Shorted caps under cpu

I’m looking for some help here.
I’ve been given a switch to repair : the symptoms were that it won’t charge and won’t start.

I’ve checked the caps around the 2 usual chips (pi3usb, mt92xx) but couldn’t find anyone that was shorted.
I’ve also managed to charge the battery using my GSM phone charger (9V, usb-C) and the battery that read 0V before charging actually showed 3.7V after a 3 hours charging session. Still, the switch would not start.

I then checked all the big caps that are located under the cpu (i.e. on the opposite side of the cpu) and found several that were shorted. I haven’t unsoldered them to check further and went asking for some advice here and what should be done or done again in a different way to correctly diagnose the issue.


Some caps under the SoC have a very low resistance. It s possible that a low cost multimeter show them as shorted but they are not.

1 Like

I would start at the usb-c port.
Does the 5V enter the board through the port?
Does the 5V pass the fuse and reach the M92T36?
Does the 5V reach the BQ24193?

Thanks you Calvin,

My MM is a a high quality one (Brymen 867). But I was using continuity mode, I’ll check the internal resistance as seen at the cap level instead.

I’ll then proceed with your measurements this afternoon (now it is children time).
WWIS, I used a 5->9V QC mobile phone charger and 9V was measured at the USB-C ports. I’ll check that up to the 2 chips you mentionned.

I have checked and :

  • the 5V enter the board through the usb port
  • it passes the fuse and reach the M92T36
  • for the BQ24193, I read 4,2V but I might be looking at an incorrect pin, I failed to follow the lead to this chip.

for the BQ24193, I probed against its first pin and I also got the 5V

I do have a Vbus of 4.98V, and around 3.90V for SYS and BAT.

Does TegraRCM recognize the Switch without NAND?

Sorry, I’m not sure what you mean.
I’ve just installed TegraRCM and its drivers and it says “RCM Device detected” when I connect the switch.
What should I do from there ?

I had twice the problem that a Switch had a drained battery during running cfw. It wouldn t boot nor charge the battery. After inject Hakate with TegraRCM, I discovered that the Switch was in AutoRCM.

you need to watch this video…

1 Like

Calvin, thanks for your help.
But I’m very new to this Switch thing. I originally took the switch to check for a faulty chip to replace using flux/reflow equipment I have.
Do you mean I should try to inject Hakate with TegraRCM ?
Then what, should I be dumping nand, checking something with it ?

I am able to inject the payload, the console seems to do something as the backlight is now on. But I cannot see anything thing on the screen.

Everything is good. Learning by doing. :wink:

If you can connect the Switch witch a pc and are able to inject a payload and start a cfw like Hakate, it shows that the soc is fine and the Switch is not bricked and in most cases fixable.

If you can inject a payload and the backlight is switching on but nothing on the screen, you might have a problem with the display. I would check with a known good screen if the switch outputs a picture.

Thanks: )
Is there a way to check the display by reading voltages, testing continuity, etc. (I don’t have a spare display) ?

I only have a map with diode mode values.

Unplug battery and put your multimeter on diode mode, red probe on ground and test with the black probe the pins. You should get similar values.

1 Like

thanks Calvin, where is this part located (face and location) I cannot find it.

sorry for the dumb question, but how I can measure voltage without power injected (wo battery) ? did you mean measuring resistance between ground on each of the LCD pin for which you have values ?

it seems I’m just a newbie more…
interesting watch on youtube -> h4hpM8JyvM regarding diode mode measurement

In diode mode the voltage is delivered from the multimeter. It measures the voltage drop to ground and is much faster than measuring resistance to ground.

1 Like