under voltage detection - Raspberry Pi 3 B+

RE: under voltage detection - Raspberry Pi 3 B+

clem57 — Thu, 17 Oct 2019 15:48:00 GMT

I used my official Pi 4 5 V 3.0 A charger on the BB-AI with no troubles. But I have not stressed out the board, since I am waiting on a fan to cool things down first.

Clem

RE: under voltage detection - Raspberry Pi 3 B+

shabaz — Wed, 16 Oct 2019 20:53:18 GMT

Hi Roger,

You're right, the Pi 4 charger is 3A, but labelled 5.1V, so still a nice way to quickly identify if people are using phone chargers, or the official chargers.

I've been meaning to purchase a few of them, now that I have a couple of Pi 4's, and the BeagleBone AI use the same connector.

I tried powering the BB-AI from a fairly new Samsung charger but that failed miserably. The older Pi 3 5.1V 2.5A charger is working fine for powering the BB-AI for now (using a micro USB to USB-C adapter) but I've not pushed it yet, so want the newer Pi 4 supply for that.

RE: under voltage detection - Raspberry Pi 3 B+

glyndon — Wed, 16 Oct 2019 11:23:29 GMT

I can supply a data point, as an example.

One of my 3b+'s has been notoriously flashing undervolt alerts. So, I built, using 16ga stranded copper and new connectors, a 1m power cable, supplied voltage to it using a very expensive 5V supply rated at 400W output (it's normally intended for use in specialized equipment, e.g. medical), set its self-monitored output to 5.25v.

And the blasted thing still flashed undervolt at a couple points during bootup.

In fact, I did this for 3 more units, with similar results.

My background across 40+ years includes a lot of electronic and electrical engineering study, so I am familiar with how current and power flows work, and have heretofore been figuring - as have others here - that the various small PSU's in use cannot handle the spiky draw by the Pi. So, I had been intending to put a 'scope on some of mine to see if that was the case.

I've also considered beefing up the supply cable with some load-end buffer caps to help handle load spikes.

But it seems like such drastic measures shouldn't be necessary for a device that was designed to be operated under less-than-ideal conditions (i.e. a classroom) by less than degreed engineers. Doesn't it?
It's starting to sound more like there may be an actual flaw in some range of the devices, something dirt-simple like a bad batch of voltage regulators or zeners, that's frustrating a lot of otherwise adequate use-cases.

RE: under voltage detection - Raspberry Pi 3 B+

rew — Wed, 16 Oct 2019 09:14:48 GMT

alinp wrote:

Current limit was set to 2.2A.

That's not enough. The pi will draw peaks that are larger than that. They may be very short. To catch such things, put the scope in "normal" or "single" mode and set it to trigger at 4.8V. Put the timescale to something much faster than 100ms/div.

quoting shabaz and cant find the "quote" button....

The original post refers to a 5V 3A rated supply. If that really is correct, then it's not an official PSU, which is marked 5.1V 2.5A, so all bets are off if that is a measurably good PSU and cable. The original poster observed 0.4V drop so it's not surprising if an undervoltage alert occurs.

IIRC, the pi4 comes with a slightly beefier power supply. 3A sounds right to me.

RE: under voltage detection - Raspberry Pi 3 B+

dougw — Wed, 16 Oct 2019 06:13:42 GMT

Check the voltages and currents measured in this blog ... the voltage does dip enough to be worrisome with some supplies and cables.

RE: under voltage detection - Raspberry Pi 3 B+

fmilburn — Wed, 16 Oct 2019 05:29:28 GMT

To reinforce the post from Shabaz: I have a Pi 2, two Pi 3s, two Pi 3B+, and two Pi 4s. Like Shabaz I have never seen a problem with a good PSU and cable on a Pi by itself and I have loaded them up 100% for extended periods. Using a phone charger or cheap cable is the best way to see the dreaded underpower icon. Powering a screen or other add-ons with a Pi 3/3B+ off of the power pins might also cause under voltage but the Pi alone should be OK.

RE: under voltage detection - Raspberry Pi 3 B+

shabaz — Wed, 16 Oct 2019 04:44:50 GMT

glyndon wrote:

The one the original post was relating: That despite a measurably-good PS and cable

I'm not sure anyone has demonstrated that. As a data-point, for what it's worth, I've probably gone through a dozen official PSUs with attached cable, across different full-size Pi's (the first model, Pi 2B, Pi 2 B+, Pi 3 and 3B+, and now the 4, expect that's got a different connector) and not seen any power related issue (with the official charger+attached cable). The times I have seen a power issue, it's been resolvable by using the correct PSU or sometimes by using a thicker cable.

I believe some users assume that mobile phone chargers make good power supplies for the Pi, but that's measurably not true in all cases - only sometimes. It can be demonstrated by measuring with a voltmeter, under load. The micro USB connector looks like it is designed for accepting a mobile phone charger, and there's an implicit assumption that it should work. That's not the user's fault, but nevertheless that's a common issue. A thinner cable or poor quality charger still often charges a mobile phone, but that's not the case with the Pi or other SBCs.

Mobile phone chargers in conjunction with a typical charging cable, sometimes cannot supply close to 5.0V at the micro USB end when under load. The Pi's official PSU has a slightly higher output voltage to compensate for voltage drop across the cable. It's not a great implementation (there were likely non-technical reasons for that decision - i.e. cost, and the hope the standard connector could allow some mobile chargers to work under lighter loads). The Pi's official PSU makes for a terrible mobile phone charger, but is a good PSU for the Pi.

The original post refers to a 5V 3A rated supply. If that really is correct, then it's not an official PSU, which is marked 5.1V 2.5A, so all bets are off if that is a measurably good PSU and cable. The original poster observed 0.4V drop so it's not surprising if an undervoltage alert occurs.

RE: under voltage detection - Raspberry Pi 3 B+

glyndon — Wed, 16 Oct 2019 02:56:40 GMT

The one the original post was relating: That despite a measurably-good PS and cable, some models of Pi's seem to complain about undervolt events.

I'm just one of (what seems like) many who register the same complaint.

RE: under voltage detection - Raspberry Pi 3 B+

clem57 — Wed, 16 Oct 2019 01:19:06 GMT

Powering the GPIO pins are possible too. But be sure to regulate that source carefully or the board will fry.

RE: under voltage detection - Raspberry Pi 3 B+

shabaz — Wed, 16 Oct 2019 00:55:08 GMT

No; that's a dual-transistor device assisting in implementing an ideal diode; if anything, not having it would further reduce any (negligible) voltage drop at that point. It's a red herring.

3B and 3B+ are solved problems, the official PSU manages to operate them. Rather than trying to supply power to different points on the board, it is easier to swap out the PSU and/or cable.

However, if you're integrating the Pi into another project, then I can see the need to perhaps supply power from elsewhere, e.g. via the 40-way connector.

What is the problem you're trying to solve?

RE: under voltage detection - Raspberry Pi 3 B+

glyndon — Wed, 16 Oct 2019 00:45:22 GMT

A notable difference between boards, unless they just omitted showing it on one schematic, is that the 3b+ does not have the current-mirror device (BCM857BS) between the power socket and the +5 bus on the board. Other recent board models have this device, but the 3B+ doesn't.

Could that be making it more susceptible to relatively minor variations from the supply?

RE: under voltage detection - Raspberry Pi 3 B+

shabaz — Wed, 16 Oct 2019 00:40:32 GMT

I'm not sure the micro USB socket could be an issue, unless it's been used thousands of times, or physically damaged or pulled off the board.

If you wish to power the Pi from elsewhere, then the 40-pin header can be used*, but there is less protection (e.g. no polyfuse). You could try soldering to some test pads close to the micro USB connector (I have done that in the past) but it is very easy to rip off the pads. It's much easier to connect to the 40-pin header (or solder to the underside of the 40-pin header if you wish to keep the pins free for plugging on other boards in future.

Example, showing wires soldered to the underside of the header, and then epoxy-glued:

* Another option is power over ethernet (PoE) on certain supported Pi's if a PoE board is fitted on the Pi.

RE: under voltage detection - Raspberry Pi 3 B+

colporteur — Wed, 16 Oct 2019 00:28:56 GMT

I have experienced the same issue you are describing. Ironically, I am using 3amp P.S. as well. At least that is what the label suggests. The problem went away when I switched to another 3amp P.S. I had seen the problem randomly over the past few months. It appears throwing a bad P.S. into a box with other P.S. doesn't resolve the issue.

I'm not saying this is your problem but it definitely was mine.

Sean

RE: under voltage detection - Raspberry Pi 3 B+

glyndon — Wed, 16 Oct 2019 00:23:04 GMT

With all the talk about cabling being the barrier to proper voltages within the Pi, why hasn't anyone mentioned that the micro-USB socket on the Pi might be introducing a bit of resistance itself?

(presuming the PS itself is sound and has the needed capacity)

Seems like a few people have had success by using a proper (i.e. one that's designed to have low resistance) power socket connected directly to the proper pins on the board.

Are there specific recommended points (solder, or pins) at which such a socket should be attached?

RE: under voltage detection - Raspberry Pi 3 B+

alinp — Wed, 15 May 2019 07:00:56 GMT

As stated in my original issue

I've measured the voltage right at the USB input socket and it's 5.0V (so the cable drops 0.1V).

The measuring procedure was done using a MSOX3104TMSOX3104T oscilloscope with a passive 10:1 probe hooked up right to the power rail smoothing capacitor C84.
*schematic from the official Rasberry Pi docs.

Measuring on the oscilloscope (MSOX3104TMSOX3104T) was done using Roll mode, at a sampling rate of 100ms/division.
At no point during the testing procedure, which involved throwing multiple power hungry tasks at the CPU, did the power supply (which again was a R&S HMC8042 OS) vary by more than 100mV. Overall, the voltage did no drop under 4.9V. Despite this, the low-voltage warnings kept coming.

The power was fed trough the USB port, using a high quality USB cable, that was cut to 5cm in order to expose the power and GND wires inside an hook them up to the bench PS.

Given above testing procedure I can confidently exclude any factor related to the USB cable or power supply being used. Thus, I conclude that there must be a faulty hardware design somewhere down the line. If someone has identified it or has a different opinion please let me know.

I still don't know if I should send in back to Farnell under warranty due to this malfunction.

RE: under voltage detection - Raspberry Pi 3 B+

luislabmo — Tue, 14 May 2019 18:54:01 GMT

The voltage drop may be caused by the USB cable used: length, cable thickness, connector quality (or specs) may be the most important factors to consider: see this comment and this one. In my Raspberry Pi Road Test I cover this in the Power Consumption section.

The easiest way to discard a USB power problem with the Pi is to power it directly to the GPIO with a regulated power supply where you can reduce some risk factors by increasing the wire thickness, using connectors with a good current rating, etc. USB power issues are harder to troubleshoot as USB cables are sealed most of them without reliable specs available that will bring a lot of unknown factors -one of the main reasons why is good to have the official micro USB power supply for Raspberry Pimicro USB power supply for Raspberry Pi to discard USB power issues.

Luis

RE: under voltage detection - Raspberry Pi 3 B+

alinp — Tue, 14 May 2019 15:33:42 GMT

I'm facing the same problem.

At first I thought my switching PS was the culprit, but then I've switched to powering the Raspberry Pi 3B+ from a Rohde&Schwarz HMC8042 bench power supply which is more than capable of driving the Pi.
Voltage was set to 5.1V.
Current limit was set to 2.2A.

According to the bench PS which shows real-time current consumption, the average (in my case, for the firmware I'm running (it's a RetroPie) it's around 600 mA. Even with this beefy bench power supply I still get the under-voltage warning. I've measured the voltage right at the USB input socket and it's 5.0V (so the cable drops 0.1V).
If someone has an explanation for this, I'll be glad to hear it.
I'm assuming it's either a bad LDO or a hardware design issue.

RE: under voltage detection - Raspberry Pi 3 B+

genebren — Sat, 29 Dec 2018 19:15:59 GMT

Doug,

This is a great cable test setup. This should help all those people out there that are having under voltage errors on their Pi devices. I have switched all of my USB cables over to ANKER PowerLine products and I have not had any power delivery issues (one of my products runs >2.0 Amps through the USB cable).

Thanks!

Gene

RE: under voltage detection - Raspberry Pi 3 B+

dougw — Sat, 29 Dec 2018 17:05:52 GMT

The USB cable can easily drop the voltage by over one volt. You can see some real tests here:

E36313A Triple Output DC P.S. - Review

Some cables drop the voltage by over 3 volts....

www.youtube.com/watch

RE: under voltage detection - Raspberry Pi 3 B+

e14 Contributor — Thu, 27 Dec 2018 18:56:41 GMT

My issue was a USB cable .it's working fine with an OEM Samsung non fast charge box and cable. 5v 2a

RE: under voltage detection - Raspberry Pi 3 B+

jomoenginer — Fri, 06 Jul 2018 21:07:17 GMT

This issue seems to have shown up on the Raspberry Pi forum where the solution was to either run the following command or to load a new SD card with the latest Raspbian image.

Try this if you have not performed sudo apt-get update:

sudo apt-get update && sudo apt-get dist-upgrade -y

If this does not solve the issue, then go the path of loading at minimum 2018-03-13 Raspbian Stretch.

https://www.raspberrypi.org/forums/viewtopic.php?t=207942

RE: under voltage detection - Raspberry Pi 3 B+

e14 Contributor — Fri, 22 Jun 2018 15:54:33 GMT

I have the same issue under-voltage deteched.

why not check the cable itself.

my cable takes 0.06v/100mA

RE: under voltage detection - Raspberry Pi 3 B+

keillrandor — Sun, 17 Jun 2018 09:30:06 GMT

I have a PI 3B +, had similar under voltage warnings, both in Debian Stretch syslog and yellow flash icon on my connected official 7" screen (I don't use any GPIO voltage bridging). I use an iFi power 2.5A rated switching PSU with micro usb connected to the PI. I also fed the screen with another separate micro usb power source. With this setup I had voltage underruns every now and then during the boot sequence.

PI<----angled micro USB cable/USB-A male 0.5m---> <----micro iFi USB female A adapter/DC pin cable 0.1m---> <--DC PIN cable to iFi PSU 1.5m? ---> wall DC 230V/50Hz

I resorted to solder the angled micro USB connector to the DC-pin adapter, removing about 0.5 meter of cabling and a possibly weak USB A female adapter used above.

It resolved the issue. It would have been interesting to use an oscilloscope, which I don't have, to monitor the improved voltage stability.

RE: under voltage detection - Raspberry Pi 3 B+

y2010x — Wed, 13 Jun 2018 13:53:48 GMT

Thank you all very much for your suggestions.

I tried them all, but the yellow lighting bolt still showed up, and the voltage drop for most cases is about -0.4 volt. The unit is functional (not trying to reboot) in one of the power supply with the yellow thunderbolt; I ordered 3 more units to see if this is a bad unit - I will let you know the results when I have a chance to test them.

Thank you all again very much.

RE: under voltage detection - Raspberry Pi 3 B+

fmilburn — Sun, 03 Jun 2018 00:38:37 GMT

I agree on the USB cables - I have had problems with them in past and I think they are sometimes overlooked. Avoid the long cheap ones in particular. There is an informative video on the topic here: https://m.youtube.com/watch?v=n70N_sBYepQ&t=14s