Gnd and 3v3 pins are connected thru wire-wrap IC socket pins.
330uFD 10V capacitor is used.
I set the objectives for a versatile 3.3V_output power supply for Omega 2+
Abbreviated design description:
Hope this info is useful for some design ideas.
In state (2) / case (2), how did the USB-serial converter connects to Omega 2+?
Is it three-wire (Rx, Tc, and Gnd) or 4-wire (Rx, Tx, Gnd, and Vcc)?
Also pls show in details how your lm1117 based circuitry supplies power to the Omega2+ board.
(Wire gauge, wire length, filtering capacitors adjacency to the regulator IC, board layout/fly-wiring, etc.)
Power supply design is a serious business, especially for a picky board as Omega 2+ is.
The fact that case (1) showed up means you have a marginal power supply.
You need to have a power supply that can truly deliver 250mA and handle transient/surge demand well.
(Don't read just the simple spec. That's where people got fooled.)
Under this situation, anything can happen...
Not yet. Pls be patient.
Also note that MT7688 does not have a crypto accelerator module inside to help the encryption/decryption process thus the poor CPU will have to do the hard labor.
I'd suggest: give that CPU/SoC a break and let it do what it is designed to do; no more and no less.
Answering OP's question, Omega2 as a USB device
I think the best solution is to use a $2 shipped board such as ATMega32u8 USB module, running some type of special firmware, such as "Rubber Duckie" emulator.
From a different viewpoint, Rubber Duckie + WiFi for remote control will be overly devious.