--Please share the power supply section of schematic.
This is how my FX3 is connected.
I tried to replicate this power distribution to my development kit (CYUSB3KIT-001), I haven't see the problem.
The difference I can trace between the kit and my board is:
CTL2 (J7) & CTL3 (H7) connected to 1V8
CTL9 (H6) & CTL10 (K5) connected to GND
DQ21 (J1) as a 10K pull-up to 5V
I2S-MCLK (C4) as a 270R in serie with a pull-up at 2V5
VBUS connected to VBAT
FX3 is in the USB fallback state so most of the pins are Tristate.
I removed the SPI Flash I had on my board and the 1V8 rail went to 3.2V (My SPI Flash was possibily the one clamping the rail).
I removed the FX3 to put in place one from another batch, same issue. I also cut the track that connect E10 to VBUS before soldering back the BGA, just in case.
This is the power sequence for my 5V, 1V2, 1V8 and VADJ (2V5). It can be seen that the 1V8 is okay and start jumping to 3V.
The moment the 1V8 rail start raising seems linked to the time when the POR release the RESET signal for the FX3.
If I force the RESET to be always low, the 1V8 still raise but only to 2V1.
After further tests it seems that somehow the 1V8 regulator "Open" its output and let it floats somehow.
Here is the voltage at Startup, I put some code in the EEPROM for the FX3 to "work" on it, few seconds at startup:
So the 1V8 is fine then once the FX3 as finished, the rail start to float. I also measured the current on my 1V8 rail. It is quite small and we can see that it reverses.
This is the power supplies for the sake of completeness:
My current workaround is to add a small resistive load at the regulator output.
-Please share the complete schematic of your design.