Quote

First of all, that's trivial to do (don't see the issue) and secondly, you can use any type of connector to provide the extra power. It could also be feasible to solder thick wires with USB-A on the end, though, I suppose.
In case you want to do this, I have to revise what I said about soldering to large VDD and GND planes: the ones I thought about are obviously not at the right potential and would skip the power regulator circuit, so you need to solder to the existing USB connector's contacts (plus side: they're nice and big).

Why add a connector when you just not need any connector ?
Need a power supply a wire and a solder point for solder directly the copper cable on the copper PCB that's the better way for have the higher voltage/current. You seem never have measured any data cable and view issue at each connector. In high voltage even a damaged insulator can cause major issue.
People who want loss power on their USB connector use their USB connector it's not an issue but add a pad somewhere on the PCB change nothing with an easy way for cut power coming from USB. Voltage can't be take from 5V usb and another from another power supply for power.

Quote

That's wrong. 2.4A is definitely in spec. All my mobile devices are charged at 2.4A through USB-A - I think it's the USB PD 2.0 spec or something like that (commonly known as 'fast charging').
If you want to go higher, of course that's outside the specification, but that's why I said that sidehack makes these with 2.4A in mind.. of course you can overclock higher if you provide more power through specialized hubs, but you could run them in-spec.

USB PD 2.0 it's 20V@5A. USB norm is a voltage and a current within 5m cable look what happen with your 5V 100A charger at the end of a 5m usb AWG28 cable by measuring the voltage and real current at the end you will understand the issue. It's not for nothing than voltage increase to 20V for transfer 5A with USB-C.
Current is limited by cable section and length or even connectors.

Quote

Okay, so the idea is to keep the lower part of the stick (power supply circuit). Not sure that's going to save you that much money. Also it is virtually the same as in Compac, NewPac, 2Pac, but it is slightly tuned to the chip(s) that it powers, so they would need to make it somehow modular (move resistor to change core voltage and similar), and I'm not sure it makes sense to save at the very most $10 off a $250 device.

An industrial low voltage power supply isn't really cheap but yes you can keep all and you not lost any time because it's like hot plug an usb stick for replace the WHOLE miner and you can exchange easily each part (cable usb hub power supply computer for controlling stick and custom voltage source for each miner) without trash all like a pod. 5V USB or 5V from industrial power supply not need any change and like I already write a "chip" exchange = whole stick miner exchange not unsolder asic chip and solder a newer. Package probably not the same vcore change so no.
I just ask two copper pad and a 0ohm resistor or removable fuse for cut usb 5V supply and provide directly 5v to the pcb like probably all other people since 7 years asking for the same thing

Imagine how many time the people who made is own pod with usb stick lost time for design smaller water block put so many pipe who can cause leak too for almost nothing because hub port aren't correctly aligned for put one big water block. You think really that the cost is low ?
It can just align 4 or more miners and put even an already existing cooler because it's like VRM IC aligned on mainboard or graphic card so something probably already exist for put that under watercooling and cost are way reduced just need to solder 2 copper cables on each miner and use standard material for all other.