Watt is correct?

Peso

Member
Hello. My first time in the Hummy Arms. I only came in to use the toilet, but while I'm here I have an electrical question. Who better to answer it than men with pints.

I've got this Raspberry Pi 4B I'm playing with. Stock PSU 5v, 3amp, 15w. I have an SSD, a 2.5" HDD and a USB stick plugged into it. I've had a few crashes and one path I was wondering is is the power draw too great.
I have one of those WiFi plugs for turning things on and off which also measures power usage. So plugged the Pi PSU into that. It tops out at 10w at boot.
Then it occurred to me, the measured 10w are for the 230v AC coming from the socket, not the 5v leaving the DC PSU. Does anyone one know to calculate the DC wattage please?
 
Does anyone one know to calculate the DC wattage please?
You can estimate the DC watts if you know the efficiency of the PSU, but really what you have to do is measure the DC current. There are in-line USB current meters available, but that would add another unknown because of the series resistance added. You might not need to do that though, if you can put a 'scope on the DC voltage and look for brown-outs. Meters are not much use for this (including your smart-plug power monitoring), they don't react fast enough.

Could you not try a process of elimination? Eg, disconnect the HDD and see whether the RPi stops crashing.
 
If you've got 10W in and less than 10W out, then where does the extra power go? The only place is heat in to the PSU itself, so assuming it's not a toaster then it's going to be roughly 10W out.
Application of the Mark I temperature testing device (your hand) to it should tell you something.
Obvious, innit?
Now about that pint...
 
If you've got 10W in
Yes, but how accurate is that?

and less than 10W out, then where does the extra power go?
Thermodynamics. No way will 10W in convert to 10W out.

But that's largely irrelevant. My point is that measurements are taking average readings over some period of sampling. What matters here is instantaneous peak demands. There are no big capacitors on the RPi nor in a standard USB power wart to cover peaks (unlike in the systems I used to build professionally, nor in a proper PSU).
 
Yes, but how accurate is that?
Who cares about the exactness of the absolute value? It's largely irrelevant.
Thermodynamics. No way will 10W in convert to 10W out.
I never said it would, but modern SMPSUs get pretty close. The rest goes as heat, as I said. So if it ain't warm, then power out is almost power in.
What matters here is instantaneous peak demands.
And what's going to draw those peaks? Nothing that I can see.
 
Last edited:
Thanks everyone. The Pi is at my parents, used, or trying to use it as a NAS. Currently not booting after its last crash. I can see it drawing power from the WiFi power switch when turned on, but it never appears on their router (LAN connection).
Will be there next week to build it again. Connect to it by VNC and I keep getting the desktop suddenly changing to a really basic one, taskbar disappears. If the file manager was still open USB drives disappear, if terminal was open a restart or shutdown command "can't be found". All I can do is pull the power, which is probably why it now doesn't boot!!

I think I'll try a process of elimination as Blackhole suggests, just in reverse. I'll splurge on a powered hub to remove any question of not enough power. Was hoping to avoid it so less cables and sockets used at my parents router. It'll also get fresh install of Pi OS, fingers crossed that's that.
When it worked it was brilliant and I felt really clever using a Pi, despite no one in many family knowing what one is, or be being remotely impressed or interested.
 
It’s a low cost 64gb SSD in a USB 3.0 caddy. Can't remember the brand, a yellow lion I believe. I'm nothing if not cheap.
The HDD is a 2TB Seagate 2.5, 2 years old at most. The HDD was plugged into my parents router first, but would become unavailable. Still present in the router settings. Restarting the router didn't help, would have to re-plug it. Moved the HDD for 3 months to my synology NAS and it was always present and responded when accessed. Ran a fixdisk scan which it passed. So assumed dodgy router firmware. Hence the foray into a Raspberry Pi, I wanted something more reliable.
 
The units are "GB".
"b" is bits, "g" is undefined in this context.
The HDD was plugged into my parents router first, but would become unavailable. So assumed dodgy router firmware.
More likely lack of current capabililty.
Moved the HDD for 3 months to my synology NAS and it was always present and responded when accessed.
With a proper power supply. Kind of proves it don't you think?
Most drives have a label on which states the voltage and current requirements, but I guess you declined to use the information provided and just ploughed on with plug'n'hope and assumptions.
I'm nothing if not cheap.
Buy cheap, buy twice.
Hence the foray into a Raspberry Pi, I wanted something more reliable.
They are (at least mine are), but you have to understand their limitations. You can't just go on plugging stuff in relentlessly until every connector is used.
 
Agreed. Power management is (IMO) the biggest issue with RPi set-ups (along with heat management). They've been developed way beyond their original intention, and now trying to power them from USB is getting silly.

Consider that 10W demand. Unless there's PD going on (negotiated higher voltage supply), @ 5V that's 2A continuous average, but peak? Who knows. For a safety margin I would allow double. Even 2A is stressing the cable and the connector contacts, and producing voltage drops along the way. Current peaks create voltage droops (brown-outs) even if the PSU (at the other end of a long wire) is able to supply the full current, which wouldn't be so bad if there was a big capacitor at the RPi end to smooth them out. But there isn't.
 
Last edited:
... now trying to power them from USB is getting silly.
Oh, I don't know, the laptop I'm typing this on came with a 180W USB-C power supply, and the only way to charge the battery and otherwise power the laptop (and GPU when fitted) is through USB-C. They currently also offer a 240W power supply (the maximum USB-C is currently specified to).

But yes, your point is taken, there are distinct limitations in how much power a RPi can supply to its peripherals. Time for Peso to get a powered hub to plug the drives into.
 
Many thanks everyone.
I think prpr responce to "Moved the HDD for 3 months to my synology NAS and it was always present and responded when accessed" of "With a proper power supply. Kind of proves it don't you think?" sums it up.
3amp 4 port hub ordered. Will be setting it up next week.
 
Update
Almost three weeks with the Pi powering itself and an SSD with its OS, while a 3amp hub powers the 2TB 2.5“ HDD.
Pleased to say I’ve had no problems.
 
Back
Top