gregeric
Posts: 1509
Joined: Mon Nov 28, 2011 10:08 am

Re: Current Consumption Figures

Thu May 12, 2016 6:41 am

The core regulator on the Pi3, a RT8088AWSC, has max 3A output, so about 4W @ tops at 1.4V . Allowing for conversion losses, that's about 1A tops on the 5V rail from compute usage alone, with all guns blazing. Add on ~300mA for network etc on the 3V3 rails, a 1200mA allowance for USB, and you arrive at the 2.5A max power requirement/recommended PSU for Pi3. I can't fathom how @MarkHaysHarris777's draw 2-2.5A unless there are some thirsty peripherals attached, in which case it's hardly comparable to the OP's test setup.

Pithagoros
Posts: 580
Joined: Wed Nov 12, 2014 8:16 pm

Re: Current Consumption Figures

Thu May 12, 2016 8:36 am

gregeric wrote:The core regulator on the Pi3, a RT8088AWSC, has max 3A output, so about 4W @ tops at 1.4V . Allowing for conversion losses, that's about 1A tops on the 5V rail from compute usage alone, with all guns blazing. Add on ~300mA for network etc on the 3V3 rails, a 1200mA allowance for USB, and you arrive at the 2.5A max power requirement/recommended PSU for Pi3. I can't fathom how @MarkHaysHarris777's draw 2-2.5A unless there are some thirsty peripherals attached, in which case it's hardly comparable to the OP's test setup.
Either MarkHaysHarris777 is using his GPIO power pins to heat his swimming pool, or he's made a miscalculation.
I've got a Fluke current clamp that is not under a calibration schedule but is not far out, but according to MarkHaysHarris777, I've never used it.

I have a Pi3B with WiFi and Bluetooth in the USB, and its attached to the official Pi 7 inch touchscreen (power link through GPIO power pins). The whole lot is powered by single Tesco Hudl charger which is rated (printed on it) at 2.1AMP. Never shows any sign of any power problems.

I'll set up a test with a shunt (I also have an Avo 8), and compare it to my Fluke clamp with a Pi Ethernet, keyboard, mouse, HDMI. I would expect 1200-1700mA.

Oh, and if "normal use is subjective" then it's a pretty nebulous parameter.

User avatar
MarkHaysHarris777
Posts: 1820
Joined: Mon Mar 23, 2015 7:39 am
Location: Rochester, MN
Contact: Website

Re: Current Consumption Figures

Thu May 12, 2016 8:51 am

Pithagoros wrote: Either MarkHaysHarris777 is using his GPIO power pins to heat his swimming pool, or he's made a miscalculation ... I've got a Fluke current clamp that is not under a calibration schedule but is not far out, but according to MarkHaysHarris777, I've never used it ... I'll set up a test with a shunt (I also have an Avo 8), and compare it to my Fluke clamp with a Pi Ethernet, keyboard, mouse, HDMI. I would expect 1200-1700mA.
Complete balderdash and poppycock!

... well, I would recommend that you continue to use your 2A PSU; you deserve it. If you search the site you'll notice that there are many official recommends to use a 2.5A supply on the 3B... now, you can do as you please (obviously), but I think there is ample evidence that you're wrong headed about this.

I think you might find it interesting to search this site for the number of people who have experienced crashed 3B units because of inadequate 2.1A power supplies... but do as you will.

And, as has been said so many times already, there appears to be a huge disparity in reported results... and that just can't be... unless there is a huge quality control problem at manufacturing. I have several PI 3B units and they all perform the same. How is it possible that all of your units are soooo different? How come your units are not crashing with 2A supplies? How is it that your fluke measures your PI and the specs are significantly different than the ones I purchased ? You might want to check the battery in your fluke...
marcus
:ugeek:

mikerr
Posts: 2770
Joined: Thu Jan 12, 2012 12:46 pm
Location: UK
Contact: Website

Re: Current Consumption Figures

Thu May 12, 2016 9:08 am

The reason 3A PSU are better than say a 2A may be that they suffer less voltage sag when supplying 1A, not that the pi itself was drawing more than 2amps

Many PSU drop the voltage slightly under load, and this upsets the pi.
Remember these are generally designed as phone chargers where a little voltage sag is not as important.

A portable HDD powered from one of the usb ports will obviously change matters ...
Android app - Raspi Card Imager - download and image SD cards - No PC required !

Pithagoros
Posts: 580
Joined: Wed Nov 12, 2014 8:16 pm

Re: Current Consumption Figures

Thu May 12, 2016 9:57 am

MarkHaysHarris777 wrote: Complete balderdash and poppycock!
... well, I would recommend that you continue to use your 2A PSU; you deserve it.
A perfectly functioning setup is what I have. "deserve" has nothing to do with it.
MarkHaysHarris777 wrote: If you search the site you'll notice that there are many official recommends to use a 2.5A supply on the 3B... now, you can do as you please (obviously), but I think there is ample evidence that you're wrong headed about this.

I think you might find it interesting to search this site for the number of people who have experienced crashed 3B units because of inadequate 2.1A power supplies... but do as you will.

And, as has been said so many times already, there appears to be a huge disparity in reported results... and that just can't be... unless there is a huge quality control problem at manufacturing. I have several PI 3B units and they all perform the same. How is it possible that all of your units are soooo different? How come your units are not crashing with 2A supplies? How is it that your fluke measures your PI and the specs are significantly different than the ones I purchased ? You might want to check the battery in your fluke...
The Fluke lets me know if the battery is a problem.
I've seen all the posts, makes absolutely no difference (and the fact that there is more to the performance of a switched power supply than just the rated voltage).

If you want to keep calling me a liar, then carry on. My stuff works.

hippy
Posts: 5595
Joined: Fri Sep 09, 2011 10:34 pm
Location: UK

Re: Current Consumption Figures

Thu May 12, 2016 11:52 am

MarkHaysHarris777 wrote:If you search the site you'll notice that there are many official recommends to use a 2.5A supply on the 3B
Indeed; and that is the Foundation's recommendation -

https://www.raspberrypi.org/help/faqs/#powerReqs

But that 2.5A figure includes capacity for 1.2A of USB consumption so, without USB devices attached, 1.5A should be adequate and 2A more than suitable.

If your Pi 3B requires a 2.5A supply and it is not taking 1.2A for USB devices it would be interesting to know what is consuming that additional current.

MarkTF
Posts: 295
Joined: Tue Mar 03, 2015 4:59 pm

Re: Current Consumption Figures

Thu May 12, 2016 12:33 pm

MarkHaysHarris777 wrote: And, as has been said so many times already, there appears to be a huge disparity in reported results... and that just can't be... unless there is a huge quality control problem at manufacturing.
There seems to be a great deal of confusion about test conditions. You, in particular, seem to be fixated on the absolute maximum power consumption while the original poster in this thread was attempting to characterize average power consumption for his scenario, which apparently has the GPU and the CPUs mostly doing minimal work and no attached peripherals except as specifically indicated. It's perfectly reasonable that the power consumption is highly dependent upon the specifics of the test.

User avatar
Gerd
Posts: 66
Joined: Wed Mar 16, 2016 10:48 am
Location: Europe

Re: Current Consumption Figures

Thu May 12, 2016 2:02 pm

I would like to point to a neutral resource, unfortunally only available in german language. But I think MarkHaysHarris777 speaks some German, because I have seen you posting in the german subforum.
https://www.elektronik-kompendium.de/si ... 912111.htm
https://www.elektronik-kompendium.de/si ... 002021.htm
Here they examine in detail power supply for the raspi and a guide to choosing the right power supply.
And the guys who have written it are not somebody on the internet. They are the authors of some books of electronic used in education for electronics, engineering etc.
Draw your own conclusions!
To add my own opinion: The rule of thumb I know for choosing power supplies: Take the maximal current of your usecase, add 20-30% safety margin. The recommentation of a company/organization selling electronic stuff must be: Take the maximal current of all usecases, add 20-30% safety margin.

boyoh
Posts: 1299
Joined: Fri Nov 23, 2012 3:30 pm
Location: Selby. North Yorkshire .UK

Re: Current Consumption Figures

Thu May 12, 2016 3:15 pm

davidcoton wrote:
boyoh wrote:Tests carried out would only prove accurate when don under test bench conditions
With instrument calibration checks, Testing two identical Pi might give different results
Then you are left with a problem , witch is correct.
My "thought experiment" is to find a way of making valid, comparable measurements under distributed conditions. Conventional wisdom says it is not possible. Many other scientific and engineering practices were not possible until someone cracked it with new insights. Unfortunately I don't have the necessary insight yet...
Best way might be to put a 10ohm high wattage resistor in the power supply lead
And check the volt drop on each test ( Ohms Law)
If the current draw is around 1 amp, a 10 ohm resistor will drop 10 volts (and dissipate 50 Watts) -- but not from a 5V supply. The resistor value needs to be nearer 0.1 ohms. I don't think the connections to it could be made consistently good enough not to affect results, not everyone can solder. That sort of issue is at the heart of the problem.
Do your calculation again, this time put the 10 ohm resistor in series with the Pi load ? Supply

And not in parallel , Then check the volt drop
BoyOh ( Selby, North Yorkshire.UK)
Some Times Right Some Times Wrong

User avatar
MarkHaysHarris777
Posts: 1820
Joined: Mon Mar 23, 2015 7:39 am
Location: Rochester, MN
Contact: Website

Re: Current Consumption Figures

Thu May 12, 2016 5:09 pm

Gerd wrote: https://www.elektronik-kompendium.de/si ... 912111.htm
https://www.elektronik-kompendium.de/si ... 002021.htm
Here they examine in detail power supply for the raspi and a guide to choosing the right power supply.
And the guys who have written it are not somebody on the internet. They are the authors of some books of electronic used in education for electronics, engineering etc.
I enjoyed reading the articles, Gerd. Thank you.
marcus
:ugeek:

User avatar
davidcoton
Posts: 3944
Joined: Mon Sep 01, 2014 2:37 pm
Location: Cambridge, UK

Re: Current Consumption Figures

Thu May 12, 2016 6:49 pm

boyoh wrote:
Do your calculation again, this time put the 10 ohm resistor in series with the Pi load ? Supply

And not in parallel , Then check the volt drop
Er, yes, it must be in series to measure current (using ohm's law to convert voltage drop to current).
In a series circuit, ALL the current passes through ALL the elements. So the 1A (or thereabouts) that a working Pi requires must ALL pass through the resistor.
A 10 ohms resistor passing 1 amp drops 10 volts. That isn't going to work with a 5V supply. But a 0.1 ohm resistor passing 1A will only drop 100mV. Not enough to upset anything, and it will give a measurable voltage to use in calculating the actual current. (Finding a standardised way for a distributed community to measure that voltage is another story.)

If I've got that wrong, please show how you think 10 ohms will work.
Signature retired

User avatar
MarkHaysHarris777
Posts: 1820
Joined: Mon Mar 23, 2015 7:39 am
Location: Rochester, MN
Contact: Website

Re: Current Consumption Figures

Thu May 12, 2016 6:58 pm

davidcoton wrote: If I've got that wrong, please show how you think 10 ohms will work.
You're both correct. Which shunt you use depends on the meter|galvanometer|fluke you're trying to take the measurement with. For very sensitive instruments you only need a very very small shunt... sometimes this is only a small number of centimeters of nichrome wire! I mean, its called a shunt because it 'shunts' across the leads of the instrument so that when placed in series with the power feed most of the current goes through the shunt, not the meter! I've used 10 ohm shunts, and I've used .1 ohm shunts too! I have a fluke, and I have analog galvanometers also... what are you using to take the measurement?
marcus
:ugeek:

jamesh
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 23078
Joined: Sat Jul 30, 2011 7:41 pm

Re: Current Consumption Figures

Thu May 12, 2016 7:15 pm

Hmm, I use my Pi2 power supply on my Pi3, seems to work fine. Think is a 2A supply. Only have keyboard mouse and DECT ULE USB adapter connected though.

Quad core ARMa53's are not that hugely power hungry.
Principal Software Engineer at Raspberry Pi (Trading) Ltd.
Contrary to popular belief, humorous signatures are allowed. Here's an example...
"My grief counseller just died, luckily, he was so good, I didn't care."

User avatar
davidcoton
Posts: 3944
Joined: Mon Sep 01, 2014 2:37 pm
Location: Cambridge, UK

Re: Current Consumption Figures

Thu May 12, 2016 7:20 pm

MarkHaysHarris777 wrote:
davidcoton wrote: If I've got that wrong, please show how you think 10 ohms will work.
You're both correct. Which shunt you use depends on the meter|galvanometer|fluke you're trying to take the measurement with. For very sensitive instruments you only need a very very small shunt... sometimes this is only a small number of centimeters of nichrome wire! I mean, its called a shunt because it 'shunts' across the leads of the instrument so that when placed in series with the power feed most of the current goes through the shunt, not the meter! I've used 10 ohm shunts, and I've used .1 ohm shunts too! I have a fluke, and I have analog galvanometers also... what are you using to take the measurement?
So -- if we need a 10 ohm shunt, and don't want to drop more than 0.1V across the measuring system, the resistance of the voltage measuring device (in parallel with the shunt) must be about ... click, whirr, clunk ... 0.1 ohms. That would be a fairly useless voltage measuring device, as it will significantly alter the test conditions. And the shunt across it will be ... useless.

A 10 ohm shunt would be useful in measuring lower currents in a high voltage circuit. Not for measuring currents of the order of 1 amp in a 5 volt circuit.

There is as yet no specific voltmeter -- but any reasonable voltmeter has high impedance -- of the order of 30 0000 ohms pewr volt for an average moving coil meter of 40 years ago (that means at 0.1 volts the meter looks like 3 000 ohms, inconsequential across a shunt of even 10 ohms). A digital voltmeter should be even higher. Note however that a voltmeter used as a current monitoring device does not actually need high impedance, but it makes life easier (linearity, inherent accuracy without calibration, etc).
Signature retired

boyoh
Posts: 1299
Joined: Fri Nov 23, 2012 3:30 pm
Location: Selby. North Yorkshire .UK

Re: Current Consumption Figures

Thu May 12, 2016 7:37 pm

davidcoton wrote:
boyoh wrote:
Do your calculation again, this time put the 10 ohm resistor in series with the Pi load ? Supply

And not in parallel , Then check the volt drop
Er, yes, it must be in series to measure current (using ohm's law to convert voltage drop to current).
In a series circuit, ALL the current passes through ALL the elements. So the 1A (or thereabouts) that a working Pi requires must ALL pass through the resistor.
A 10 ohms resistor passing 1 amp drops 10 volts. That isn't going to work with a 5V supply. But a 0.1 ohm resistor passing 1A will only drop 100mV. Not enough to upset anything, and it will give a measurable voltage to use in calculating the actual current. (Finding a standardised way for a distributed community to measure that voltage is another story.)

If I've got that wrong, please show how you think 10 ohms will work.
David Please accept my apologies, you are correct
Must be my old age. Back to ohms law for me,
It's good for somebody to give me some stick

Regards BoyOh
BoyOh ( Selby, North Yorkshire.UK)
Some Times Right Some Times Wrong

occupied
Posts: 42
Joined: Mon Apr 25, 2016 3:50 pm

Re: Current Consumption Figures

Mon May 30, 2016 7:53 am

I now have a Pi Zero to play with :)
But I don't have the same test conditions that I used for the initial batch of testing, and I definitely don't have the time to repeat :|

So...

The clear champ from the first round of testing was the Pi 2B.
Since then, I have been using the Pi 2B to record from the camera module using a 10,400mAh powerbank. I get around 25 hours, give or take.

Therefore I decided to pit the Pi 0 against the Pi 2B.

The test: disable HDMI, LEDs and USB bus. Record from the camera module..
Pi 2B = 342mA
Pi 0 = 216mA

My test showed a clear new winner, with the Pi 0 consuming over a third less than the Pi 2B.

In theory, I should be able to stretch the endurance from 25 hours up to around the 40 hour mark by using the Pi0... Results to follow.

User avatar
alexeames
Forum Moderator
Forum Moderator
Posts: 2848
Joined: Sat Mar 03, 2012 11:57 am
Location: UK
Contact: Website

Re: Current Consumption Figures

Mon May 30, 2016 12:59 pm

I've been doing power measurements since the first Pi came out. Most up-to-date results are here...

Image

Image

These days I use my eMeter and a bench PSU and they both agree with each other fairly closely. My figures also agree closely with those of Dave Akerman and Mike Redrobe. So I am reasonably confident that they are fairly accurate.

Note: my shoot video test is without recording to the SD card, but with an HDMI display attached.
Alex Eames RasPi.TV, RasP.iO

blaablaaguy
Posts: 623
Joined: Sun Sep 27, 2015 3:26 pm

Re: Current Consumption Figures

Mon May 30, 2016 2:47 pm

Wow, the B used a LOT of juice compared to the others
This signature intentionally left blank.

mikerr
Posts: 2770
Joined: Thu Jan 12, 2012 12:46 pm
Location: UK
Contact: Website

Re: Current Consumption Figures

Mon May 30, 2016 3:07 pm

Surprisingly or not, quite a bit less power is used for lower resolution video, e.g. 720p vs 1080p

Setup
Pi Zero + Camera,
WIfi dongle (but its turned off in script below) ,
no HDMI

Script

Code: Select all

#!/bin/sh
sudo tvservice -o
echo 0 | sudo tee /sys/devices/platform/soc/*.usb/buspower >/dev/null
sleep 5
raspivid -t 20000 -w 1280 -h 720-o /home/pi/test720.h264
sleep 5
raspivid -t 20000  -o /home/pi/test1080.h264
sleep 5
echo 1 | sudo tee /sys/devices/platform/soc/*.usb/buspower >/dev/null
- always worth turning USB off even if not using it.

Results:
80mA at the first sleep with USB off i.e. idle

180mA recording 720p
260mA recording 1080p
Android app - Raspi Card Imager - download and image SD cards - No PC required !

occupied
Posts: 42
Joined: Mon Apr 25, 2016 3:50 pm

Re: Current Consumption Figures

Mon May 30, 2016 4:57 pm

Thanks for posting up more figures :)

A few things that have become apparent to me over the past few weeks:
  • There are frequent spikes in consumption on all models, no matter what they are doing. I'm presuming this is caused by other things going on in the background. Therefore I've found I need to average consumption over 20-30 minutes in order to achieve repeatable results.
  • The zero'ing function on the USB multimeter I use (see first post) only zeros out the minutes. Not the seconds. Similar deal with the mAh figure. Therefore I have to wait for the minute and the mAh increments to roll over (usually a few seconds apart), then zero the device.
Hope that helps someone

occupied
Posts: 42
Joined: Mon Apr 25, 2016 3:50 pm

Re: Current Consumption Figures

Mon May 30, 2016 4:58 pm

mikerr wrote: Results:
260mA recording 1080p
Broadly similar to my results. The only difference being I also turn off the LEDs.
Would love to see if that figure changes if you run the test for longer than 20 seconds.
mikerr wrote: Setup
Pi Zero + Camera,
WIfi dongle (but its turned off in script below) ,
no HDMI
I have noticed that my Wifi dongle is illuminated, even with the USB power off. Therefore I always physically remove it.
Last edited by occupied on Mon May 30, 2016 5:13 pm, edited 1 time in total.

occupied
Posts: 42
Joined: Mon Apr 25, 2016 3:50 pm

Re: Current Consumption Figures

Mon May 30, 2016 5:09 pm

alexeames wrote: Image

Note: my shoot video test is without recording to the SD card, but with and HDMI display attached.
Again, interesting to compare your HD shoot with my figure and mikerr:

Me: 216 mA with HDMI off, USB off, LEDs off, recording to SD
alexeames: 240 mA with HDMI on, USB on (?), LEDs on (?), without recording to SD
mikerr: 260 mA with HDMI off, USB off, LEDs on, recording to SD

How long do you run your tests for?

User avatar
alexeames
Forum Moderator
Forum Moderator
Posts: 2848
Joined: Sat Mar 03, 2012 11:57 am
Location: UK
Contact: Website

Re: Current Consumption Figures

Mon May 30, 2016 5:23 pm

occupied wrote:
alexeames wrote: Image

Note: my shoot video test is without recording to the SD card, but with and HDMI display attached.
Again, interesting to compare your HD shoot with my figure and mikerr:

Me: 216 mA with HDMI off, USB off, LEDs off, recording to SD
alexeames: 240 mA with HDMI on, USB on (?), LEDs on (?), without recording to SD
mikerr: 260 mA with HDMI off, USB off, LEDs on, recording to SD

How long do you run your tests for?
USB on, LEDs on. I run for about 30s to 1m and watch the meter readout and bench PSU readout.
Alex Eames RasPi.TV, RasP.iO

occupied
Posts: 42
Joined: Mon Apr 25, 2016 3:50 pm

Re: Current Consumption Figures

Mon May 30, 2016 5:31 pm

alexeames wrote: USB on, LEDs on. I run for about 30s to 1m and watch the meter readout and bench PSU readout.
Thanks :)

VahidD
Posts: 16
Joined: Sat Apr 09, 2016 6:48 pm

Re: Current Consumption Figures

Tue Aug 02, 2016 5:53 pm

Hi and thank you for your interesting post. :-)

I want to measure the current consumption on RPi3 using a UNI-T UT61E True RMS multimeter. Is it possible to do that without utilizing any other devices? I'd be thankful if you answer my question.

Thanks,
Vahid

Return to “General discussion”