Ran a test on my Pi 3B+, today.
Keyboard, Mouse, HDMI & Ethernet.
Official Pi Plug Pack, measured 5.22V. All ok.
Plug Pack rated at 5V 2.1A, measured 4.79V. Icon triggered.
So it triggers slightly higher than the 4.63V +/- 5%.
Agree with it being because of school situation and beginners.
I think the Official Plug Pack is a better switch mode than my ordinary plug pack.
Cheers
Jim
Hi Jim,
Good to have some real-world testing to check the theory!
We do find the failure rate of the official power supplies to very low, nothing is perfect after all, but for power supplies they seem to be exceptionally robust.
Hi Guys
That is one particular unit. The “real world” could be a bit different with other devices.
4.79V is within 5% of 4.63V so it is within quoted figures. 4.63V plus 5% is 4.86V.
What I am getting at is the difference between 5.1V and 5V should not be a concern. As I stated above some of this could well be meter error. I believe a piece of kit such as RPi or Arduino should be reliable over a spread of voltage as much as 10%. that is +/- 5% or +3% to -7% or something of that order. The manufacturer may publish something like this I haven’t had a close look to be sure. I have not had any problems with this sort of thing myself as I tend to overestimate the loads anyway. As for actual points where a device becomes unstable or unusable I doubt that a manufacturer could publish figures or would want to as each individual circumstance would be different.
I don’t think some contributors pay enough attention to the power supply in use regarding current supply capabilities before they start. I have seen instances where a contributor has measured V at the power supply and measured something less at the device. Advice has been to increase V at power supply to compensate for the loss. I think this is quite wrong and one should find out where and why the V is dropping and do something about that. Granted there are some cases I have seen where the voltage at the load is critical and when supplied from a regulated supply the voltage sensing is actually done at the LOAD to compensate for any drop in the wiring, but I don’t think this situation is applicable to this forum topics.
Personally I tend to over do this a bit. I am currently mucking about with some XMAS lights and so far have 2 strings of 150 Neopixels with an Arduino Pro Mini as the clever bit. Brightness currently at 100 (out of 255) and max current so far 2.5A. To power this I purchased a MeanWell supply of 5V @ 14A. That just about should cover anything more I need to throw at it. May be an overkill but it is something I should not have to worry too much about. The main worry is inrush current possibly tripping circuit breakers but so far so good. I might have to measure this and if a concern do the switching at the 5V level. Another story later.
Cheers Bob
Hey Bob,
My understanding of the reason they target 5.1v nominal instead of 5v is to give themselves a little extra voltage to compensate for voltage droop under heavy load, while still being USB compliant.
Actually, I just took a bit of a dive into the USB specs. USB Type C voltage requirements are defined in the USB 2.0 specification.
Interestingly, there was an ECN (Engineering Change Notice) issued in August 2014 which increased the maximum allowable USB voltage from 5.25V to 5.50V - and they’ve even stated cvompensating for voltage droop as the reason:
I’d be willing to put money on the 5.1V specification being a direct result of that ECN. The original USB spec was 5.0 ±0.25V, but now that it’s 5.0V (-0.25V +0.50V), it’s equivalent to 5.125V ±0.375V - hence the 5.1V of the official Pi supplies.
Percentage wise, that’s 5.0V +10%, -5% or 5.125V ±7.32%
All this is to say that a 5.0V ± 0.25V supply capable of 3A isn’t strictly USB spec compliant and hasn’t been since 2014 - though if the cable and connectors are low enough impedance it’d work fine.
This post says the voltage input requirement of the on-board supply is 4.00V to 5.50V - which is actually the same as the USB 3.2 spec:
Ah, but don’t forget to account for the impedance of the connector and traces on the board! If you’re pulling 1A and measuring 4.79V at the supply, as little as 90 milliohms would be enough to trip the Power warning on a 3B+.
FYI, here’s the datasheet for the voltage monitoring chip on the Pi 3 boards:
https://www.mouser.com/datasheet/2/115/APX803-82466.pdf
And here’s the one for the Pi 4s:
Hi Oliver
All very interesting.
What I was really referring to was the publication or lack publication of the reliable operating voltage range of the devices themselves, RPi, Arduino etc. Or a % referred to a nominal 5V.
Cheers Bob
Thanks for all the excellent research you have done @Oliver. Provides a good light on what the Raspberry Pi corporation is up against in powering the Pi via a USB connection; considering the requirements of version Pi 4. And has refreshed my memory from a few years ago.
The voltage measurements were made at the 40 pin GPIO connector. The easiest place to get the board voltage.
I think this thread has gone way beyond the original intent.
But hopefully it may help other users of Pi devices understand the need for a quality robust USB power source.
Cheers
Jim
Great deep dive there Ollie!
And some qualitative feedback after distributing thousands of RPi’s; I can confirm we get our fair share of “undervoltage warnings” that miraculously go away when the official RPi 5.1V Power Supply is used. It’s one of the most common “faults” reported, and no surprise given how often it appears online in other communities.
The overall design (electrically, physically) of the RPI PSU is better suited for power delivery compared to chargers and other 5V options that most people have within arms reach. It will ensure RPi will function fine out of the box, and with as many peripheries as you like, right up to its rated specs without the disappointment of unexpected power issues/warnings.
Such a comprehensive discussion - thanks to everyone for sharing their perspectives!
@Robert93820 I get where you’re coming from, but I think the short of it is, it’s easiest/best to just use the official supply with a device whenever possible. If you want/need to use another power source, assume the official supply specifications are the same as the device requirements (5.1V ± 5% ie. 4.85V - 5.36V for a Pi).
That said, (more for others reading) you do need to watch out for Fast charging protocols. Even though your Samsung Adaptive fast charger says 9.0V @ 2.3A you can’t just stick 9V into your phone - it’ll blow it up! It’s expecting 5V first before the phone and charger ‘negotiate’ jumping up to the higher voltage. There are many different proprietary fast-charging protocols - check out this video - but USB ports/supplies implementing these will still be backward compatible with older USB devices, so they shouldn’t damage a normal USB device if plugged in.
If a device doesn’t have an official supply, but it’s got a USB port, the power supply requirements are defined in the relevant USB spec(s).
If it’s powered another way, the requirements should be specified explicitly (eg. the Uno R3 is specified to accept 7V-15V via V_in).
The problem with USB is for a “Universal” serial bus, it’s far from Universal these days. There are so many different specifications, revisions and ECNs it’s hard to keep track. USB 2.0, USB 3.0, USB 3.1 (each different pre and post 2014), USB 3.2, USB Type-C, USB Power Delivery 2.0 and 3.1, and USB Battery Charging 1.0, 1.1, and 1.2, as well as the deprecated USB 1.0 and 1.1 all have similar, but slightly different, voltage requirements. The exception is USB 4 (essentially Thunderbolt 3), which doesn’t specify voltages - it leaves that up to one of the other specs, depending on the implementation (I’m not sure if this is more or less confusing).
Of course, this all means that while most USB power sources and power sinks will work fine, there are edge cases (like the Pi’s) where it’s really important not just which USB spec it’s compliant with, but also when it was compliant with that spec.
It also depends on what you call/where you measure the input voltage! Here’s the worst-case topology for a low-power USB 2.0 device (100ma) from the spec:
And here’s the one for USB 3.0 and USB 3.1 which applies for both low and high power devices. :
They swapped it for an equivalent resistance diagram in USB 3.2:
Matching up the correct USB supply and USB device is a hell of a task (and the historical specs are a pain to track down as the USB IF doesn’t publish them on their site), so honestly, leaving that job up to the engineers that designed the parts to work together (by just buying an official Pi supply to use with a Pi) is a very wise move and for the $15 can save a lot of headaches - particularly since some early versions (Rev 1.1) of the Pi 4 weren’t USB compliant! This issue was fixed in hardware Rev 1.2 (released back in Feb 2020).
@James46717 It’s also worth noting your meter is most likely showing RMS voltage (averaged over a period of several seconds), whereas the chip on the Pi 3 will respond to a voltage dip within 20 µs. Because the processor on the Pi is running at 1.4 GHz, the power spikes/voltage dips can be far faster than what a DMM can measure. For monitoring supply voltage on an embedded device like a microcontroller or SoC (even one operating in the high MHz range) you really need a high-speed scope or a specialist measurement device like the Qoitech Otii Arc to catch the transient behaviour (it always seems to be the transients that ruin an otherwise good design).
FYI, the USB spec does allow for transient voltage dips, so devices can experience voltages below what’s specified above. At 4.79V DC I’d be quite surprised if it wasn’t dipping below the threshold momentarily - especially when the official Pi Supply bottoms out at 4.845V, in the absolute worst case.
@Graham I got a bit carried away last night. I actually ended up updating some of the Wikipedia pages on USB since they were inaccurate! 
Hi Oliver
Methinks it is all getting too hard.
Thanks for that great historical insight into what has been happening. It has been 20 years or a bit more since I worked full time and to be honest I just haven’t kept up. Somehow I am not unhappy about that as I think if I had been still working I would have been pulling my hair out in a padded room somewhere.
It is all a bit strange after working in situations involving military and other customers requiring high reliability and MTBF numbers where there are very few if any grey areas. Something is right or wrong, not many “in betweens”. From your last couple of informative posts it seems that occasionally (or frequently) someone wanders off on a tangent and gets a little bit away from the recognised “norm”. After a while this is likely to become some form of “standard”. We then finish up having lots of these “standards” and going in lots of little circles.
Don’t get me wrong here as lots of good things come out of people going off on tangents. Think of Sony and Phillips who gave up waiting for the industry to make decisions and went off and did their own thing and we now have a standard called SPDIF for digital audio transmission over cable and fibre. But your reference to universal not being very universal is very true and can be quite confusing to those (like me) who have not been involved enough to keep right up with trends.
Thanks for a couple of enlightening posts.
Cheers Bob
A little while back I commented …
If anyone is interested, I have posted a suggestion in the “official” Raspberry Pi forum that RPT consider adding a project oriented power supply to their product range.
It’s a good discussion going on over there. My first thought - at least at the development stage - would be to use an old ATX PSU. Sparkfun actually do a few DIY power supply kits like this:
I know the guy from DIY Perks is a big fan of recycling old server power supplies in his projects, although these are much more complex (and potentially dangerous - depending on the design) to implement. Also, this is not directly useful for you but may be of interest, Raspberry Pi has recently released a Lego Mindstorms Build HAT along with a dedicated 48W PSU for it (6.0A @ 8.0V):
Is this the official power supply for the Raspberry 5 too?
NO.
This is for a Pi 4, but it can be used with a Pi 5.
Pi 5 official.
regards
Jim
Hi. I purchased the official Raspberry Pi 4 PSU from you some years ago now. My question is whether you perhaps have a solution to the following problem. I have a UPS (yes, this isn’t a typo) to which I currently have a monitor for the Raspberry Pi 4 and an extension cord that connects the RPi4 PSU to the UPS (yes, this is strange). This problem has arisen because of the odd shape of the RPi4 PSU; I can’t simply plug both the PSU and the monitor cable directly into my UPS, which has only two standard 3-pin slots that are close together. I have searched the web for a product that would provide a better solution, such as an alternative PSU without the large head, but have found none. Any ideas, please? Cheers, Glenn
Hi Glenn.
Have a look at a “sideways” double adaptor. They can plug into a standard double outlet and still enough space to plug something into the other socket alongside. This gives you effectively 3 sockets in a line so you can plug your larger device into the one furthest away.
Cheers Bob
Hi, Bob. Many thanks. Yes, that could work. I don’t have the right sideways double adapter at the moment but will get one soon.
Cheers, Glenn
Hi Glenn
Readily available at anywhere they sell electrical bits. They are available both ways too so work out which way is best for you and go from there.
Cheers Bob
Hi, Bob. Yes, I had to order a “left” one. Thanks again.
Cheers, Glenn
And you can get our latest projects and tips straight away by following us on:
