Smoke after direct 19v

#1

According to the wiki, the Rock5B supports being powered by up 20v fixed through the USB C port.

https://wiki.radxa.com/Rock5/hardware/5b#USB_Power_.283.29

I tried that, the power LEDs illuminated, but what I’m guessing is the power supply chip immediately started smoking :cry:

PXL_20230421_050852005
PXL_20230421_050852005.jpg1572×980 223 KB

(Small square chip middle-right is visibly damaged). Is my understanding incorrect here? What could have gone wrong to cause this, and can the board be repaired?

Here is the connector I used

https://www.aliexpress.us/item/3256804444151540.html

Could this have been a polarity issue? Would that have such catastrophic results?

#2

Digging through the schematics and I found that this is a FUSB302BMPX, the USB controller. I’ve placed an order for a replacement chip, but I’d really like to understand what went wrong here before I try anything else. Any thoughts or leads or info would be greatly appreciated.

#3

Maybe, the plug you used had reverse polarity?

#4

So far

Seems like there is something wrong with your cable.

But also there was commit that added 12v via overlay

But in main dtb it’s still 20v

So in short - don’t try this cable on your phone, would be better if you checked it on Type-C tester before using it on the board

#5

I exactly had the same problem on my board. FUSB burned randomly, my assumption was that the ripple in the converted DC was so high it exceeded >= 20V for too long. (i dont know the withstand of it).

However you can still push up to 19V to vbus of the type C header (could be more but put some margin due to crappy converters DC ripples) and your device should work fine. There are back to back regulators in the chain to give the correct voltage. But your PD is potato now. At least that was the case in my case.

#6

This 20V config works fine for my PD power. Straight DC will ignore this.

#7

I’m guessing your barrel to type-C connector applies incoming power to both VBUS and VCONN, which is unsafe (only VBUS is allowed to be higher than 5V, VCONN is not).

#8

Seems correct there 4 central pins, which is actually data

browser
Screenshot_2023-04-25-11-02-33-566-edit_com.opera.browser.jpg660×574 86.2 KB

Fig1m11292018.webp (25.5 KB)

#9

@Googulator It would appear that your conjecture is correct. It’s hard to say since the probes on my multimeter aren’t quite that fine, but my testing makes me fairly confident that there is continuity between the VBUS and CC1/CC2 pins. That sucks! Not only do I have to fix or replace my board, but I’ve got to find another part that doesn’t have that behavior :frowning:

#10

I found an ordered an alternative part that I think should work properly.

https://www.aliexpress.us/item/3256804730593984.html

The 3Pin R1 version breaks out VBUS, GND, and CC. If I just leave CC disconnected I think it would have the desired behavior. I’ll definitely be continuity testing beforehand and I’ll report back here if it works/doesn’t work.