ESP32 high power draw (fake chips?)

[no_esperience]

Hi!
I have a bunch of different esp32 bare modules that I jut can manage to power with voltage regulators due to voltage drop causing brownouts.

Unfortunately I have no way of measuring such low currents ATM, only a DMM.

I first tried a bunch of low iq LDOs from aliexpress: MPC1700, HT7833 and HT7333, they provide up to 500mA and people are using them in ESP32 projects successfully, No such luck for me, I added capacitors, electrolytic 1000uf and ceramic (10uf, 100nf) but it did not help.
I just figured I got fake regulators, but different models from different vendors to all be fake, seemed weird.
Then I tried using a few different LM1117 I had, which should be good for 800mA, same thing. But it was also form Aliexpress, so questionable.
Then I went and got a locally sourced KA78R33, which should be good for 1A current draw, same issue, brownout detected and voltage output drops from 3.3 to 2.9 according to my DMM.

I'm starting to think maybe there is an issue with the ESP32 modules.
The weird thing is that the ESP32 modules are working in an esp programmer, which uses an ams1117.

Any ideas?

[ESP_Sprite]

How are your modules connected to the LDO? If there's long wires or traces in the path, the inductance there may be an issue. Connecting something like an 10uF decoupling cap between gnd and 3v3 may solve that issue.

The weird thing is that the ESP32 modules are working in an esp programmer, which uses an ams1117.

Not that weird. The ESP32 CPU and memory don't use that much power, it's mostly the WiFi subsystem that leads to brownouts, and that isn't used when programming the ESP32.

[MicroController]

How are you supplying power to the regulators? Is the upstream supply sufficient in terms of voltage, current, and pulse response?

[no_esperience]

How are your modules connected to the LDO? If there's long wires or traces in the path, the inductance there may be an issue. Connecting something like an 10uF decoupling cap between gnd and 3v3 may solve that issue.

It was the bloody wires and or the breadboard I use to prototype this.
I soldered the components on a protoboard and suddenly it worked and is still working. Using MPC1700 (250ma max) a 1000uf and a 10uf ceramic.
I would still have to do more testing, trying the other LDOs, soldering more sensors to see if there will be any issue, but considering the wifi is the biggest power consumer, I very much doubt it.

Are there any other ways to prototype without having to solder stuff?

How are you supplying power to the regulators? Is the upstream supply sufficient in terms of voltage, current, and pulse response?

1 x 18650 li-ion but also tried 5A 5v PSU and 2 x 18650. It should not be an issue.

[ESP_Sprite]

It was the bloody wires and or the breadboard I use to prototype this.

Yeah, I'm not surprised tbh.

I soldered the components on a protoboard and suddenly it worked and is still working. Using MPC1700 (250ma max) a 1000uf and a 10uf ceramic.
I would still have to do more testing, trying the other LDOs, soldering more sensors to see if there will be any issue, but considering the wifi is the biggest power consumer, I very much doubt it.

Note that 250mA is under-dimensioned for an ESP32: the datasheet calls out a power supply that can handle 500mA at least. Your overkill 1000uF capacitor may be saving you here in most instances, but you may still see brownouts in some circumstances.

Are there any other ways to prototype without having to solder stuff?

I'd grab a devboard rather than a module; that way you at least are not dependent on the breadboard for the signals that critically need to be low impedance. From there, just go straight to the PCB design; Espressif devboards have schematics that are available, so you can simply copy those. (I personally don't even bother with breadboards anymore if I can help it: I simply design an initial PCB, potentially with multiple footprints, test points and zero-ohm resistors to optionally connect things, then prototype on that.)