12

I am using Android things on Raspberry Pi 3 B with a 7-inch touch screen. Currently, I always see a yellow thunder bolt sign on the screen which means that the Raspberry Pi isn't receiving a high enough voltage from my 5.3 V 2 A power supply.

According to this and this and many other links, they all suggest that I should increase the voltage supplied to the Raspberry Pi until the yellow bolt disappears.

I will be customizing another power supply manually using voltage dividers, so in order to make sure I do not damage my Raspberry Pi, what is the maximum voltage I can supply to it without damaging it?

5
  • 9
    How do you know it is 5.3 V? Is it a measured value or is it printed on the power supply? Jun 17, 2019 at 6:17
  • 4
    Tip: get yourself a USB power meter. They're very cheap and handy in loads of situations. Jun 18, 2019 at 9:07
  • "I will be customizing another power supply manually using voltage dividers" This is almost sure to result in a very damaged Pi. Voltage dividers (resistor dividers) depend on the amount of current running through them. Aside from needing some very big resistors, if the current draw by the Pi changes, the voltage will change and you may end up with some very large spikes. Dividers are good for measuring voltage with a known current, but if that current is not constant, the voltage will change, a lot.
    – Ron Beyer
    Jun 19, 2019 at 15:05
  • @RonBeyer Let's be positive and assume (s)he was referring to voltage dividers in the feedback loop of the power supply :-)
    – Roland
    Sep 7, 2019 at 20:52
  • @PeterMortensen Exactly. That is the question. My official Pi power adapter is labelled 5.1 V, but with two hobby multimeters in the 100 euro range, I measured 5.25 V and 5.35 V. Accuracy is 1% plus 5 digits, hence +/- 0.1 V. So, without a voltmeter in the 300-500 euro range with 0.01% accuracy, and recently calibrated, which assumes you are a somewhat more professionally skilled electronic technician, don't bother with the exact voltage limits of the Raspberry Pi :-)
    – Roland
    Sep 7, 2019 at 21:01

9 Answers 9

26

No, you should not increase the voltage any further... and at least this answer linked in the question does not suggest to do that.

From Raspberry Pi Power Limitations:

Power sources SHOULD provide 5±0.25V ...

and:

The newer Pi(3/2/B+) have a voltage monitor chip (APX803) which triggers at 4.63±0.07V. The Pi3B+ uses a MxL7704 chip to manage power, which has the same nominal trigger point. [..] The GUI had an rainbow indicator (replaced by a lightning bolt) which comes up in the top right if the voltage is inadequate.

The point is, if there is a 5.3V power supply connected and the lightning bolt still appears two issues are likely:

  • current rating of the power supply is insufficient with respect to the current required by the Pi and accessories such as the display. This usually means that the output voltage of the supply will decrease.
  • poor cables with high resistance, i.e. low cross section and/or lenght, causing a voltage drop.

Replace the current supply and/or cable but do not increase voltage beyond 5.25V.

... and as Milliways' answer got it covered: do not try to use a voltage divider to adjust the output voltage of a power supply.

8
  • 1
    A high risk way to reduce cable resistance (since the USB spec calls for 0.5A only) is to directly feed 5V to the GPIO power headers. However, be warned, this bypasses the polyfuse that protects the pi.
    – Aron
    Jun 17, 2019 at 2:07
  • 7
    A poor cable with high resistance is definitely a possibility. I once encountered a USB cable with a resistance about four times higher than normal (for all four wires), triggering a brown-out detector circuit to reset a device that used about 150 mA. As the current was lower at startup it could start, but it was reset after a few seconds, resulting in a cycle. I suspect they used iron instead of copper for the wires (the cable was much more stiff than normal and the electrical resistivity of (pure) iron is 5.7 times higher than copper). Jun 17, 2019 at 6:44
  • I can confirm this, my rPI 3B was displaying the undervoltage notification even with a tailored usb power supply (Not sure if it was OEM, but was explicitly marketed for raspberry by a decent local retailer), unless I used one specific USB cable. Not sure why, but i had to try several different cables until I found a working one. Jun 17, 2019 at 10:38
  • @Aron Using another MicroUSB cable and power supply would work just as well, and have far less risk. If your power supply only supplies 0.5A, it's not going to work, regardless of whether it is hooked up through the MicroUSB or the GPIO headers. Whether the Pi is hooked up through GPIO or MicroUSB doesn't have anything to do with cable resistance or amperage, what matters is whether you have a bad cable/power supply or not. Jun 17, 2019 at 21:06
  • @Hitechcomputergeek The USB specification does not allow for greater than 500mA current. The micro USB port was never designed for "high loads". Cables than can carry the current required is out of specification, and therefore quite hard to find. If he is handy enough with a potentiometer, then he could directly solder a SMPS (using 22 AWG Wire) to the GPIO header, which is what I did for my NTP server. Plus the micro USB port has terrible contact resistance, hence many SBC use barrel connectors instead, and USB-PD using USB-C.
    – Aron
    Jun 18, 2019 at 1:45
16

The MAXIMUM is 5.25V, although this should NOT be the target. You should not apply more than 5.1V

That is not to say the Pi will be damaged by the higher voltage, because nothing uses 5V - the on-board regulator supplies the voltages used by the Pi. There is a point when the transient protection diode will trigger - causing the poly fuse to blow.

No one can say what will happen to the peripherals attached by the Pi.

Anyone who suggests using a higher voltage is irresponsible! This is just a substitute for a proper power supply and cable.

You CANNOT power a Pi (or any other device) with a voltage divider! This is equivalent to using a very poor high-resistance cable and poorly regulated supply.

1
  • A high resistance cable is probaly what he has already.
    – Jasen
    Jun 17, 2019 at 8:26
7

I will be customizing another power supply manually using voltage dividers

You can't do that. And this suggests that you're a novice with electronics.

Voltage dividers are fine for signal conditioning or creating a relative reference point for something, e.g. an A/D converter. They are not a way of regulating a power supply.

Any modern computer varies its energy consumption based on what it's doing. This is true for RPi as well. If it's doing nothing, it uses less energy than if it's busy calculating and writing to flash memory.

Voltage dividers will not respond well to changes in the load.

3
  • 1
    you can do that by adjusting the internal voltage divider that feeds the feedback node of the power supply. it's not the right solution here though.
    – Jasen
    Jun 17, 2019 at 8:28
  • Although you're right that a voltage divider is a very bad means for regulating a power supply, it is technically speaking possible to use one. You just need to ensure that its impedance is much smaller than the load's – i.e., you need to use very small (but power-capable) resistors, which will then drain much more current than the load you actually indend to feed; this way load fluctuations won't make much of a difference. This is for sure very wasteful, and almost certainly exacerbates the OP's problem. For very small, predictable loads, a voltage divider can be ok though. Jun 17, 2019 at 16:04
  • 2
    ...What is a good idea is to put a proper voltage regulator right next to the Pi, and then supply the whole thing with significantly more than 5V. Pretty sure there are shields available that implement this. Jun 17, 2019 at 16:06
6

Nobody can answer that question as it will be down to the peculiarities of the components in each Pi.

As you know 5V +/- 5% (4.75 to 5.25 V) is the USB spec.

In your case I suspect your power supply claims more than it delivers. Try a better cable or known truthful supply.

That said I have powered higher at 5.8V without causing any known damage. However I had nothing connected to the USB or HDMI ports.

2

Let's take a peek at the Pi 3B schematic. I'll assume you'll use the micro USB input.

Maximum reverse input voltage: 5V

Anything past -5V would probably result in catastrophic failure of the BCM857BS (V_EB absolute maximum rating violation on pins 2 and 1, but then, there are resistors.)

Absolute maximum input voltage without peripherals: 6V (but I wouldn't bet on it)

Looking at the compute module schematics over here, there doesn't seem to be a direct connection between the power supply and the BCM core itself. All voltages are regulated either on-chip or external 3V3/1V8 regulators. The compute module datasheet over here (Table 4) specifies the maximum rating at 6V for the BCM VBAT (the only power rail directly connected to the BCM chip it seems).

Also, a lot of the components listed on the 3B schematic have their absolute maximum rated at 6V (typical for 5V parts). A lot of them (RT9741, RT8088A, PAM2306, APX803) recommend not going above 5.5V.

Compliant HDMI Vout (Pin 18) sets the maximum to 5.3V. Pin 18 is directly* connected to the 5V rail.

The camera and display connectors are 3.3V, so no worries there.

I think it's safe to say that you should stick to 5V, 5.25V tops according to USB specs, as seen from the Pi.


Cool, the technical stuff is out of the way. In your case, sure, you may have a 5.3V power supply, but

  1. Is that the voltage at the connector end itself? Sometimes, the cable isn't factored in. That 5.3V is only at the power brick. The cable drops some voltage.
  2. Is the cable snugly plugged in? It might be loose or not making good electrical contact. Most likely esp. with phone chargers and cheap/non-branded cables.
  3. Are those ratings any good? Is it still 5.3V when you're drawing 2A?

There are test points on the Pi itself on the bottom side. Try checking the voltage between PP1 and PP3 while the Pi is fully loaded.

I think those "increase the power supply voltage" suggestions simply try to compensate any voltage drops in the cable itself and/or on any loose connections. It's quite dangerous in my opinion. What if you're at 9.5V and your loose cable suddenly connects properly?

1

While the other Anwsers are correct as in 'you should not provide any higher voltage then 5.25V', the solution to your initial problem (yellow thunderbolt sign) is to use a power supply which provides a higher current. the official raspberry pi power supply is labeled at 2.5A

And in my experience (running a >100 RasPi research network) a 2A power supply is never really enough, even with no additional hardware attached.

1

The problem of building a proper power supply for the Raspberry has been discussed at lots of places, and the canonical answer is to buy the official Raspberry power adapter.

I just bought one myself, price was not too high, something like 14 euro, and this includes a contribution to some charitable projects. It is rated 5.1 V and 2 A, and has pretty thick copper wires (labelled 18 AWG (0.8 mm2), L = 1.45 m) to the usb plug. It powers the Raspberry including the touchscreen monitor without showing any of those nasty lighting bolt icons.

As for the maximum allowable voltage: I did not yet measure the exact output voltage of this adapter, but it appears to be within the USB spec of 5 V +/- 5%, i.e. 4.75 - 5.25 V DC. It is with 5.1 V slightly higher than the 5.0 V rating of regular usb chargers.

Hence there is no need to try higher voltages than the official usb voltage spec of max 5.25 V.

In fact, your lightning bolts are not shown because the rated voltage is too low, but indicates that the minimum voltage is below some threshold. The voltage may be sagging at moments of higher currents, and those voltage sags are not that easy to measure. You need at least an oscilloscope. Those voltage sags may also be very short, milliseconds to microseconds.

The current rating of the power supply is also not very important, as the actual load, even with monitor, is well below 1 A. One may expect that a higher current rating may help prevent voltage to sag too much, but I have tried several usb chargers rated 2A, 2.5A and 3A, that still cause lightning bolts to show. USB chargers are designed to be cheap, not to power a raspberry pi.

As the official raspberry power adapter is rated 2 or 2.5 A, you could try to power more than one raspberry pi from one adapter. It would be interesting to see if the lightning bolts reappear. If so, you should ignore the current rating, and use it to power only one pi.

0

As many have already mentioned, do NOT use higher voltage. It'll fry your device. Rather try a charger/power supply that can power up to 3amps of current at 5V. You may have a hard time finding those, and even if you find, you'd see that most Chinese adapters are wrongly advertised, hence will not actually provide 3 amps. So unless you have a fast charger (With at least 5V 2.5A or similar wattage rating) lying around, it'd be wise to buy the official RPi power supply.

0

I kept getting low voltage warning with a beefy 3 A universal connector, adjustable voltage radio shack (remember that place?) wall-wart supply I rigged directly to the pins to power stuff including a pi, and save the pi USB port some plug/unplug cycles, while set to 5 volts. It has switchable like 5, 6, 9, 12, 18 V settings or something. This is my "test setup".

The Pi had subpar performance on what seemed like tasks my old Pi 3B+ from 2015-2017 should be capable of.

I was hesitant to try kicking up (all the way!) to 6 V with these tight specs listed here (and on the mfg website), but knew there must be at least 0.5 V drop on my 4-6' USB replacement wire, yet it was reading ~4.95 V open circuit (no load).

Since my former employer wanted my calibrated and extremely accurate Fluke meter back (read: calibrated), my chinesium Harbor Freight Tools meter is maybe +/- 0.5 V on a good day, and even the $100 "premium" HFT meter is +/- 0.1-0.2 V depending on where on the nonlinear curve you are (not accurate or precise and completely uncalibrated), why bother setting up a complex fixture to try and measure voltage at the pins under various CPU load when the chinesium pi jumpers will cut all voltage as soon as you start separating them from the pins, changing the load conditions with an inadvertent forced reboot and invalidating the test, for a $35 SoC. So we continue to speculate about theoretical voltages in this thread, while I wish I had ponied up for a used Fluke instead of that $100 HFT meter. Waste of money.

One day got mad at the Pi, and the project was likely to damage the pi anyway, so kicked it up to 6 V! And it didn't explode. insane! ... it performs much better! Going strong several months now. Learned the RPi definitely has a "throttling" mode where it throttles the CPU, even triggered the low voltage warning a few times on 6 V, but performance is much improved since it's usually not throttling now.

TL;DR: if your Pi is lagging, or you're getting low voltage warnings non-stop, check for CPU throttling and don't fear MOAR VOLTS (within reason) and just go for it until the symptom resolves, so the current supplied by the PSU while actually using it with a RPi load doesn't cause voltage sag with even a decent wall-wart (info for you CSE types) -- these datasheet numbers cited here are academic theory for pencil pushers and were meaningless in practice! $35 RPi on to the next project. Christmas is coming! Get a bench PSU or at least an adjustable wall-wart if you're enough of a tinkerer to have a RPi. Sorry StackExchange, the personal experience was right in this case, and this is based on more than opinion. (BS, MS EE -- surprisingly fun)

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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