Reducing ESP32-S3 current peaks during WiFi

[Someone]

Here is the official ESP32 data sheet:
https://documentation.espressif.com/esp32_datasheet_en.pdf

Page 52.

Transmit 802.11b, DSSS 1 Mbps, POUT = +19.5 dBm
Typ 240 mA

RF front ends usually operate at 10...20 % efficiency to avoid distortion. To output  100 mW, you will need about 1 W. It seems things agree. 3.3 V x 240 mA: 792 mW

Except the OP is having trouble with the peak current, which you conveniently chopped out of the "quote" (and only typical, not worst case).
Others measure pretty similar to the OP:
https://www.microelectronicos.net/measuring-power-on-esp32-modules/

Yes sure, I very conveniently chopped that out...

If you read the OP first post you will see that the peak current is 300 mA. See the screenshot which has been attached by the OP.

Orly?

How about the OP in full:

I have an ESP32 board that pulls about 150mA nominal, but 350mA peak during wifi transmission.
The peaks are only about 2ms, but trigger the over-current on the 250mA supply of the system (energy metering port).

The power flow is 5V > TLV62568A 1A buck > ESP32-S3
Currently I already reduced the wifi power from 21 to 11 dBm and added a 220uF elco to the 5V input.
The peak flattens and reduces a bit, but is still noticably there. Current profile measured with Nordic PPK2 is attached.
Now it is close to the 250mA limit, but with significantly reduced WiFi range.

How would I go about reducing these peaks so they get closer to the baseline?
More capacitance? Different type of capacitors? On the 5V rail, or on the 3V3 rail?
Some time of current clamp circuit?

Seems your claim of 300mA is nowhere to be seen, and walking away from the 240mA you waved about as if it meant something.

OP says 350mA peak on the 3.3V side, and more than 250mA on the 5V side with a plot having some exponential decay peaks at 240 something and 255 something.

Come back with some more attitude? This thread majority rubbish, mostly from one poster... you. Did you even simulate your proposed solutions? They aren't solving the problem of a 1000us+ long pulse of demand being passed through to the source.

OP is on the right track already:

Am looking into options with a shunt, op-amp and FET but not fully sure if that works out.

Yes, something is needed between the input and 1000 or so uF of storage (as an inductor >> 10mH is plainly impractical!). I thought through and simulated a current source as you're suggesting and it does seem to work ok but of course works best if you allow a significant droop on the intermediate capacitor. Opamp is non critical for bandwidth or precision so cheap and even 1c pFETs come with SOA covering startup and transients, you'd probably spend more on the passives in the circuit. Note you'd want to setup the UVLO on the following buck converter to handle the slow startup (more passives).

[temperance]

Come back with some more attitude? This thread majority rubbish, mostly from one poster... you.

For all you capacitor lovers, the entire thread is yours and yours only. I removed my entire contribution including a spice sim providing the evidence you were seeking so the OP can read about capacitors solutions involving very large capacitors solutions without getting disturbed by what I wrote.

I hope you like this fantastic news. Best regards.

[lutkeveld]

The ESP32 has indeed high current peaks. I don't think just a buck converters and regular capacitors are really going to solve the problem. The impedance of the P1 port's power supply is going to be ballpark lower than your capacitor, so it's going to try to supply this power.
You can go a few ways, basically you need to set a current limit lower. You can have a current limited boost converter go to let's say 24V and run an unlimited buck converter from that.
Or you can charge a capacitor, 5V with the P1 port, current limited, and have a buck-boost instead of a buck for the ESP32. This current limit can be as simple as a resistor. Maybe even a supercapacitor could work, something like 1F has plenty of energy to cover high current peaks and it's not as expensive as it sounds.

That is also what I am suspecting, because without a current limiter the peak only goes away to a certain extent when increase capacitance to quite high values, indicating it's still supplied by the P1 port. 1F would be quite bulky / costly, so trying to avoid that.

---------------------

I'm curious how far I could get by just buck IC selection and tweaking the output LC. Maybe have to get some SPICE going to see its effects.
Increasing the input capacitor can only be done to a certain extent before I need to add a current limiter for inrush anyway.
At that point, something like a AP22652 might work, although for a guaranteed max 250mA limit, the minimum limit will be ~180mA due to accuracy.
Will have to check if the baseline current of the ESP32 never gets that high, otherwise it will get clamped.