powering ESP32-S3 dev module on it's 3.3V "OUT" pin?

[eTimesV]

I've been looking all over for an answer to this but nothing concrete came up.

I have an ESP32-S3 development board:
https://www.aliexpress.com/item/1005006963491611.html

The product description has a pin diagram showing one pin as "3.3v (OUT)".

I would like to feed it 3.3V power via its "3.3V (OUT)" pin because I have a 3.3V source in the small device it's part of.  In order to get 5V I'd have to add another dc-dc converter or regulator or something, which I'd like to avoid to save space and cost.

I found some people saying you can power the ESP32-S3 via the 3.3V pin.
(not sure if that's the same as the "3.3V (OUT)" pin on this dev module).
Other people saying you cannot.  I can't really figure out for sure who is right and don't want to damage my S3 dev board.

I messaged the aliexpress shop, EGBO, and the response was "3.3V can be powered"
but I have doubts as to whether the person replying really knew what they were talking about.

This discussion,
https://www.reddit.com/r/esp32/comments/1clbmee/how_to_power_esp32_s3_mini_from_33v_source/?rdt=55564
has two conflicting reports: One person says you can use the "3.3V (OUT)" as input power as long as the dev board is not connected via USB-C.  The other says "You CAN'T power this board from 3.3V *reliably*."

Can anyone say for sure which it is?  The "(OUT)" on the 3.3V pin diagram does make me wary of using it as an "in."

[ppTRN]

The 3.3V are generated from the +5V (from the USB) via a linear regulator. If you apply a voltage to the output of the regulator, like in this case, nothing terrible should happend.

Many linear regulators have a bupass diode that bypass them if a voltage higher than the input is present at the output. If so, it could be even less riskyer.
But regardless of the bypass diode inside the regulator, i would do it.

[langwadt]

The 3.3V are generated from the +5V (from the USB) via a linear regulator. If you apply a voltage to the output of the regulator, like in this case, nothing terrible should happend.

Many linear regulators have a bupass diode that bypass them if a voltage higher than the input is present at the output. If so, it could be even less riskyer.
But regardless of the bypass diode inside the regulator, i would do it.

if you want to be sure just feed both 5V and 3.3V pins with 3.3V then there won't voltage difference across the regulator
, just don't plug in USB at the same time

[Smokey]

The 3.3V are generated from the +5V (from the USB) via a linear regulator. If you apply a voltage to the output of the regulator, like in this case, nothing terrible should happend.

Many linear regulators have a bupass diode that bypass them if a voltage higher than the input is present at the output. If so, it could be even less riskyer.
But regardless of the bypass diode inside the regulator, i would do it.

Good luck finding a schematic for an aliexpress board, but all the parts are on top and the picture is pretty clear.  Apart from the ESP32 there is only really one other chip, so that's probably your 3.3V regulator.

[ppTRN]

No need for a precise schematic. It must be a linear regulator, it's obvious since there are none of the typical components used to implement a buck converter.

But even if it was a buck converter, I still think that it would not pose a problem to just feed 3.3V on its output.

[globoy]

Unfortunately the regulator could blow up (or it could work).  It all depends on the internal circuitry of the regulator chip and that's most certainly not going to be documented for an Aliexpress special.

You might try using a lab power supply set to 3.3V output and a low current limit (e.g. 25 mA) and see if the regulator shorts the 3.3V to ground.  If it doesn't appear to be shorted to ground then you could gradually increase the current limit.  Probably by 75-100 mA the ESP32-S3 will be getting enough current to run (w/o any radios turned on).  Assuming the output is maintaining 3.3V then increase it further and see if the regulator gets hot.  If it runs cool then you're probably OK with this particular board.  Be aware that if you buy the board again in the future it could come with an entirely different regulator.

Or buy two boards and just give it a go with one.  If it blows up then you know (and you could replace the blown regulator with a new part for some future use).

Also remember that if this works and you decide to program the board in-situ then its regulator will be driving the other regulator in your circuit.

[eTimesV]

The 3.3V are generated from the +5V (from the USB) via a linear regulator. If you apply a voltage to the output of the regulator, like in this case, nothing terrible should happend.

Many linear regulators have a bupass diode that bypass them if a voltage higher than the input is present at the output. If so, it could be even less riskyer.
But regardless of the bypass diode inside the regulator, i would do it.

if you want to be sure just feed both 5V and 3.3V pins with 3.3V then there won't voltage difference across the regulator
, just don't plug in USB at the same time

Thanks for the replies.  In the meantime I see the esp32-s3 board product description says the onboard regulator is a "ME6217C33M5G Low voltage drop (LDO), current (Max)800mA".  I found that data sheet,
https://www.lcsc.com/datasheet/lcsc_datasheet_1912111437_MICRONE-Nanjing-Micro-One-Elec-ME6217C33M5G_C427602.pdf
I'm not sure what I can glean from that but this seemed like maybe it's relevant: "The ME6217 Series uses a low on-resistance P-channel MOS FET as the output transistor.  Be sure that VOUT does not exceed VIN + 0.3 V to prevent the voltage regulator from being damaged due to inverse current flowing from VOUT pin through a parasitic diode to VIN pin."

If I power VOUT could it be that VIN on the other side is floating at some voltage less than VIN+0.3V?
Does that 0.3V imply some kind of protection diode in the ME6217 to prevent reverse currents?

As for tying my 3.3V supply to both the 5V and 3.3V pins, could I just put 3.3V on the 5V pin (which I think is in fact designed as a power input pin)?  I mean, the 3.3V output pin on the dev board can remain unconnected as far as my requirements go.

Are there any non-destructive tests I could do to see if input on the 3.3V output pin is working safely or not?  ChatGPT suggested to use a multimeter to check for reverse current on the regulator's input pin when applying 3.3V to the output.  Does that make sense?  Any other tests I could perform?

[eTimesV]

You might try using a lab power supply set to 3.3V output and a low current limit (e.g. 25 mA) and see if the regulator shorts the 3.3V to ground.
... Assuming the output is maintaining 3.3V then increase it further and see if the regulator gets hot.  If it runs cool then you're probably OK with this particular board. 
...Also remember that if this works and you decide to program the board in-situ then its regulator will be driving the other regulator in your circuit.

I'm not sure how I'd see if the regulator is shorting to ground.  The esp32-s3 certainly uses more than 25mA and if limit the current supply to 25mA I would see it max out at 25mA in either case, shorting to ground or the esp32 pulling all the current it can get? 
Ditto for increasing the current limit.  It will go up to whatever the esp32 wants, something like idk, 100-150mA maybe (not transmitting).  I guess if it goes over ~250mA that could indicate a short to ground?
I'm missing something on how to execute this test.

"decide to program the board in-situ" ...I think you mean plug in the USB cable.  I would definitely try to avoid that.  Maybe I'll build some kind of hardware sliding cover into the case that prevents that from happening when I go between battery power and USB programming mode.

[eTimesV]

One other thing, some context for my original question since maybe someone has an entirely different and better idea on how to power this little project board I'm making.

I am using the ESP32 to read a vacuum pressure sensor.  It will run off two 18650's in series, which give 7.4V nominal but the battery output could go from ~6V min. discharged to 8.4V max charged.  The sensor requires 9-28VDC power.  The sensor outputs RS232 so I'm using a MAX3232 RS232-to-TTL converter, which requires 3.3V, to get the serial data to the ESP RX/TX pins.

I currently have one tiny dc-dc convert to give 12VDC to the sensor (I wanted to give it something above it's bare minimum 9V since I thought that might help minimize any effect of waning battery voltage) and one dc-dc converter to give me 3.3V for the MAX.  I was hoping I could get away with just two dc-dc converters by powering the ESP32 on its 3.3V output (or maybe giving 3.3.V to its 5V input?), the subject of this post.

Ideally I'd like some kind of tiny buck-boost board that takes 6-8.4VDC input and has three outputs; 3.3V, 5V and 12V (12 or whatever in that 9-28V range).  I didn't find any although I did find this board which takes 7-30VDC in (I just realized that 7V is higher than I'd like and won't let me discharge the batteries too much) and give 5 & 3.3V out:
https://www.aliexpress.com/item/1005007501353930.html
I ordered a couple in case it turns out the 3.3V into the ESP32 doesn't work.

Does anybody know of a little buck-boost board that can run off of two 18650's in series and has 3.3, 5 and ~12V outputs?
A totally different question, which I don't want to sidetrack the 3.3V discussion, but would designing a board like that be hard?  I'm no electronics expert but when I look at those little switching dc-dc converter boards they only have a dozen or so components.

[globoy]

Thanks for the replies.  In the meantime I see the esp32-s3 board product description says the onboard regulator is a "ME6217C33M5G Low voltage drop (LDO)

I stand corrected   

Looking at the spec I think that regulator might be ok to backdrive.  What they mean by "voltage regulator from being damaged due to inverse current flowing from VOUT pin through a parasitic diode to VIN pin"is that if Vout was > Vin and Vin was connected to something that could sink appreciable current then current would flow from Vout to Vin and heat up the parasitic diode that exists across the regulating MOSFET's Source and Drain pins which could damage it.  But if you have nothing to sink current it might be ok (which should be the case with no USB plugged in).  I think you shouldn't have any issue with Vin connected to Vout as langwadt suggested (other than the risks and items others have listed).

I'm not sure how I'd see if the regulator is shorting to ground.  The esp32-s3 certainly uses more than 25mA and if limit the current supply to 25mA I would see it max out at 25mA in either case, shorting to ground or the esp32 pulling all the current it can get?  Ditto for increasing the current limit.  It will go up to whatever the esp32 wants, something like idk, 100-150mA maybe (not transmitting).

It's possible this idea wouldn't work.  What I guessed, based on some experience, is that the PSU output would go to near 0 if there was a short.  Otherwise it would be something higher.  Perhaps it would hover around the under-voltage reset of the ESP32 as the chip went in and out of reset (while in reset it will consume very little power).  As you increase the current eventually the ESP32 will be happy and the rail will be 3.3V.  I was thinking to start with very little current because usually that's safe no matter what the regulator circuitry looks like.  It's likely that even at 75mA it would be safe since we're looking at heating related damage and the ESP32 will also be happy.

[eTimesV]

Here is a test of using a 3.3V supply attached to the module's GND and 3.3V output pin and cycling the current limit between ~25-70mA. It does look like under ~60mA the ESP32 is doing some kind of cycling as you can see by the regular variation in the displayed voltage level.  I think this is probably what globay meant with "hover around the under-voltage reset of the ESP32 as the chip went in and out of reset" ?



I forgot to check the regulator temperature.  Would that be meaningful to know with this test w.r.t. understanding if it's ok to power the ESP32 board via it's 3.3V OUT pin (as long as no usb connected)?

[Smokey]

The "cycling" is probably the RF radios transmitting.

[eTimesV]

Assuming the output is maintaining 3.3V then increase it further and see if the regulator gets hot.  If it runs cool then you're probably OK with this particular board.

Here's a thermal image video of giving 3.3V to the 3.3V OUT pin on the S3 board with current limit higher than the board wants to draw.


The regulator is (hard to see) to the right and just above the main S3 processor.  It does not seem to get hot although you can see the processor goes from ~24° to 27°C.  So...all systems go?  Any other tests I could do?  Run some program on the S3 to use the wifi and see what happens to the regulator when the board wants 200-300mA?

[globoy]

I think you're good to go!  Nice job on the experiments.

[Peabody]

The regulator will have a feedback resistor divider across its output.  So if you apply power to the output, with the input open, then a little current will flow through that divider.  But we're talking microamps.  But then there's the question of whether current will flow back through the rest of the regulator to its ground pin.  The only way to know for sure is get one of those regulators out of the circuit, apply a 3.3V supply to its output, and see how much current flows.  But I think in general this will work fine.

Of course one option would be to remove the regulator, and just connect power to the 3.3V pin, but then you wouldn't be able to use USB.  I do that routinely on my Pro Minis when I want the lowest possible sleep current when powered directly from a battery.  But the Pro Mini doesn't have USB in the first place, so nothing is lost.

I've never seen a report of any problems when powering at the regulator's output pin so long as nothing is connected to its input, at least with linear regulators.  So I suspect it will work for you.  One test you might do is put the ESP32 to sleep, and measure the current it's drawing from USB.  Then do the same measurement when powered at the 3.3V pin.  Since it's a linear regulator, the comparison should be valid, and I think the 3.3V supply should draw slightly less current.  But you just want to be sure the 3.3V supply doesn't draw substantially more current.