The Raspberry Pi Zero uses about 1/3 the power of the A+, or 1/6 the power of the model 2 B.
This is pretty awesome, as it's only about twice the power usage of a typical Arduino setup, so battery-powered embedded projects can start considering the Pi as an option. It helps, of course, that this is just $5!
I just got my mini-HDMI adapter in the mail today, so I'll be running a few more tests to see if there are any other major differences between the Zero, the model 2 B, and the A+/B+ (I currently use the A+ for most small projects right now, and the 2 B when I need a little more CPU).
if power and cost were 1 and 2 this is pretty good for you! Could I ask what the specific applications were? I'm very interested and wonder what other people have done. May have follow-up questions for you :)
I may also disable the HDMI interface as well, to see if it makes any real difference in power consumption. It's rumored to do so, but I haven't seen any proof that it helps.
Can't edit this post anymore, but I'll reply here: I just ran tests with HDMI disabled, and it looks like it does save ~30 mA. Running more tests, will leave a top-level comment with more notes soon.
> The Raspberry Pi Zero - quite a small Linux computer!
Okay, I have a smalllll complaint. In popular modern-day English, R-pi is not really a 'computer'. Because you can't run a full-fledged operating system like Windows or Linux on it -- you have to run embedded version of those OSs (like Windows CE or whatever). I think I may not be totally and technically correct in my point here, but I think it's an important point to make or at least mention somewhere along the way anyway -- because, as it happens, I told someone that R-pi is basically a cheap computer... and this person was eventually miffed to find that he couldn't run the Windows applications he wanted to run on it, because those applications only could run on a x86 Windows install.
What? You can run full-fledged operating system like Linux on Raspberry Pi (all of the variants). Full Linux distribution with desktop environment, browser, e-mail client and package manager.
All Raspberry Pis are computers, even with this mangled definition: just because it can't run Windows doesn't mean it can't run, say, Debian, which is absolutely a "full-fledged operating system".
You can totally run a full desktop Linux install on a Pi. Your facts are incorrect.
pen2l actually you are wrong, because it runs full Linux, but also it's PRECISELY the thing that makes this so attractive to so many people: the fact that it's a full, normal, desktop, Linux PC. That gives it amazing possibilities. You can prototype on a laptop (running Linux) and just move your code, AS-IS, over to a 1 GHz Linux Desktop with 512 MB of RAM: the Raspberry Pi Zero that is pictured.
Now you are correct that it's not capable of running Windows. Even the Linux that it runs is a version that doesn't support everything that a desktop version supports. (Like Photoshop etc.) It's extremely underpowered for a desktop, and not every Linux package supports its architecutre, though most do.
So it's a complete PC in the sense that it can run desktop Linux applications - you can surf the web on it - write emails, do whatever you want, and you won't be using special embedded versions of applications. But for that usage it is also radically underpowered.
Hard to understand what else someone might expect from a $5 computer on a chip, though, when a typical desktop costs $250-$500. The specs I listed are absolutely remarkable, but in a real desktop you use on your desk you would want 8 times as much RAM (4 GB) at a minimum, or even 64 times as much ram (16 GB). I mean if you're advising your friends.
This is more about experimentation for applications where you embed a small PC. If your friend doesn't have much of a budget, I suggest they get a used PC. This PC is about as powerful as a desktop PC from over fifteen years ago. (Which they could also get, and still run some desktop applications on.)
The solution to "cheap windows" is the various Bay Trail devices; so far the best offer I've seen is 7" tablet for £45 which comes with a year's subscription to Office 365 (worth >£50).
This is a very strange comment for HN. I get the gist of what you're trying to say here, but your terminology is completely off base. You really should look up the definitions for "computer," "touring complete," and well, actually learn a bit about the R-pi and arm computers in general.
The R-PI is very much a full fledged computer or "PC" even. Microsoft's reluctance to fully port a desktop operating system to the device has nothing to do with it's technical merits. It's perfectly capable of running a full windows OS, if Microsoft were to choose to do so. BTW, their IoT platform is rumored to eventually offer desktop variants, so this argument will probably be null sooner than later as well.
Otherwise, it's capabilities are damn near identical (functionally) to an X86 machine running Linux, BSD, Android, and a few other operating systems. There are limitations with IO and that sort of thing, which one has to take into consideration for any PC, but otherwise they are 100% functionally equivalent.
Anyway, yes yes I agree with the things you (and others) have said. I failed to say what I wanted to say correctly. The short and sweet of it is, sometime ago I proposed to a friend that I'd help him do some things really cheaply ... (because the R-pi is cheap!) and we eventually couldn't because we couldn't use things like Labview and so on. I shouldn't have extended this issue to terminology, whether R-pi is a computer or not, sorry about that.
Basically the source of my complaint is being sliiightly bothered by the fact that when people call R-pi a computer, the everyday joe-shmoe takes from it that they can run x86 windows on it. They can't. They can maybe run Windows embedded on it (I think?) but certainly not proper Windows. I know and understand that R-pi is a computer by any reasonable definition, I'm just lamenting the reality of things in this one small respect.
Very interesting. Still far from solar-powered territory for outdoor projects, for example. Unless there were a way to wake the pi very easily. However, the raspberry pi includes no power management whatsoever; it's either off or on[1]. Why not? Does the chip not support it?
--> Note that the article specifically calls out disabling USB, disabling ethernet (not relevant here), and disabling video output if not using it. I wonder if the author could try disabling HDMI to see if power is reduced....
This can be mitigated if there were a way to boot quickly on demand. See this writeup on sleepypi, an expansion-board that does just that:[2] So it should be possible to wake on demand by powering up and going through a boot cycle. Since the hardware is SO very fixed (not going to change short of ripping parts of the board off) I wonder why the boot time couldn't be further reduced...what really takes 10 seconds, during which the SD card could have read 200 MB or so? It's nearly enough time to wake from hibernation, were that supported. (SD card bus speed is limited to reading 20 MB/second or so). I'm sure it could be improved.
This other write-up mentions using the previous generation for up to 24 hours for remote-location time-lapse photos [3] - also contains an interesting idea about underclocking. Not sure if this suggestion is relevant, possible, or works.
(could you clarify why you include the second link? Your second link is our article link.)
The comment that you mention mentions savings of "28ma" - could it really be so high? This is hard to believe, or hard to believe that it would translate the raspberry pi zero, as our article link says the idle is 50-70 mA (third photo caption on page) - so if true and the same for this board, disabling HDMI would reduce current by up to 56%! However, on the other hand that might not be true at all. Perhaps the A has a dedicated chip for HDMI, whereas this chip drives it directly; or perhaps the hearsay on current savings from the HDMI is simply wrong. I've submitted a version of this comment as a comment to the author's blog, since he's all set up to investigate. (I think the post is pending moderation.) Probably there will be 0 power savings.
I just tested this three times, and it looks like it's right around 30 mA saved, so the rumors are true!
I'm going to tinker a little bit and see how low I can get the power consumption (disable HDMI, disable LED, and just have Pi running by itself). I'll post back here with more details.
Wow! That is an astounding difference. Thanks for your quick action. If your numbers are correct looks like you just reduced idle usage by close to 50% - 0.2 watts to 0.1 watts or something - if a solar panel has to get through fog and night-time on an attached battery, that is a huge difference!
going to sleep so you won't be getting more comments from me, but thanks for the interaction and your article! great result. I hope you'll have a chance to incorporate the new numbers/savings into your write-up.
I'd really like to find a USB battery pack for one of these which can be used to make the device portable. I know there are loads, but they usually have one or more of the following problems:
1.) Can't power the PI at the same time as being charged
2.) PI consumes power faster than the battery can charge up
3.) Neither of the above two problems, but when you plug the battery into the mains it momentarily disrupts the PI causing it to reboot.
Anyone know of a battery which will last with one of these for the best part of a day or more and which I can just plug into the mains when I'm near a socket to keep it topped up?
[edit] FWIW, my plan (if I can sort the power) is to migrate my email and xmpp services from the fixed in place Intel NUC under my TV at home, to a Pi Zero in my backpack.
I am not a battery or electronics expert but my understanding is that it is not ideal to simultaneously charge and discharge a LiPO battery, so most don't support passthrough.
I use a RAVPower Xtreme (26800mAh - though I'm not sure if I believe that rating) to power a dashcam and had similar requirements to you. However, I've implemented a DPDT relay and a 12V boost regulator to switch between the battery pack and car power (ACC), depending on the ignition status of the car. So when the car is on, the battery is being charged only and the dashcam is getting power from the car. When the car is off, the battery is only being used to power the dashcam.
I've not tested it with a Pi but there is not an interruption in power long enough to disturb the dashcam (the relay spec is 8ms operate time). The battery pack is charged using a 2A 5V charger. I can't properly assess the charge status/history of the battery as it only has 4 LED's and I don't want to take it apart but so far my commute and other driving is long enough to keep it ticking over.
Its definitely possible, think about the circuitry in your phone and how the batteries work with that.
As far as I'm aware, most modern phones can't turn on without a battery pack present in the phone, even when plugged in, which would suggest that the juice is actually coming from the phone battery at all times and not direct from the charger regardless of charging state.
32 comments
[ 4.6 ms ] story [ 80.3 ms ] threadThe Raspberry Pi Zero uses about 1/3 the power of the A+, or 1/6 the power of the model 2 B.
This is pretty awesome, as it's only about twice the power usage of a typical Arduino setup, so battery-powered embedded projects can start considering the Pi as an option. It helps, of course, that this is just $5!
I just got my mini-HDMI adapter in the mail today, so I'll be running a few more tests to see if there are any other major differences between the Zero, the model 2 B, and the A+/B+ (I currently use the A+ for most small projects right now, and the 2 B when I need a little more CPU).
It can be disabled with the command:
Okay, I have a smalllll complaint. In popular modern-day English, R-pi is not really a 'computer'. Because you can't run a full-fledged operating system like Windows or Linux on it -- you have to run embedded version of those OSs (like Windows CE or whatever). I think I may not be totally and technically correct in my point here, but I think it's an important point to make or at least mention somewhere along the way anyway -- because, as it happens, I told someone that R-pi is basically a cheap computer... and this person was eventually miffed to find that he couldn't run the Windows applications he wanted to run on it, because those applications only could run on a x86 Windows install.
You can totally run a full desktop Linux install on a Pi. Your facts are incorrect.
Now you are correct that it's not capable of running Windows. Even the Linux that it runs is a version that doesn't support everything that a desktop version supports. (Like Photoshop etc.) It's extremely underpowered for a desktop, and not every Linux package supports its architecutre, though most do.
So it's a complete PC in the sense that it can run desktop Linux applications - you can surf the web on it - write emails, do whatever you want, and you won't be using special embedded versions of applications. But for that usage it is also radically underpowered.
Hard to understand what else someone might expect from a $5 computer on a chip, though, when a typical desktop costs $250-$500. The specs I listed are absolutely remarkable, but in a real desktop you use on your desk you would want 8 times as much RAM (4 GB) at a minimum, or even 64 times as much ram (16 GB). I mean if you're advising your friends.
This is more about experimentation for applications where you embed a small PC. If your friend doesn't have much of a budget, I suggest they get a used PC. This PC is about as powerful as a desktop PC from over fifteen years ago. (Which they could also get, and still run some desktop applications on.)
The R-PI is very much a full fledged computer or "PC" even. Microsoft's reluctance to fully port a desktop operating system to the device has nothing to do with it's technical merits. It's perfectly capable of running a full windows OS, if Microsoft were to choose to do so. BTW, their IoT platform is rumored to eventually offer desktop variants, so this argument will probably be null sooner than later as well.
Otherwise, it's capabilities are damn near identical (functionally) to an X86 machine running Linux, BSD, Android, and a few other operating systems. There are limitations with IO and that sort of thing, which one has to take into consideration for any PC, but otherwise they are 100% functionally equivalent.
Anyway, yes yes I agree with the things you (and others) have said. I failed to say what I wanted to say correctly. The short and sweet of it is, sometime ago I proposed to a friend that I'd help him do some things really cheaply ... (because the R-pi is cheap!) and we eventually couldn't because we couldn't use things like Labview and so on. I shouldn't have extended this issue to terminology, whether R-pi is a computer or not, sorry about that.
Basically the source of my complaint is being sliiightly bothered by the fact that when people call R-pi a computer, the everyday joe-shmoe takes from it that they can run x86 windows on it. They can't. They can maybe run Windows embedded on it (I think?) but certainly not proper Windows. I know and understand that R-pi is a computer by any reasonable definition, I'm just lamenting the reality of things in this one small respect.
--> Note that the article specifically calls out disabling USB, disabling ethernet (not relevant here), and disabling video output if not using it. I wonder if the author could try disabling HDMI to see if power is reduced....
This can be mitigated if there were a way to boot quickly on demand. See this writeup on sleepypi, an expansion-board that does just that:[2] So it should be possible to wake on demand by powering up and going through a boot cycle. Since the hardware is SO very fixed (not going to change short of ripping parts of the board off) I wonder why the boot time couldn't be further reduced...what really takes 10 seconds, during which the SD card could have read 200 MB or so? It's nearly enough time to wake from hibernation, were that supported. (SD card bus speed is limited to reading 20 MB/second or so). I'm sure it could be improved.
This other write-up mentions using the previous generation for up to 24 hours for remote-location time-lapse photos [3] - also contains an interesting idea about underclocking. Not sure if this suggestion is relevant, possible, or works.
[1] No sleep mode on pi: http://raspberrypi.stackexchange.com/questions/4773/raspberr...
[2] * Pi and SleepyPi: http://marktmarshall.com/2014/08/experimenting-with-raspberr...
* An interesting FAQ on SleepyPi: http://spellfoundry.com/sleepy-pi/sleepy-pi-faq/
[3] Powr saving on raspberry pi A in remote installations: http://babaawesam.com/2014/01/24/power-saving-tips-for-raspb...
from comments ~ http://www.midwesternmac.com/comment/3216#comment-3216 in this article, "Controlling both the PWR and ACT LEDs on the Raspberry Pi 2" ~ http://www.midwesternmac.com/blogs/jeff-geerling/raspberry-p...
The comment that you mention mentions savings of "28ma" - could it really be so high? This is hard to believe, or hard to believe that it would translate the raspberry pi zero, as our article link says the idle is 50-70 mA (third photo caption on page) - so if true and the same for this board, disabling HDMI would reduce current by up to 56%! However, on the other hand that might not be true at all. Perhaps the A has a dedicated chip for HDMI, whereas this chip drives it directly; or perhaps the hearsay on current savings from the HDMI is simply wrong. I've submitted a version of this comment as a comment to the author's blog, since he's all set up to investigate. (I think the post is pending moderation.) Probably there will be 0 power savings.
EDIT: Thanks, author!
I'm going to tinker a little bit and see how low I can get the power consumption (disable HDMI, disable LED, and just have Pi running by itself). I'll post back here with more details.
1.) Can't power the PI at the same time as being charged
2.) PI consumes power faster than the battery can charge up
3.) Neither of the above two problems, but when you plug the battery into the mains it momentarily disrupts the PI causing it to reboot.
Anyone know of a battery which will last with one of these for the best part of a day or more and which I can just plug into the mains when I'm near a socket to keep it topped up?
[edit] FWIW, my plan (if I can sort the power) is to migrate my email and xmpp services from the fixed in place Intel NUC under my TV at home, to a Pi Zero in my backpack.
I use a RAVPower Xtreme (26800mAh - though I'm not sure if I believe that rating) to power a dashcam and had similar requirements to you. However, I've implemented a DPDT relay and a 12V boost regulator to switch between the battery pack and car power (ACC), depending on the ignition status of the car. So when the car is on, the battery is being charged only and the dashcam is getting power from the car. When the car is off, the battery is only being used to power the dashcam.
I've not tested it with a Pi but there is not an interruption in power long enough to disturb the dashcam (the relay spec is 8ms operate time). The battery pack is charged using a 2A 5V charger. I can't properly assess the charge status/history of the battery as it only has 4 LED's and I don't want to take it apart but so far my commute and other driving is long enough to keep it ticking over.
As far as I'm aware, most modern phones can't turn on without a battery pack present in the phone, even when plugged in, which would suggest that the juice is actually coming from the phone battery at all times and not direct from the charger regardless of charging state.
With a WiFi dongle (so it can operate on the local network at least) and HDMI disabled, it hovers around 120 mA. Very, very nice!