0
\$\begingroup\$

I am trying to power the esp32 with the mic29302 regulator. I have used ams1117 for various applications but in this particular circuit I need more current (2 A) so I switched to mic29302 which can provide more current.

However, my ESP does not power on. I get 3.28 V on my power lines, and all other LEDs and modules power up but my ESP does not function. it doesn't send anything through a serial port, and the LCD remains white.

when I replace the mic29302 with an ams1117 the processor works fine. But I am struggling with the question of what happens when I connect mic29302. the fun fact is that I get 3.28 V on my power lines and mic29302 works fine. but the ESP malfunctions.

does anybody have an idea? thank you for your help in advance.

schematic of the mic29302 circuit. I have used the values in the datasheet.

\$\endgroup\$
2
  • 1
    \$\begingroup\$ Have you checked the power supply with an oscilloscope when connecting power? What happens if you force the ESP32 to reset with the reset pin while power is on and then release reset? \$\endgroup\$
    – Ralph
    Commented Apr 27 at 9:23
  • 1
    \$\begingroup\$ More information is required to determine the answer: please include oscilloscope waveforms of the input and output voltages, and the type of C14. \$\endgroup\$ Commented Apr 27 at 10:12

2 Answers 2

1
\$\begingroup\$

The problem cannot be solved by measuring output voltage with multimeter. Oscilloscope should be applied. LDO is very sensitive to output capacitor and may oscillate. Try to use output capacitor according to manufacturer recommendation.

\$\endgroup\$
2
  • \$\begingroup\$ He has followed capacitor recommendations in the data sheet as far as I can tell. \$\endgroup\$
    – Andy aka
    Commented Apr 27 at 10:03
  • 1
    \$\begingroup\$ @andyaka how far can you tell and how you got that information? Tantalum capacitor should be used. \$\endgroup\$
    – user263983
    Commented Apr 28 at 8:35
1
\$\begingroup\$

Here's an example good datasheet transient response from LDL1117: note the horizontal axis is 50µs/div, it only uses a cheap 4.7µF MLCC on the output, preload current is only 10mA, and all the necessary details are specified. This makes it a useful graph, you know what you're getting. On ESP32 boards I always replace AMS1117 and the tantalum cap with LDL1117 with 10µF MLCC, and that solves all power related crashes.

enter image description here

Looking at the datasheet for MIC29302 there are several things that I don't like. One is the text that says "fast transient response" but there is no usable transient response spec.

Normally you'd expect the transient response graph to have a horizontal scale in the 10-100µs/div to show all the details, but here we get 1ms/div which means all we see is a meaningless blip. Also I have no idea what the vertical scale is.

enter image description here

In addition, the test circuit uses a 1000µF low-ESR (40 mOhms) tantalum capacitor which is more expensive than doing the whole thing with a buck converter. And the LDO is preloaded with 200mA before the test (current steps from 200mA to 3A). It is much harder to get good transient response when stepping to/from very low current like a few mA because the transconductance of the pass transistor falls at low current. But 200mA to 3A is much easier.

So it looks like they did everything they could to put lipstick on it, and you won't get the same performance with a general purpose 100µF cap on the output because its ESR will be much higher.

But that's only a hypothesis. In order to confirm, you need to probe both the input and output with an oscilloscope.

Maybe the LDO's output drops under load, but it could also happen on the input depending on what the source of that 5V is.

If you don't have a scope, you can try putting a 1000µF cap with reasonably low ESR on the input, then on the output.

If putting it on the input solves the problem, then the prime suspect would be the 5V power supply dipping under load.

If putting it on the output solves the problem, then it's the LDO transient response.

Note ESP32 will pull 250mA spikes when the wifi transmits, so it really needs a lot more than 100nF low-ESR/ESL (ceramic) capacitance. 10µF ceramic would be a good starting point, and it's the first thing you should try.

\$\endgroup\$

Not the answer you're looking for? Browse other questions tagged or ask your own question.