I have a switch lite I have just been sitting on for ages because I saw a bluescreen and just went no, not yet.
But I needed to test a new screen so plugged it in to this one as I figured I could at least test the backlight and… blue.
But what I found was that it behaved very inconsistently.
Most of the time, it would go to a black screen, and it would flash the home light once when doing this. Sometimes is would flash the home light a lot, and then go to a bluescreen. Sometimes it would be solid blue, sometimes it would only stay blue for a quick flash and sometimes it would flicker on and off.
I did a quick scan around the board to see if any of the obvious things were wrong, M92 and BQ caps all seemed reasonable, but I found that the fuel gauge was a mess. After a few goes with aliexpress chips not wanting to stick, I took one off another board and reballed, which went on ok.
I now still get the black screen most of the time, SoC gets warm and it draws 0.45A. When it does bluescreen, it seems to flash on and off, and the Home light is no longer flashing at all.
Obviously the bluescreen makes me think that we have a loose ball under the SoC or RAM or something, but the inconsistency confuses me. Has anyone seen this before?
So after burning my finger on the back light ic, i replaced it and now when i get a blue screen it is no longer flickering. But I still usually get a black screen. Is there something I should check first before doing.things like reflowing the RAM and Soc? I have a one of the cheap pi based boards if installing that would give more info.
Nothing boots when applying pressure, but when it is just on a “black screen”, the RAM gets too hot to keep pressure on, more so that the apu which seems weird.
Check primary rails resistance to ground, particularly 1V8PDR and 3V3PDR
Ram shouldn’t be getting too hot to touch in any circumstance so I’m guessing either bad prior rework on one of these IC’s or a failed IC (maybe caused by bad rework), if the SoC is getting hot too and seems hotter than the Ram then same possiblity I guess. You can measure resistance to ground on one of the caps alongside ram (1V8PDR ) and note the reading, do the same on the other alongside the other Ram IC and then again on the caps on the SoC wafer which correlate with this rail and if the rail was short / low, see which readiing constitutes lowest resistance to identify most likely culprit. (provided your meter has the resolution in resistance of course)
If it turns out to be one of the Ram modules, I’d go ahead and replace them both as a matter of course.
Noticed some missing caps, else where on the board, so sorted those out, and activated some flux under the RAM in-case there was something under there shorting things. No change. I measured the heat of the RAM, the top chip gets up to about 50c, the bottom one 45c, the CPU is sat somewhere in the 30’s.
Seems about right off the top of my head for Lite/Mariko, though best to compare to a known good Mariko/Lite to be sure (same with the other rails too)
There is another ram rail of interest too, I think it measures something like 1V15 (provided by the PMIC too) as far as I remember this also shows up on the other cap/s alongside the Ram IC’s.
Make sure to check resistance to ground on 3V3PDR also.
Seems a bit too hot to me, especially just in idle. Top one would be the prime candidate but prior to removal I’d first verify resistance on the caps alongside each IC and note the readings, if one is lower/short you can verify and then confirm following removal to ensure this was actally the fault and the fault is cleared.
Might be the case it’s some ram I/O lines are short in which case primary rails may not be afected or affected significantly, so rail checks may not show… though in my eperience, you do generally see a slight skew, though it’s small in these cases.
At the PMIC,
Top right is: 3.58k
Top left:2.8M and rising
Left: 18.5r
Bottom left: 324.5r
I don’t have a known good, but I do have a know very bad, and even that has 1k on the bottom left.
The tiny caps next to ram, some are at the 333r area, and some at the 75r area.
On my very bad board, those 333 caps are 1k, and the 75r ones are… in the k’s, but fluctuate.
This rail can alternate, also ram IC manufacturers / variants can give different readings too.
In any case, I don’t think your meter has the resolution to help you much here and the lack of a known good doesn’t help
Given that, I think your best bet is just to replace the ram ICs. I wouldn’t bother attempting a reball of the existing as by the sounds of it they have been operating at somewhat extreme temps. Just make sure your replacement IC’s which you use are the same brand, if your patient is using Samsung then stick to Samsung if it’s using Micron then stick to Micron etc
You may be right, the 333-ish readings were about 3 ohms different between the two chips, but with the top one having the higher reading. I do have a donor board with the same type of RAM chips, so I guess I start there.