3.53V on input, 2.61V on the output.

There’s your problem! The BLE chip isn’t getting enough voltage, likely because you’re overloading the port with that device requiring more current than the NES port can supply…

I don’t know enough about the NES to walk you through how to mod it to increase the available current, and I’m unfortunately not seeing any immediately available guides on the problem your facing but your two options would be to see if there’s some current limiting inside the NES for those ports (and risk full device brownouts, overloading, damage to power further upstream) or isolate the existing power rail and essentially replace it with the USB power adapter… Or just use the external power adapter…

From what I got, the SNES, N64 & GameCube controller ports are outputing 3.3V directly, not 5V.

Doing a quick google, this excerpt about the GameCube is enlightening:

There are two power rails on the connector, a 3.43V supply that is probably used for the logic, and a 5V supply that appears to be used to power the rumble motor (and perhaps logic also).

I’m willing to bet the 5V for the rumble is what is being used to power that module as it had significantly higher current capacity and would explain why it works on that device but not the NES.

Given that the Blueretro is taking 3.3V apparently, is it possible to step down from 4.6V to 3.3V instead? Is it wiser than stepping up?

That’s what the AMS1117 you identified does! One of the pins on that IC will be the 4.6V input, one will be the 3.3V output. Looking at the datasheet it has a dropout (minimum vin-vout) of 1.3V meaning that voltage regulator doesn’t have much margin…

Power issues/brownouts do seem like a possible explanation. Great job at tracking the issue down as far you did, but I think it’s a bit to early to jump to the conclusion that that is definitely the issue.

• What’s is the voltage you measure on the AMS1117?

• Does the voltage you measure change when you connect via Bluetooth?

• Do your measurements change when USB powered?

• Does the 4.6V output from the controller drop when you connect over Bluetooth?

• Are you measuring the ports when something is connected or when the ports are open?

• Does your blueretro work on the ports of the other NES devices?

My hunch is the ports don’t output enough current for reliable Bluetooth which isn’t going to be fixed without some NES surgery… You might be better off just using the USB power.

More abstractly what you’re doing with the resistor is creating a very crude linear regulator, which is fine for most applications and if you’re careful about keeping your source voltage close-ish to the forward voltage of the LED this method can be fairly efficient.

Using an active constant current supply (as an example or many dedicated LED driver ICs do something very similar) can be marginally better as it allows you to reduce the waste from the linear regulator.

However, if efficiency is what you really care about you’ll need to go with a switching regulator. Here’s an app note going over the basics of that approach. and again you can usually find dedicated ICs for that approach.

Overall I’d recommend doing a detailed power budget and really seeing whether it’s worth the cost/trouble of implementing that because while you are correct it is usually more energy efficient it can be significantly less labor/material/maintenance/longevity efficient (hence the prevalence of the humble resistor…)