then it wouldn't power on with the button. Maybe something to do with me checking continuity between everything else and GND the go around?
Maybe pulling RST to ground made it boot? I don't know. I unplugged the battery, plugged it back in, and it boots normally again. I think this is probably all intended behaviour though.

If need be, I'll upload photos when I wake up, but it's taken me several hours, and the ultraviolet light they include with the solder mask stuff is completely useless (and I can't find any proper consumer-grade ultraviolet lights in New Zealand), so for now, I'm tidying up and going to bed.
If anyone can help though, I'd really appreciate it.

Definitely need pictures and details of what you're using.

Sounds like you were not taking your time with the mod. Using a multi=meter, even if it itself is not turned on, is asking for shorts because your probing with metal endings. Always disconnect the battery when performing the mod or doing tests/diagnostics. The fact that it still works in OFW is a sign that you got lucky because the Lite is notorious for being voltage sensitive and probably has the highest bricking rate when it comes to performing this mod.

What gauge wires are you using for the data lines and power/ground lines? What mosfets are you using? I initially had trouble with the irfhs8342 and I'm assuming you're using them too. I ditched them for a single AON6554 and its chunky and big but its easier to work with and glitching is faster. Downside is, it has to be installed on the rear of the motherboard but at least that way I don't have to cut any shielding between the ram and cpu to fit it in. Others may have different preferences, you got to find yours. You can ground anywhere where ground is, probe around and find a place that is easier to attach to.

What brand flux are you using and what type of solder are you using? Have you cleaned flux residue off of the cmd point more than adequately. If I remember correctly, the Lite's Dat0, cmd and clk points are quite close together so be thorough with the IPA99 or flux remover of your choice. Are all points getting the correct ranged diode readings? What values are you getting?

Why did you bridge cmd and ground? Are you using a microscope?
The UV light in 'bundled' mask packages are useless. People have found they're just purple coloured lasers and don't beam out in the UV spectrum. I think @lightninjay used a uv laser or you can invest in a UV Lamp. Or at the very least, get it all working and worry about masking up later. You can even get away with kapton tape over the points if you'rereally strapped for cash. Better than nothing.

 

[NintendoDisco]

Having the ** code (two short yellow blinks)

Used the guide to figure out it's the RST connection.

I went to resolder and I noticed that the pad on the ribbon is gone... what are my options?

 

[Phantomas77]

...
First, yes I also made the mistake of using low melt to help in wicking off the old solder from the motherboard and eMMC. Posted about the same issue a few months back (6?) and was also told to stop using it. At the time, I would get returns with the same issue you described and I suspect it's from people who constantly dock their OLEDs. I know sthetix just posted a video talking about the different Dat0 methods and he had the same suspicions.

Guys, why have you used low-melt for cleaning the remaining solder in the first place? Where did you get that info from?

I imagine it was from one of the sthetix's videos, I like to watch them and in overall he has good insight about modding procedures, but when I saw that he used low-melt in one of his videos to clean the solder before the reball in OLED - it was a huge red flag, I'm not surprised that now he's saying that reballing is not reliable, lol. Low-melt is not necessary for that application for sure, nonetheless for many others. If you have good equipment and some basic skill, you can do well with the Sn / Pn 63/37 or even with Pb-free solder and avoid future headaches.

Reballing/reflow soldering is a well know technique, when using correct materials and following an adequate temperature profile it's solid. [/rant_off]

Reballing, kamikaze, adaptors, black magic, etc for connecting DAT0 --> by the end of the day it's a personal preference and should be balanced by the risk/reward ratio in my opinion, and that's what I love about the modding scene, people coming together with different techniques in order to make this thing work

 

[MopSec]

Guys, why have you used low-melt for cleaning the remaining solder in the first place? Where did you get that info from?

I imagine it was from one of the sthetix's videos, I like to watch them and in overall he has good insight about modding procedures, but when I saw that he used low-melt in one of his videos to clean the solder before the reball in OLED - it was a huge red flag, I'm not surprised that now he's saying that reballing is not reliable, lol. Low-melt is not necessary for that application for sure, nonetheless for many others. If you have good equipment and some basic skill, you can do well with the Sn / Pn 63/37 or even with Pb-free solder and avoid future headaches.

Reballing/reflow soldering is a well know technique, when using correct materials and following an adequate temperature profile it's solid. [/rant_off]

Reballing, kamikaze, adaptors, black magic, etc for connecting DAT0 --> by the end of the day it's a personal preference and should be balanced by the risk/reward ratio in my opinion, and that's what I love about the modding scene, people coming together with different techniques in order to make this thing work

Just chiming in here because I have the exact same problem.
I've reballed and resoldered the eMMC in my OLED twice now (Samsung in my case) and both of those times it only worked for three or four weeks. I also followed sthetix's video and used low melt solder for wicking.
I'll try again next weekend and I'll definitely only use Sn63Pb37 and I'll make sure to clean the flux under the eMMC more thoroughly. Fingers crossed.

 

[Mr Skinner]

My dearest feelaas,

I received this Switch Lite the day before yesterday with a purple screen. I proceeded with the installation of the chip to try to boot Hekate in the hope of saving 400 hours of Pokemon Violet. From what I understand, the console entered the purple screen after installing a Gulikit (Hall Effect Joystick).

Picofly was successfully installed. However, when turning it on with SD and the Hekate files, a strange screen... A kind of strange purple flickering appears. And surprisingly, the CPU gets slightly hot... Even with it turned off.



Has anyone experienced this? I just want to boot Hekate and try to save nand + prod keys to extract the save.

 

[bvang913]

Having the ** code (two short yellow blinks)

Used the guide to figure out it's the RST connection.

I went to resolder and I noticed that the pad on the ribbon is gone... what are my options?View attachment 446150
View attachment 446151

Already tried to help you in the AiO thread, but most people here will tell you the same thing here. Your only option is to wire directly from the RST point on the motherboard to the RST/B point on the chip you have. There is a reason why those chips have pads on there; for those that choose to wire directly instead of using those ribbons.

 

[Danook28]

Any One have info about sx lite chip mosfet digram DRIN and S And cpu gate point.... I desolder this mosfet for use.

 

[Takezo-San]

Guys, why have you used low-melt for cleaning the remaining solder in the first place? Where did you get that info from?

I imagine it was from one of the sthetix's videos, I like to watch them and in overall he has good insight about modding procedures, but when I saw that he used low-melt in one of his videos to clean the solder before the reball in OLED - it was a huge red flag, I'm not surprised that now he's saying that reballing is not reliable, lol. Low-melt is not necessary for that application for sure, nonetheless for many others. If you have good equipment and some basic skill, you can do well with the Sn / Pn 63/37 or even with Pb-free solder and avoid future headaches.

Reballing/reflow soldering is a well know technique, when using correct materials and following an adequate temperature profile it's solid. [/rant_off]

Reballing, kamikaze, adaptors, black magic, etc for connecting DAT0 --> by the end of the day it's a personal preference and should be balanced by the risk/reward ratio in my opinion, and that's what I love about the modding scene, people coming together with different techniques in order to make this thing work

Not sure where exactly but I saw a bunch of Northridge Fix videos where he uses low melt when wicking away non-leaded solder. I should've thought more logically about it. The pads will have that low melt residue and the balls formed atop will have THAT residue as its base. Makes more sense now. Never again. Also the D24 tip is an awesome buy
Do you have any recommendations for a good J tip?

 

[TankedThomas]

Definitely need pictures and details of what you're using.

Sounds like you were not taking your time with the mod. Using a multi=meter, even if it itself is not turned on, is asking for shorts because your probing with metal endings. Always disconnect the battery when performing the mod or doing tests/diagnostics. The fact that it still works in OFW is a sign that you got lucky because the Lite is notorious for being voltage sensitive and probably has the highest bricking rate when it comes to performing this mod.

What gauge wires are you using for the data lines and power/ground lines? What mosfets are you using? I initially had trouble with the irfhs8342 and I'm assuming you're using them too. I ditched them for a single AON6554 and its chunky and big but its easier to work with and glitching is faster. Downside is, it has to be installed on the rear of the motherboard but at least that way I don't have to cut any shielding between the ram and cpu to fit it in. Others may have different preferences, you got to find yours. You can ground anywhere where ground is, probe around and find a place that is easier to attach to.

What brand flux are you using and what type of solder are you using? Have you cleaned flux residue off of the cmd point more than adequately. If I remember correctly, the Lite's Dat0, cmd and clk points are quite close together so be thorough with the IPA99 or flux remover of your choice. Are all points getting the correct ranged diode readings? What values are you getting?

Why did you bridge cmd and ground? Are you using a microscope?
The UV light in 'bundled' mask packages are useless. People have found they're just purple coloured lasers and don't beam out in the UV spectrum. I think @lightninjay used a uv laser or you can invest in a UV Lamp. Or at the very least, get it all working and worry about masking up later. You can even get away with kapton tape over the points if you'rereally strapped for cash. Better than nothing.

Thanks for the reply. Just realised I didn't post in the AIO thread in my tiredness, oh well.

I'm ashamed to admit I spent about 12 hours on the whole mod, so I definitely took my time. And yes, I agree on using a multi-meter. Tired me was stupid top do it with the battery connected.

Quick-fire answers:
Wire gauge: 40AWG (have some 30AWG if necessary, but only used 40AWG)
MOSFETs: IRFHS8342TRPBF (sourced from Mouser - https://mouser.com/ProductDetail/942-IRFHS8342TRPBF specifically)
Flux: MG Chemical 8341 No Clean Flux Paste
Solder: Asahi 0.25mm Sn62/Pb36/Ag2 (contains ROL0 (rosin) flux)
Soldering iron: Pinecil V2 with ILS and C1 tips (incidentally, I don't like it as much as my old, underpowered TS80 and wish I'd just gone Hakko this time)

I've got about 20 years of soldering experience (of varying quality, mind you) so I try to only use quality products now. I'm no expert but I'd like to think I know what I'm doing (sometimes, anyway).

I used Kapton tape in the end since the "UV" light is indeed useless. I've spent a long time looking for proper ones for retrobrighting but they just don't seem to be available here in anything other than tiny torch size (and even then, they're very rare). Maybe that'd be adequate for this though.

I'm using an Andonstar 246S-M microscope - I doubt I could do this otherwise, honestly.

I cleaned the flux (mostly with 99% IPA - after swapping the two 47ohm resistors with 100ohm resistors on the RP2040-Tiny, I washed it with 70% IPA) as best I can though it's not easy, between tiny, fragile joints and cotton buds (q-tips) leaving fluff rather easily.

I was checking for continuity between the RP2040-Tiny and all the end points to make sure the wires were okay. I did do continuity checks before putting it all back together and plugging the battery in. My concern was I was getting continuity between DAT0 and CMD when measuring from the RP2040-Tiny and the end points on the board. I later found that I was just touching the shielding next to DAT0 when checking the continuity, hence why I was getting continuity between CMD and DAT0 (or more precisely, CMD and GND). Hence why I checked CMD and GND had continuity normally.

Bit of a pain in the ass if I have to get different mosfets since I'm only on holiday until the end of this week and I'll have to do another ~NZ$80 Mouser order but I will if I have to (need other parts anyway, just didn't want to spend the money right now).

I mounted the IRFHS8342s in the gap next to the processor substrate like in the PDF guide. Originally used Kapton + double-sided tape to hold them down but it didn't stay, especially once they were heated up with the soldering iron. Tried the UV crap, didn't work, and so I ended up throwing Kapton tape over the top in the end, which seemed to hold everything in place.

I assume I don't need SDA and SCL soldered. Doesn't seem like it'll help much.
Haven't done diode checks because I thought it was only for the chip itself. Must have skimmed that, need to find the details and have another look. It's not in the guide so I assume I'll find the info from searching the threads. Just need to know which points to measure and what the expected values are.

Gonna need to clean the thermal paste off the processor and such, then I'll get some photos under the IHS.
I recorded most of it with my microscope so I'll look through that shortly and get some new photos too.
Only got up an hour ago so I'll get some photos sorted out shortly and update this reply with them then.

EDIT:
I've attached photos of everything (sorry for the out-of-focus 3.3V one but I doubt it's retaking - the RPi wouldn't work if it was bad).
Looks like I forgot to clean the flux properly on the four points above the processor, so I'll do that now, but I have a feeling it's probably related to the mosfets. I bought 10 mosfets, so I can try another pair if necessary.
Everything else looks okay to me. Not amazing but adequate. Clearly it isn't though, since it doesn't work.

Worth noting that I routed the CPU wires from the mosfets through the hole in the side of the metal bracket and bent that one little flap up on the IHS so they weren't squashed. Doubt this is an issue but you never know.

EDIT 2:
My multimeter is pretty crap - I need to import a better one at some point. The diode mode is a combined continuity mode on this, as far as I'm aware (unless HFe is the proper diode mode but I'm pretty sure that's just for transistor testing).
The readings drop off immediately but rarely seem to be the same so I've provided ranges from the initial readings which I presume are the forward voltage:

GND: 60-150 (sometimes above or below, but usually around 115-150)
3.3V: 70-200 (usually ~150)
DAT0: 70-200 on average ( usually closest to ~75 but sometimes jumps up into the 100s or 200s)
CMD: 100-900 (usually around 300-500)
CLK: 80-150 (saw some 800s and 900s but they weren't frequent)
RST: 150-220 (saw one ~80 and a couple around 300 but mostly around 170-ish)
CPU: 50-120 (usually around mid-50s, sometimes went up to ~200-300)

To be honest, I'm not sure that I'm doing the diode test correctly. It's certainly not my forte (and a better meter would certainly help - this is just the hardware store special; a cheap Chinese multimeter basically).

Figured it out after spending more time searching, thanks to @cilione in the AIO thread. Black probe on ground, red probe on RPi points. I assume the unit of measurement is Volts.

3.3V: .684 (holding the wire in place, as I desoldered this and CPU wires so I could reopen IHS)
DAT0: .538
CMD: .535
CLK: .595
RST: 1.486
CPU: not tested because it was disconnected to open IHS and is two separate wires (saw a comment saying it wasn't necessary but can test if needed)

 

[lightninjay]

Any One have info about sx lite chip mosfet digram DRIN and S And cpu gate point.... I desolder this mosfet for use. View attachment 446209

If it's anything like the dual-mosfet that I have a diagram for, here's a top and bottom layout that may work for you.

 

[Takezo-San]

Thanks for the reply. Just realised I didn't post in the AIO thread in my tiredness, oh well.

I'm ashamed to admit I spent about 12 hours on the whole mod, so I definitely took my time. And yes, I agree on using a multi-meter. Tired me was stupid top do it with the battery connected.

Quick-fire answers:
Wire gauge: 40AWG (have some 30AWG if necessary, but only used 40AWG)
MOSFETs: IRFHS8342TRPBF (sourced from Mouser - https://mouser.com/ProductDetail/942-IRFHS8342TRPBF specifically)
Flux: MG Chemical 8341 No Clean Flux Paste
Solder: Asahi 0.25mm Sn62/Pb36/Ag2 (contains ROL0 (rosin) flux)
Soldering iron: Pinecil V2 with ILS and C1 tips (incidentally, I don't like it as much as my old, underpowered TS80 and wish I'd just gone Hakko this time)

I've got about 20 years of soldering experience (of varying quality, mind you) so I try to only use quality products now. I'm no expert but I'd like to think I know what I'm doing (sometimes, anyway).

I used Kapton tape in the end since the "UV" light is indeed useless. I've spent a long time looking for proper ones for retrobrighting but they just don't seem to be available here in anything other than tiny torch size (and even then, they're very rare). Maybe that'd be adequate for this though.

I'm using an Andonstar 246S-M microscope - I doubt I could do this otherwise, honestly.

I cleaned the flux (mostly with 99% IPA - after swapping the two 47ohm resistors with 100ohm resistors on the RP2040-Tiny, I washed it with 70% IPA) as best I can though it's not easy, between tiny, fragile joints and cotton buds (q-tips) leaving fluff rather easily.

I was checking for continuity between the RP2040-Tiny and all the end points to make sure the wires were okay. I did do continuity checks before putting it all back together and plugging the battery in. My concern was I was getting continuity between DAT0 and CMD when measuring from the RP2040-Tiny and the end points on the board. I later found that I was just touching the shielding next to DAT0 when checking the continuity, hence why I was getting continuity between CMD and DAT0 (or more precisely, CMD and GND). Hence why I checked CMD and GND had continuity normally.

Bit of a pain in the ass if I have to get different mosfets since I'm only on holiday until the end of this week and I'll have to do another ~NZ$80 Mouser order but I will if I have to (need other parts anyway, just didn't want to spend the money right now).

I mounted the IRFHS8342s in the gap next to the processor substrate like in the PDF guide. Originally used Kapton + double-sided tape to hold them down but it didn't stay, especially once they were heated up with the soldering iron. Tried the UV crap, didn't work, and so I ended up throwing Kapton tape over the top in the end, which seemed to hold everything in place.

I assume I don't need SDA and SCL soldered. Doesn't seem like it'll help much.
Haven't done diode checks because I thought it was only for the chip itself. Must have skimmed that, need to find the details and have another look. It's not in the guide so I assume I'll find the info from searching the threads. Just need to know which points to measure and what the expected values are.

Gonna need to clean the thermal paste off the processor and such, then I'll get some photos under the IHS.
I recorded most of it with my microscope so I'll look through that shortly and get some new photos too.
Only got up an hour ago so I'll get some photos sorted out shortly and update this reply with them then.

EDIT:
I've attached photos of everything (sorry for the out-of-focus 3.3V one but I doubt it's retaking - the RPi wouldn't work if it was bad).
Looks like I forgot to clean the flux properly on the four points above the processor, so I'll do that now, but I have a feeling it's probably related to the mosfets. I bought 10 mosfets, so I can try another pair if necessary.
Everything else looks okay to me. Not amazing but adequate. Clearly it isn't though, since it doesn't work.

Worth noting that I routed the CPU wires from the mosfets through the hole in the side of the metal bracket and bent that one little flap up on the IHS so they weren't squashed. Doubt this is an issue but you never know.

EDIT 2:
My multimeter is pretty crap - I need to import a better one at some point. The diode mode is a combined continuity mode on this, as far as I'm aware (unless HFe is the proper diode mode but I'm pretty sure that's just for transistor testing).
The readings drop off immediately but rarely seem to be the same so I've provided ranges from the initial readings which I presume are the forward voltage:

GND: 60-150 (sometimes above or below, but usually around 115-150)
3.3V: 70-200 (usually ~150)
DAT0: 70-200 on average ( usually closest to ~75 but sometimes jumps up into the 100s or 200s)
CMD: 100-900 (usually around 300-500)
CLK: 80-150 (saw some 800s and 900s but they weren't frequent)
RST: 150-220 (saw one ~80 and a couple around 300 but mostly around 170-ish)
CPU: 50-120 (usually around mid-50s, sometimes went up to ~200-300)

To be honest, I'm not sure that I'm doing the diode test correctly. It's certainly not my forte (and a better meter would certainly help - this is just the hardware store special; a cheap Chinese multimeter basically).

Your wiring looks good. Though your 3.3v line looks like a tsunami of flux hit it. Genreally on PCBs you'd never want to use anything less than 99% IPA or, even better, an actual flux remover. You mentioned you used 70% IPA. That other 30% is water. Your asking for problems down the line but you have a working switch so your fine. Though I would stop using that asap for flux clean up. Get 99% asap. Water is conductive and corrosive.

Since you are getting the *= CMD error and not =*= or ==* then that means the CPU line is fine so your mosfets might not be needing replacing. I started with an el-cheapo DMM too and thats good enough to get your readings for this mod. Some DMMs have continuity with diode mode and some have with resistance mode. As long as your on diode mode and the black probe is on a random ground and the red probe is on your seleceted point. You should be getting a voltage reading. I went to get some readings out of my donor lite board for you and they should be +/- 0.100v/0.200vish off these sorts of numbers (note these are taken at the end points from the pcb of the switch and not from the pico board points):

cpu capacitors (both) - 0.018v
clk - 0.778v
cmd - 0.782v
dat0 - 0.789v
RST - 0 or somtimes 1.5 (depends on your DMM)
3.3v - 0.831v
GND - 0v

Double check these with your settings on your DMM and see if you can get or are getting readings similar to those. Start with getting your readings to 0 on a ground point. Ground should always be zero or next to zero. If the Pico is flashing CMD as the problem and your DMM is not working. Great, at least we can do one thing. re-do that line. Its a pain but atleast doing that, you'll rule out the CMD line being the problem. Meanwhile, priority is getting the DMM to read the correct readings or get a better one.

Though looking closely at your stills, the CMD point look like its not making a solid connection. And the wire used looks thicker than 40awg. They look more like 30awg. 40awg looks like a strand of hair under the scope. They look like a bunch of strands.

You said you weren't getting your readings to remain stable. That indicates to me a cold solder joint and or a bad connection or your not holding down the probes of your DMM securely enough to the connection. Can you check if this happens on a random point on the board. A cap or test pad or something. If it still happens, your multi-meter might just be the issue.
You've got a bunch of leads to go on here so I hope I've been of some help to you.
You can do this!

 

[TankedThomas]

Your wiring looks good. Though your 3.3v line looks like a tsunami of flux hit it. Genreally on PCBs you'd never want to use anything less than 99% IPA or, even better, an actual flux remover. You mentioned you used 70% IPA. That other 30% is water. Your asking for problems down the line but you have a working switch so your fine. Though I would stop using that asap for flux clean up. Get 99% asap. Water is conductive and corrosive.

I'm aware. Normally I'd have dried it out for a day first but I already spent a couple of days on prep (for better or worse).
I have a big bottle of 70% and a small bottle of 99% (the latter costs more), so I didn't have a ton of 99% to spare. Not ideal, I know, but I did make sure it was dried as best I could with a paper towel before it ever came near the Switch.
I only used it on the RP2040-Tiny by itself though, everything else was the 99% IPA.

Since you are getting the *= CMD error and not =*= or ==* then that means the CPU line is fine so your mosfets might not be needing replacing. I started with an el-cheapo DMM too and thats good enough to get your readings for this mod. Some DMMs have continuity with diode mode and some have with resistance mode. As long as your on diode mode and the black probe is on a random ground and the red probe is on your seleceted point. You should be getting a voltage reading. I went to get some readings out of my donor lite board for you and they should be +/- 0.100v/0.200vish off these sorts of numbers (note these are taken at the end points from the pcb of the switch and not from the pico board points):

cpu capacitors (both) - 0.018v
clk - 0.778v
cmd - 0.782v
dat0 - 0.789v
RST - 0 or somtimes 1.5 (depends on your DMM)
3.3v - 0.831v
GND - 0v

Double check these with your settings on your DMM and see if you can get or are getting readings similar to those. Start with getting your readings to 0 on a ground point. Ground should always be zero or next to zero. If the Pico is flashing CMD as the problem and your DMM is not working. Great, at least we can do one thing. re-do that line. Its a pain but atleast doing that, you'll rule out the CMD line being the problem. Meanwhile, priority is getting the DMM to read the correct readings or get a better one.

Though looking closely at your stills, the CMD point look like its not making a solid connection. And the wire used looks thicker than 40awg. They look more like 30awg. 40awg looks like a strand of hair under the scope. They look like a bunch of strands.

You said you weren't getting your readings to remain stable. That indicates to me a cold solder joint and or a bad connection or your not holding down the probes of your DMM securely enough to the connection. Can you check if this happens on a random point on the board. A cap or test pad or something. If it still happens, your multi-meter might just be the issue.
You've got a bunch of leads to go on here so I hope I've been of some help to you.
You can do this!

Just updated with the correct measurements right as you were typing, it seems! The unstable readings were me doing a continuity check instead of a proper diode check. Didn't realise I had to always be grounded on one side. Now they're stable.
They're about 200mV below what you got though, with RST just under 1.5. I'll check the CPU caps after lunch and get back to you on those.

Definitely 40AWG. I have 30AWG too (not used for this) and it's noticeably thicker. CMD is definitely a little shoddy so I'll try to clean that up. It got quite flattened out on the tinned end there for some reason - probably why it looks like 30AWG.

Here's hoping between the two of us, we get it working. Much appreciated already.

 

[Takezo-San]

I'm aware. Normally I'd have dried it out for a day first but I already spent a couple of days on prep (for better or worse).
I have a big bottle of 70% and a small bottle of 99% (the latter costs more), so I didn't have a ton of 99% to spare. Not ideal, I know, but I did make sure it was dried as best I could with a paper towel before it ever came near the Switch.
I only used it on the RP2040-Tiny by itself though, everything else was the 99% IPA.

Just updated with the correct measurements right as you were typing, it seems! The unstable readings were me doing a continuity check instead of a proper diode check. Didn't realise I had to always be grounded on one side. Now they're stable.
They're about 200mV below what you got though, with RST just under 1.5. I'll check the CPU caps after lunch and get back to you on those.

Definitely 40AWG. I have 30AWG too (not used for this) and it's noticeably thicker. CMD is definitely a little shoddy so I'll try to clean that up. It got quite flattened out on the tinned end there for some reason - probably why it looks like 30AWG.

Here's hoping between the two of us, we get it working. Much appreciated already.

Good stuff.

As long as the readings are above 0.500v ish, then they're ok. And I'm assuming the cmd lines and all other lines were checked at both ends!?

Let me know how you get on.

 

[jkyoho]

Thanks for the reply. Just realised I didn't post in the AIO thread in my tiredness, oh well.

I'm ashamed to admit I spent about 12 hours on the whole mod, so I definitely took my time. And yes, I agree on using a multi-meter. Tired me was stupid top do it with the battery connected.

Quick-fire answers:
Wire gauge: 40AWG (have some 30AWG if necessary, but only used 40AWG)
MOSFETs: IRFHS8342TRPBF (sourced from Mouser - https://mouser.com/ProductDetail/942-IRFHS8342TRPBF specifically)
Flux: MG Chemical 8341 No Clean Flux Paste
Solder: Asahi 0.25mm Sn62/Pb36/Ag2 (contains ROL0 (rosin) flux)
Soldering iron: Pinecil V2 with ILS and C1 tips (incidentally, I don't like it as much as my old, underpowered TS80 and wish I'd just gone Hakko this time)

I've got about 20 years of soldering experience (of varying quality, mind you) so I try to only use quality products now. I'm no expert but I'd like to think I know what I'm doing (sometimes, anyway).

I used Kapton tape in the end since the "UV" light is indeed useless. I've spent a long time looking for proper ones for retrobrighting but they just don't seem to be available here in anything other than tiny torch size (and even then, they're very rare). Maybe that'd be adequate for this though.

I'm using an Andonstar 246S-M microscope - I doubt I could do this otherwise, honestly.

I cleaned the flux (mostly with 99% IPA - after swapping the two 47ohm resistors with 100ohm resistors on the RP2040-Tiny, I washed it with 70% IPA) as best I can though it's not easy, between tiny, fragile joints and cotton buds (q-tips) leaving fluff rather easily.

I was checking for continuity between the RP2040-Tiny and all the end points to make sure the wires were okay. I did do continuity checks before putting it all back together and plugging the battery in. My concern was I was getting continuity between DAT0 and CMD when measuring from the RP2040-Tiny and the end points on the board. I later found that I was just touching the shielding next to DAT0 when checking the continuity, hence why I was getting continuity between CMD and DAT0 (or more precisely, CMD and GND). Hence why I checked CMD and GND had continuity normally.

Bit of a pain in the ass if I have to get different mosfets since I'm only on holiday until the end of this week and I'll have to do another ~NZ$80 Mouser order but I will if I have to (need other parts anyway, just didn't want to spend the money right now).

I mounted the IRFHS8342s in the gap next to the processor substrate like in the PDF guide. Originally used Kapton + double-sided tape to hold them down but it didn't stay, especially once they were heated up with the soldering iron. Tried the UV crap, didn't work, and so I ended up throwing Kapton tape over the top in the end, which seemed to hold everything in place.

I assume I don't need SDA and SCL soldered. Doesn't seem like it'll help much.
Haven't done diode checks because I thought it was only for the chip itself. Must have skimmed that, need to find the details and have another look. It's not in the guide so I assume I'll find the info from searching the threads. Just need to know which points to measure and what the expected values are.

Gonna need to clean the thermal paste off the processor and such, then I'll get some photos under the IHS.
I recorded most of it with my microscope so I'll look through that shortly and get some new photos too.
Only got up an hour ago so I'll get some photos sorted out shortly and update this reply with them then.

EDIT:
I've attached photos of everything (sorry for the out-of-focus 3.3V one but I doubt it's retaking - the RPi wouldn't work if it was bad).
Looks like I forgot to clean the flux properly on the four points above the processor, so I'll do that now, but I have a feeling it's probably related to the mosfets. I bought 10 mosfets, so I can try another pair if necessary.
Everything else looks okay to me. Not amazing but adequate. Clearly it isn't though, since it doesn't work.

Worth noting that I routed the CPU wires from the mosfets through the hole in the side of the metal bracket and bent that one little flap up on the IHS so they weren't squashed. Doubt this is an issue but you never know.

EDIT 2:
My multimeter is pretty crap - I need to import a better one at some point. The diode mode is a combined continuity mode on this, as far as I'm aware (unless HFe is the proper diode mode but I'm pretty sure that's just for transistor testing).
The readings drop off immediately but rarely seem to be the same so I've provided ranges from the initial readings which I presume are the forward voltage:

GND: 60-150 (sometimes above or below, but usually around 115-150)
3.3V: 70-200 (usually ~150)
DAT0: 70-200 on average ( usually closest to ~75 but sometimes jumps up into the 100s or 200s)
CMD: 100-900 (usually around 300-500)
CLK: 80-150 (saw some 800s and 900s but they weren't frequent)
RST: 150-220 (saw one ~80 and a couple around 300 but mostly around 170-ish)
CPU: 50-120 (usually around mid-50s, sometimes went up to ~200-300)

To be honest, I'm not sure that I'm doing the diode test correctly. It's certainly not my forte (and a better meter would certainly help - this is just the hardware store special; a cheap Chinese multimeter basically).

Figured it out after spending more time searching, thanks to @cilione in the AIO thread. Black probe on ground, red probe on RPi points. I assume the unit of measurement is Volts.

3.3V: .684 (holding the wire in place, as I desoldered this and CPU wires so I could reopen IHS)
DAT0: .538
CMD: .535
CLK: .595
RST: 1.486
CPU: not tested because it was disconnected to open IHS and is two separate wires (saw a comment saying it wasn't necessary but can test if needed)

 

[Takezo-San]

View attachment 446349

Bloody good eyes

 

[thesjaakspoiler]

I imagine it was from one of the sthetix's videos, I like to watch them and in overall he has good insight about modding procedures, but when I saw that he used low-melt in one of his videos to clean the solder before the reball in OLED - it was a huge red flag,

I think that @Phantomas77 is absolutely right with his suggestion to avoid using low melt solder.

Out of curiosity, I took a look at some emmc datasheets.
The power consumption of a Samsung emmc is like 200mA @ 3V.
The SKHynix emmc is like 600mA @3v.
That is 3 times as much current and thus also 3 times as much heat.
That is like 2 watts of heat in that tiny area.
The Samsung emmc gets up to 30~40 degrees celcius so I guess the SKHynix will get really hot.
Officially most emmcs can operate up in environments up to 85~105 degrees Celcius.
But I guess those SKHynix ones actually do actually get that hot when extensively used.

Low melt solder can have a melting point as low as 50 degrees Celcius.
Although most of us trust what is written on the package, cheap ass solder from Amazon/Aliexpress can be anything.

 

[TankedThomas]

View attachment 446349

I know it looks shoddy, but I'm pretty sure it's attached. I checked when I first attached them. I can reposition it again though.

Good stuff.

As long as the readings are above 0.500v ish, then they're ok. And I'm assuming the cmd lines and all other lines were checked at both ends!?

Let me know how you get on.

Yes, I checked them at both ends.

CPU readings are both 0.017mV.

I'm going to clean up CMD, that resistor, and the flux around those top points right now, then report back again.

Oh, as for 3.3V, it's not flux there, just that UV solder mask resin stuff that hasn't been (mostly) cured.

I found a few little UV torches from one of the big electronics retailers here but they're pretty expensive and don't seem powerful enough. I did find this though: https://www.pbtech.co.nz/product/MPPSGP05944/Spigen-Galaxy-S23-Ultra-5G-Premium-Tempered-Glass
Not cheap but coincidentally matches my phone and, more importantly, has a UV light to cure it. I'm thinking this might be worth getting just for the UV light but I'm not sure.

 

[Takezo-San]

I know it looks shoddy, but I'm pretty sure it's attached. I checked when I first attached them. I can reposition it again though.

Yes, I checked them at both ends.

CPU readings are both 0.017mV.

I'm going to clean up CMD, that resistor, and the flux around those top points right now, then report back again.

Oh, as for 3.3V, it's not flux there, just that UV solder mask resin stuff that hasn't been (mostly) cured.

I found a few little UV torches from one of the big electronics retailers here but they're pretty expensive and don't seem powerful enough. I did find this though: https://www.pbtech.co.nz/product/MPPSGP05944/Spigen-Galaxy-S23-Ultra-5G-Premium-Tempered-Glass
Not cheap but coincidentally matches my phone and, more importantly, has a UV light to cure it. I'm thinking this might be worth getting just for the UV light but I'm not sure.

Good stuff.
Can't say I'd go for that UV lamp. Something like this is what I'd go for personally.
What happened to that inductor near the Ground line? Mistaken point for ground at first!?

Post automatically merged: Jul 10, 2024

I think that @Phantomas77 is absolutely right with his suggestion to avoid using low melt solder.

Out of curiosity, I took a look at some emmc datasheets.
The power consumption of a Samsung emmc is like 200mA @ 3V.
The SKHynix emmc is like 600mA @3v.
That is 3 times as much current and thus also 3 times as much heat.
That is like 2 watts of heat in that tiny area.
The Samsung emmc gets up to 30~40 degrees celcius so I guess the SKHynix will get really hot.
Officially most emmcs can operate up in environments up to 85~105 degrees Celcius.
But I guess those SKHynix ones actually do actually get that hot when extensively used.

Low melt solder can have a melting point as low as 50 degrees Celcius.
Although most of us trust what is written on the package, cheap ass solder from Amazon/Aliexpress can be anything.

I don't know about you guys but I'm putting some thermal putty ontop of any SKHynix EMMC I come across. Just a pea sized portion but...I wonder