Hi together,
I have 2 questions on the voltage of the 5V GPIO pin. I use an official RPi5 power supply.
1) With an RPi5 8GB that is off and has no periphery connected and no SD card inserted, I measure the voltage on the GPIO 5V pin at 5.05V.
When pressing the "on" button, the voltage on the GPIO 5V pin drops to 4.7V almost immediately.
This happens for two different RPi5 (purchased from different vendors) and also with 2 other non-Raspberry power supplies that are rated at >[email protected].
Is this normal and/or what can be done about it? I cannot imagine that any issue comes up with power consumption as it should be extremely limited in the setup.
2) When having a buffer capacitor (for short power-out coverage) connected to the 5V GPIO, the RPi5 power supply does not provide any current after getting plugged into the wall socket, while the other non-Raspberry power supplies do (with a current limited to about 5A as expected based on the specification). Is this due to some overcurrent protection in the official power supply that shuts it down completely instead of limiting the current (which just is so high before the capacitor reaches a certain voltage)?
Thanks for your input!

Last edited by MN_RPI on Wed Nov 26, 2025 7:16 pm, edited 1 time in total.

MN_RPI wrote: ↑

Wed Nov 26, 2025 7:15 pm

I have 2 questions on the voltage of the 5V GPIO pin.

There are no 5 volt GPIO pins on the 2x20-way header of a 'flagship' RPi board.


GPIO pins work at 0 volts and 3.3 volts, and must not be subjected to an applied voltage outside that range. Not all physical pins on the 2x20-way header are GPIO pins. There are also fixed 3.3v and 5v pins, and ground connections, leaving 26 actual GPIO pins.

Physical pins #2 and #4 on the header carry a fixed 5 volt level. They are not connected to the GPIO circuitry on the RPi processor and cannot be software controlled.

Beware of the Leopard

If you don't want measurable voltage drop, buy a high quality lab power supply and connect thick copper or silver wires between the power pins on the 40-pin header. Rated 16A or more for mains AC cables thicker than the GPIO pins themselves and not longer than 10cm should do it. Or calculate yourself. Or sell your RPI5 and buy a computer that has an onboard step-down converter so it can be fed with some higher voltage 9-20V. There are no circuit schematics available for the RPi5, so it is guessing what components cause voltage drop or maybe dirt in/on connectors. Also there is firmware that does things that cannot be seen what it would or can do.

[DELETED]

Last edited by hippy on Sat Nov 29, 2025 10:00 pm, edited 1 time in total.

MN_RPI wrote: ↑

Wed Nov 26, 2025 7:15 pm

2) When having a buffer capacitor (for short power-out coverage) connected to the 5V GPIO, the RPi5 power supply does not provide any current after getting plugged into the wall socket, while the other non-Raspberry power supplies do (with a current limited to about 5A as expected based on the specification). Is this due to some overcurrent protection in the official power supply that shuts it down completely instead of limiting the current (which just is so high before the capacitor reaches a certain voltage)?

I asked the purring programmers here, but kittens are no use when it comes to hardware. Do you have any resistance in series with the capacitor?

Can't see this with a Pi5 here on my scope.
Pi5 with a 27W supply, SSD hat attached. See small ripple on the 5V header pin (header pin 2, measured to GND Pin 6, https://pinout.xyz/), all well above 5V. When switching power button off→on or on→off, then there is no drop below 5V.

redvli wrote: ↑

Thu Nov 27, 2025 10:33 am

buy a high quality lab power supply and connect thick copper or silver wires

Sell your thick silver wires, and get a power supply that has remote sense terminals.

redvli wrote:Also there is firmware that does things that cannot be seen what it would or can do.

Thanks for your feedback so far!
On 1)
I just tried it out of curiosity with a 5.25V / 1A and a 5V / 2.4A rated power supply.
Results are very much the same with a voltage measured between the pins 2 or 4 on the Pin-Header (which in most sources is referred to as "GPIO header" even though those few pins are not related to GPIO as I learned). 5.35V > 4.8V for the 1A and 5.05V > 4.70V for the 2.4A supply.
As I measure off/on the same way, I would assume the wires have no relevant impact as the "off" voltage seems to be totally fine...
Also, I would guess that the official power supply is sufficiently dimensioned regarding wire conductivity/length.
On 2)
I have no resistor added to the capacitor, mainly because I assumed that modern PSUs limit the current to a safe level (e.g. 5A) in such cases and additional resistance would reduce the few seconds the capacitor can keep the Pi running in short powerouts even more.
Maybe it is a feature (short circuit scenario) for the offical RPi supply to cut the power off completely here & the others just limit the current until after 15sec the capacitor is charged enough and the voltage suffices to start the Pi.

pidd wrote: ↑

Fri Nov 28, 2025 6:14 am

And to be petty, the 5V pins can be used as both inputs and outputs on most (if not all?) Pi's.

i think only the pi400 and probably the pi500's dont allow input on the 5v pin

[DELETED]

Last edited by hippy on Sat Nov 29, 2025 10:01 pm, edited 1 time in total.

hippy wrote: ↑

Fri Nov 28, 2025 2:52 pm

ghp wrote: ↑

Thu Nov 27, 2025 4:21 pm

Can't see this with a Pi5 here on my scope.

Does this point to it being a Pi 5 batch issue ?

I don't know. Did you test with a large capacitor wired to 5V on the GPIO header?

[DELETED]

Last edited by hippy on Sat Nov 29, 2025 10:01 pm, edited 1 time in total.

There is no such thing as a "5V GPIO pin" !

The nominal output voltage of a GPIO pin is 3.3Volt!

Voltage drop of the power supply is mainly caused by the
cable between the PSU and the RPI, that is why the oficial PSU outputs 0.1Volt extra to compensate this (5.1 Volt).

The Regulator on the RPI creates the 3.3Volt that feeds the GPIO Drivers. The GPIO drivers are designed to try to keep the output voltage stable, dependant of the output current (15m max) see the schematic of the GPIO driver in the technical documentation.

If you actually MUST have a 5V GPIO, then add a level converter behind the GPIO pin!

This conversation is painful to read :lol:

I think OP is talking about the +5V pin int he GPIO 2x20 pins area. That is, pins 2 and 4 in https://pinout.xyz/

And he/she is wondering because voltage is 4.7V "only" with a good power supply and proper USB wiring, like the original Raspberry Pi power supply.

MN_RPI wrote: ↑

Wed Nov 26, 2025 7:15 pm

When pressing the "on" button, the voltage on the GPIO 5V pin drops to 4.7V almost immediately.
This happens for two different RPi5 (purchased from different vendors) and also with 2 other non-Raspberry power supplies that are rated at >[email protected]

Where did you buy the computers + power-supplies?

I theory you should get your money back maybe, because such a low voltage is I think not according to spec for USB. So USB-stick or SSD via USB3-to-SATA adaptor might become unstable and cause crashes/corruption. Have you measured at USB Vbus?

My feeling is that the goods are 'refurbished' or so. It might be that something is replaced/repaired w.r.t. the path from USB-C conector to 5V,GND pins on the header. We know there are RPi5 with 8GB that is replaced 2GB after factory production. So this could be something similar. It is black friday (already for several weeks(!) ), that means red-flag because there are many fake shops and fake prices as well (prices are higher as nomal but still advertised as big discounts, etc, etc).

Thread cleaned.

Software guy, working in the applications team.

MN_RPI wrote: ↑

Wed Nov 26, 2025 7:15 pm

1) With an RPi5 8GB that is off and has no periphery connected and no SD card inserted, I measure the voltage on the GPIO 5V pin at 5.05V.
When pressing the "on" button, the voltage on the GPIO 5V pin drops to 4.7V almost immediately.
This happens for two different RPi5 (purchased from different vendors) and also with 2 other non-Raspberry power supplies that are rated at >[email protected].

Don't think I've seen it mentioned specifically, but presumably you are using the USB C connector as power input. Does seem a rather high voltage drop to me. Not done a full test (i.e. sufficient to print real numbers), but I think I'm seeing barely any voltage drop on a Pi5 with SSD. What measurement points are you using?

MN_RPI wrote: ↑

Wed Nov 26, 2025 7:15 pm

2) When having a buffer capacitor (for short power-out coverage) connected to the 5V GPIO, the RPi5 power supply does not provide any current after getting plugged into the wall socket, while the other non-Raspberry power supplies do (with a current limited to about 5A as expected based on the specification). Is this due to some overcurrent protection in the official power supply that shuts it down completely instead of limiting the current (which just is so high before the capacitor reaches a certain voltage)?

Sounds perfectly possible. Basically there are two methods of overcurrent protection in power supplies generally:
a) Simple current limit, where the supply switches from its normal 'constant voltage' operation to a 'constant current' mode, during which the output voltage drops. Can cause problems because anything connected to the power rail will likely be operating outside its specified limits, so there are no guarantees what will happen.
b) 'Foldback' current limiting - where if the rated current is exceeded for a short time, the output turns off completely. In some cases you have to turn the supply input off then on again for the PSU to reset; in others the output will be turned on at intervals to see if the problem causing the overcurrent has cleared (often called 'hiccup' mode). [I've actually encountered this problem, where the equipment we were powering had a switch mode power supply that had a big inrush current at startup. We solved it by having an adjustable time where the output was forced on durung a retry, coupled with a constant current limit. The sad thing was that the switchmode IC used in the equipment actually had a pin to limit the inrush current, which the designers had chosen not to use.]