Under-voltage detected! Which is the maximum power voltage for Raspb3?

[Barabba]

Hi mates I've put the DC/DC converter (3Amper + 10.000uF at output) to 5.05volt to power up Raspberry 3, but seems it's not enough, about 3 to 7 minutes I've this warning on syslog. It looks I've to rise a bit the voltage, which can be the right one to don't exceed? Maybe 5,3volt?
Thank you

[HermannSW]

Interesting, the official Raspberry power supply has 5.1V with only 2.5A, and I never saw undervoltage when powering Pi 3B+ with camera with that supply:
https://www.raspberrypi.org/products/ra ... er-supply/

[pcmanbob]

What size are the wires between the DC-DC converter and the pi ?

You might also want to look at the wire size feeding the DC-DC converter.

If the low voltage alarm only comes up after a few minutes of use I would check the DC-DC converter is not over heating and limiting its output.

[mahjongg]

The maximum allowable (long time) voltage is that of an USB device, 5.25 Volt.

To try to avoid permanent damage to attached equipment (and itself) the RPI (except zero's) shorts the 5V line once the voltage reaches 6.0V, and then the polyfuse should blow to protect the overvoltage device.

[Ebraheem Bambotia]

Hi mates I've put the DC/DC converter (3Amper + 10.000uF at output) to 5.05volt to power up Raspberry 3, but seems it's not enough, about 3 to 7 minutes I've this warning on syslog. It looks I've to rise a bit the voltage, which can be the right one to don't exceed? Maybe 5,3volt?
Thank you

I've run a pi 3 on 2 amps and it worked fine. It only used to show me the undervoltage sign when i stressed it out like with multiple chrome tabs open or such....otherwise when it was idling the undervoltage warning never showed. Maybe you've connected too many power heavy devices to your USB ports.

[Barabba]

Thank you mates, interesting topic!
Well, no USB devices on it, raspberry is used for Node Red only, it detects undervoltage every 3-7 minutes, so also when its cpu is close to idle.. that doesn't have sense.. I also installed a capacitor of 10.000uF on DC/DC output to suppress current peaks.. if there may be, and it looks enough stored energy to cover them eventually. About cables I'm using 0,75mm2, just for power the Raspberry I cutted 30cm of the cable with microusb of the power adapter 5V 2.1A, the section of wires isn't big.. but I guess should be enough. Anyway I can test voltage on the gpio interface, since I connected on it a flat cable to power on some I2C devices, that also should not use too much power.

Chances are the DC/DC is not providing enough, maybe I set it when there was no device conected to it and with some load it went down a bit, I need to open the box and check it, since now I couldn't for work reasons, I'll report it here. Slightly I'll set it to 5,2 volt, and check if there may be other faults

[Barabba]

I would report you my experience, I turned it to 5,25 volt, the amount of low voltages was really reduced, I tested it for some hours and I had only one event, to avoid it I turned to 5,32v and now after many days I had no logs about it.
ANyway I decided to enforce the size of the main powerline, I'll use 2x0,75mmq cable and sold it directy to a micro usb male. I'll then reduce a bit to 5,2volt the power and report it here.

[Andyroo]

Do you have anything plugged in (inc keyboard, mouse etc)?
Are you running any programs or is this at the GUI / command prompt?

[Barabba]

Hi, thanks for keping posting here,
No I don't have *any* USB connected to it, but I have some I2C hardware conncted to the GPIO connector, this hardware should not drain much current, especially when no buttons are pressed and not pullup resististors are grounded. Ther are 2 16lines I/0, 2 16libes PWM drivers (connected to 5V relay, powered separetely), one RPC, one ADS1115 ADC, one LCD 20x4. Maybe all is around 50-100mA

[Barabba]

I noticed that with 5,32v input I have 5,07 volt on Gpio.. maybe a bit of voltage drop on cable, but maybe something drops also inside RB on the power management stage

[Burngate]

So you're losing 250 mV.
There's a polyfuse, MF-MSMF250/x, between the input and the rest of the 5v including the GPIO header.
It should have between 15 and 100 mΩ resistance, so to drop that much you would have to be taking 2.5A

[Barabba]

Well, I've seen that in RB1 there is only fuse, but on RB2 there is also a mosfet to check inverse polarity, I couldn't find the scheme of RB3 to understand better.
Well I dedicated a DC/DC (based now on XL4015, instead of other before lm2596) only for RB3 with a condenser of 10.000uF at output, and a cable of 2x0,75mm solded to the microUSB, bette I don't know what to do. Let's see how it reacts, I lowed a bit to 5,2 volt to stress less the components inside, now on the GPIO I have 5,07 volt, better. The microUSB cable I used had for sure some loss, and maybe the 4015 is now keeping more stability on peeks of cpu load.
The 2596 I use now only for relay coil drive (before they were powered with RB, but not on from the GPIO power line! Directly form DC/DC, and with the condenser too).

The reason why I use DC/DC instead of AC/DC is because I'm using a 12V lead battery to keep the power on even if the main power goes off, this is cheaper than UPS and better "clean" from noise.
I'll post results.

PS: to keep more stability I've just figured out is is possible to keep the log files in RAM, instead of using alwasys the SD. If the SD can have some problems (and more you write, more you will have) it may affect stability too. This program writes every hour the logs to SD, and boot automatically:
$ sudo apt -y install git
$ git clone https://github.com/azlux/log2ram.git
$ cd log2ram
$ chmod +x install.sh
$ sudo ./install.sh