I'm designing an embedded system that will be based around a RPi3, combined with the official touch display. In working out how to power the device, I'm getting stuck because the Pi I'm using keeps showing an under-voltage warning when running with the power supply I'm planning on using.

When I connect the official power supply, no under-voltage warnings are shown, using this supply is no option, however, as the device will need to run from a battery. When I power the Pi through a TI LM2596-based power supply (left over from a previous iteration of the project) it shows under-voltage warnings, especially when the display is on full brightness and I'm interacting with it. My multimeter shows a nice 5.0V on the output from the power supply under load, but measuring the 5V gpio pin shows only about 4.65V.

This is what I tried already:

  • I've eliminated voltage drop in the power cable by cutting the official power supply cable and connecting it to my power supply using a mini-fit jr plug. I also connected a mini-fit jr socket to the official power supply itself, so I can still use that for more testing.

  • The LM2596 is rated for 3A, so that should not be an issue

  • The only USB peripherals are a keyboard and a USB ethernet adapter

  • I tried using a separate power source for the display, but this made no difference

I know the official power supply outputs 5.1V, but from what I understand, the Pi should run fine on 5.0V.

Am I missing something? Should I simply suck it up and use an adjustable power supply set to 5.1V?

  • why not just use the working powersupply? – Steve Robillard Mar 18 '17 at 21:50
  • @SteveRobillard I see my question is a bit unclear. I'm going to be running the Pi from a battery, so an AC adapter is not going to work. – Marten Jacobs Mar 19 '17 at 7:02
  • This question contains many assumptions. Measure the voltage on PP1 and PP35, while the Pi is running. Sounds more like a dodgy PSU to me. – Milliways Mar 19 '17 at 10:42

My multimeter shows a nice 5.0V on the output from the power supply under load, but measuring the 5V gpio pin shows only about 4.65V.

The voltage on the output of your power supply doesn't matter. The voltage your RPi receives is what you measure on the 5V pin, and it is indeed too low.

Increasing the PSU output voltage will in turn increase the voltage the RPi gets, but that's not an ideal solution. You should find out why you get such a high voltage drop and where. Typical suspects are the cable and the connector: if the voltage drop is in the cable, get one with thicker and shorter wires. If the problem is in the connector, get a different one or solder the power cable directly to the RPi.

| improve this answer | |

The Pi circuitry is the arbiter of whether it is underpowered or not (and 4.65V is the underpowered threshold).

Perhaps you need to use a variable output version of the LM2596 and feed in 5.3V or so.

| improve this answer | |
  • OK, that is one of the options I have open, but I'm reluctant to go that high. The power supply will probably be powering some other USB devices, and 5.3V is outside of the USB-spec. But isn't it strange that it still shows this error when powering it with a 3A capable 5V power supply? Could it be that my Pi's polyfuse is wonky or something? – Marten Jacobs Mar 19 '17 at 8:33
  • I'm a software person so treat anything I say with a pinch of salt. Yes, I did wonder if the polyfuse had tripped at some stage. That would lead to a slight degradation in the power chain. – joan Mar 19 '17 at 8:35
  • I'm mainly a software person too, so I'm somewhat out of my depth here. I'll get a new RPi3 to test with. The one I'm testing with now has been used as a OSMC before, and I know it was actually underpowered back then (5V, 1.5A PSU), so that may have broken something I guess... – Marten Jacobs Mar 19 '17 at 8:40
  • @MartenJacobs I'm assuming you have a heatsink on the LM2596. If you don't that might well account for what you are seeing. – joan Mar 19 '17 at 8:50
  • Yes I do, the PSU is actually part of an earlier iteration of the design, and it's an SMD version of the LM2596 (TO-263). It's soldered to the board and vias connect it to the ground plane on the underside of the PCB, as per the recommendations from the datasheet. – Marten Jacobs Mar 19 '17 at 8:56

This could be current problem, not voltage problem. Check the amps too under the load.

| improve this answer | |
  • Current and voltage vary together under load. If there is a sudden demand for current (as is apt to happen with a running Pi), the supply may drop voltage to provide it, which is why the undervoltage alarm flickers on and off. – goldilocks May 15 '19 at 12:41
  • it depends on the "hardness" of the power source. Try the car battery (which is extremely hard source) and it will give you 0.01~60A at stable 12V. – andrej May 15 '19 at 17:44
  • Just to add: having stable voltage even under load, is a quality issue of a good power supply. – Ingo May 15 '19 at 17:48
  • I didn't mean that undervoltage is inevitable no matter what you use...the root cause is a particular supply can't deliver. I just don't think checking the amperage is very meaningful because if it is "underamped" it is almost certainly also undervoltage at that point and the Pi will simply fail. Further, deciding what the amperage should be is guesswork. – goldilocks May 15 '19 at 18:25

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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