A customer send in a Switch Lite for repair. It needed a new Gamecard connector. The Switch came in parts and so I started with the replacement of the GC connector without testing. Old connector out, new in with in one take without any adjustment needed.
After reassemble I tried to start, but the backlight stays off. Touch works. With a external light source you can see the screen.
Next issue: No button works, only the home button, which goes directly to the SoC and not to the STM32 controller.
I replaced the backlight ic twice, but no change. Vin is present, EN is high, PWM is there, ISET only outputs 0.7V (should be 1.xV)
I should have put the mainboard in the shell and check if it is running fine before changing the GC connetor. I think it is the 31 pin one.
From my measuring with some oL findings, I assume that I delaminated the pcb partial while soldering the new connector. Would be the first Switch mainboard which I damaged this way with my temp settings. And it was a smoth change. No long heat, no reheating for adjustment… So now I have to look for a replacement for my customer or offer him to buy it.
doesn’t BL run over the ribbon to the right daugher board on Lites? isn’t the trace burning out a semi common fault on lites or am I misremembering?.
Either way, can’t you just verify you have continuty from the output of the BL IC to the connector incase you have the break you mentioned? as it seems image is working fine behind the scenes for you
Yes, the BL run over the ribbon to the daughter board. I tested the daughter board assembly with a known working mainboard and it worked without any problems (daughter board buttons and BL).
And I tested the continuity from the daughter board through the flex connector of the mother board in question. All lines are present.
The STM32 ic which manage nearly all buttons and both analog sticks get the inputs but isn’t communicating with the SoC. The one line to the daughter board speaker isn’t working (oL). It is missing contact to the other side of the mainboard. The GC reader isn’t recognizing a GC. (At least I hear no signal for a new game).
I see, but BL should still work right regardless of if the STM32/GC is doing right?
Are you sure LCD is good on your patient (incl ribbon for BL) ? Not really documented anything relating to BL on Switch and the datasheet is a bit of a mammoth so I’ll have to go over it properly but, if I recall, there is a diode providing the output via the BL IC, then if I recall it has to have the load connected too (the LEDS in the panel, unsure if a common connection or an isolated line) otherwise I don’t think it will boost the output for BL (so if your LEDS in the panel are bad for example) Given your enable is high and your seeing PWM activity the only remaining variable is BL support circuitry being at fault, or the traces, BL ribbon to the panel / panel LEDs etc assuming I’m remembering right… I could well be wrong…grain of salt
the yellow marked area is raised and it can be pushed down. It feels like two or three layers deep. In the layer scans there are many vias in this area. So I guess a repair with jumper wires will not be a good solution.
The BL should be technical working. But I guess that the PWM signal is not submiting min freq for “ON”.
ah I see, yeah that’s a shame. I first discovered this sort of problem myself on a board which I ultrasoniced and did not bake for long enough afterwards, moisture was still in between the layers it seemed and during rework it evidently came up to boiling point before the moisture could escape through existing channels and the steam took the path of least resistance (i guess) and the layers bubbled up like in your case… learnt my lesson after that and baked at 70c after that for over an hour following IPA bath to disperse as much moisture from the ultrasonic as possible. Of course over and extended heat abuse can cause it too like you mentioned and also just regular liquid damage and lack of drying too.
That board looks quite shiny, did you or your client / former tech put it through the ultrasonic prior?
Sometimes you get lucky and it’s just the upper and/or lower layers that have bubbled and you can scrape them off and back like you would for a char case and conduct the repair and all is well, but depends on the area covered i suppose if it’s worth doing.
not sure. Did you verify IFB1 and IFB2 are making it back to the connector / BL IC ?
Only reason I ask is
When you say output do you mean BL IC side (input) ? because as far as I remember this is just a pull up resistor to a primary rail (I forget which) and just sets current, or if you mean the otherside of the resistor voltage is low then that’d be reflecting that primary rails voltage is low (?)
In this case I’m lost some how. the red readings are from a working Switch Lite and the yellow ones are from the Switch in question.
EN is high on both
PWM is high with 1.8V for full brightness on both
PWM is low with 0V for min brightness on both
COMP is the only difference with 3V steady on the defect one.
After the IFBx pin status detection, the COMP pin voltage starts ramp up, and the boost starts switching.
When the load current decreases along with the dimming duty, the output of
gm amplifier — COMP pin voltage decreases until it is clamped at an internal reference voltage. Because COMP
pin voltage controls the inductor peak current, when it is clamped the inductor peak current is also clamped and
cannot decrease. As a result, more energy than needed is transferred to the output stage, and the output voltage
and IFBx pin voltage increase. An internal hysteresis comparator detects the minimum IFBx pin voltage. When
the minimum IFBx voltage is detected as higher than the regulation voltage 90 mV by around 120 mV, the boost
stops switching. Then the output voltage, as well as IFBx pin voltage, decrease. When the minimum IFBx voltage
is lower than the hysteresis (around 40 mV), the boost switches again. Thus, during PFM mode the boost output
trips between the low and high thresholds. When the load increases along with the dimming duty, the COMP pin
voltage will exit from the clamped status, and the boost will exit the PFM mode and return to the PWM operation,
during which the minimum IFBx pin voltage is regulated at 90 mV again
Did you verify that from the SW diode, IFB1 and IFB2 (if applicable) has continuity alll the the way back to the LCD panel backlight connector on the daughterboard?
Only reason I ask is because the only thing I can think of is the comp pin is being driven higher in order to compensate for an issue with the load, which the only thing i can think of would be a short on the LED’s in the panel or more likely an open as mentioned on the lines above. 3V on the COMP line is the max, so probably indicating a gross condition such as a complete open or complete short, this may explain why the voltage measured on ISET is skewed too (I mispoke before, thought they were using a pullup on this line but it’s actually a pulldown) . The only other thing outside of what i mentioned that I could think of would e a bad IC or bad related passive causing the issues but nothings jumping out at me from your image
Given it was indeed an open line, then best i could say for the inverted image would be the same, though you’d be leaning more towards the LCD in that case… outside of that I’d be stumped
Where the real fun begins, keep us updated with photos of the carnage
I finally found the missing trace which prevent the STM32 from communicating with the SoC. Restored the line from pin 33 to the other side and the inputs are back again.
Not sure if you scraped back the layers or not when repairing the vias/traces but if you didn’t, might be worth just scraping the solder mask off on the upper/lower layers of the affected area, around traces as best you can and filling the voids in with UV mask, just in order to prevent the delamination from spreading in the future.
Probably will be fine without and I’d be more concerned if it was closer to a larger BGA but can’t hurt
Even if the Switch is acting ok now, I will keep it for testing and not sell it to someone as working unit.
The layer on top and beneath are full of lines. It isn’t an option for me to go through the this layers to reach the void. I tried to open the section from the side, but there isn’t a visible crack between layers.
not sure. Did you verify IFB1 and IFB2 are making it back to the connector / BL IC ?
