User avatar
FiddlerJones
Posts: 17
Joined: Wed Sep 18, 2013 5:44 pm
Location: Italy

Running Raspberry on 3.3 V

Mon Oct 28, 2013 11:26 pm

I know that is possible to power the raspberry via GPIO, connecting 5V to pin 2 and GND to pin 6. Here's my question now: is it possible to power the raspberry connecting 3.3V to pin 1 and GND to pin 6?
Of course I won't be use anything that runs on 5V!

User avatar
mahjongg
Forum Moderator
Forum Moderator
Posts: 12259
Joined: Sun Mar 11, 2012 12:19 am
Location: South Holland, The Netherlands

Re: Running Raspberry on 3.3 V

Mon Oct 28, 2013 11:33 pm

no it shouldn't be possible, as the SoC too is directly connected to 5V. For its internal SMPS for the adjustable core voltage.

That and the 3V3 low drop regulator won't tolerate having 3V3 on its output, but nothing on its input.

DejaVue with this: http://www.raspberrypi.org/phpBB3/viewt ... 98#p443798

elatllat
Posts: 1337
Joined: Sat Dec 17, 2011 5:05 pm

Re: Running Raspberry on 3.3 V

Sat Nov 02, 2013 12:12 am

SBC with 32GB RAM: https://hardkernel.com

FAQ : https://raspberrypi.stackexchange.com

Unanswered: https://www.raspberrypi.org/forums/search.php?search_id=unanswered

User avatar
Hove
Posts: 1205
Joined: Sun Oct 21, 2012 6:55 pm
Location: Cotswolds, UK
Contact: Website

Re: Running Raspberry on 3.3 V

Sat Nov 02, 2013 7:45 am

Yes, it can be made to work like a dream - follow the link from the previous post showing the hydrogen balloons sent into near space running off 3.3v (instructions included) - it does involve desoldering bits of the Pi so any warranty will be void thereafter.
www.pistuffing.co.uk - Raspberry Pi and other stuffing!

karlkiste
Posts: 184
Joined: Tue Jan 22, 2013 8:50 am
Location: berlin, germany

Re: Running Raspberry on 3.3 V

Sat Nov 02, 2013 8:02 am

I've read the link, and think it might be possible. There should however be no need to desolder the regulator - it should be possible to just feed 3.3V to both the 3.3 and 5V pins on the GPIO header.

The regulator shouldn't mind having input and output at the same level as long as it's not higher than 3.3V.

User avatar
mahjongg
Forum Moderator
Forum Moderator
Posts: 12259
Joined: Sun Mar 11, 2012 12:19 am
Location: South Holland, The Netherlands

Re: Running Raspberry on 3.3 V

Sat Nov 02, 2013 1:20 pm

according to Dave Akerman 5V for the SoC is only needed for
A battery sense pin on the BCM2835
.

but if you look at the schematic : http://www.raspberrypi.org/wp-content/u ... .2_027.pdf that is not true!

5V is connected to a pin VDDBAT2, which could be the battery sensing pin Dave is talking about, but it is also connected to four other pins named VDD_BAT1 to VDD_BAT4 on the block called SMPS (switched mode power supply) that have on the opposite end of the block the inductor of the SMTP block pins called VDD_OUT1 to VDD_OUT4 that are connected through an inductor (part of the switcher) to many VDD_COREn pins.

that can mean only one thing, namely that the SoC contains its own SMPS power supply, for the core power, that is fed from the 5V supply!

The final nail in the coffin that the SMPS block is fed from 3V3 is that the block doesn't have any 3V3 inputs!

the core of the PI's SoC is fed from the 5V line, not from the 3V3 line!

Now I'm willing to believe that the SMTP can be fed with lower voltages than 5V. An observation can be made that this power input doesn't need to get a specific voltage, so there is no need to stabilize the voltage to 3V3 for the SoC! in principle you can feed the SMTP inputs of the SoC any voltage between 5V and 3V3!

unfortunately the SoC also has other 3V3 inputs that might not tolerate voltages much higher than 3V3, and its probably impossible to separate the SoC power from the other 3V3 inputs the SoC uses, so you cannot simply put any voltage between 3V3 and 5V on the combined 3V3/5V lines!

but what Dave says is completely true, except for that 5V is not needed just for a battery sense line. The SoC's core is actually fed from the 5V line, but can probably handle a lower voltage like 3V3, so the 3V3 on the 5V line works, as its enough for the PI's SMPS.

User avatar
jojopi
Posts: 3085
Joined: Tue Oct 11, 2011 8:38 pm

Re: Running Raspberry on 3.3 V

Sat Nov 02, 2013 4:23 pm

mahjongg wrote:that can mean only one thing, namely that the SoC contains its own SMPS power supply, for the core power, that is fed from the 5V supply!
And the "BAT" designation on the input pins suggests that it is intended to work with the variable voltage from a battery, rather than fixed 5V. The most likely type of battery would be a nominal 4.2V cellphone Li-ion, which depending on temperature could go down to 3.0V before exhaustion.

(I had always assumed that the alpha boards fed the SoC SMPS from the 3.3V rail, but I can not find anything to indicate this now.)

Gert has said in the past that the SoC should run on 3V3 only, but he had not tried it: http://www.raspberrypi.org/phpBB3/viewt ... 656#p11656. I do not think he ever elaborated on what board modifications would be best, and he even once suggested that the GPU firmware might need to be changed.
unfortunately the SoC also has other 3V3 inputs that might not tolerate voltages much higher than 3V3, and its probably impossible to separate the SoC power from the other 3V3 inputs the SoC uses, so you cannot simply put any voltage between 3V3 and 5V on the combined 3V3/5V lines!
Right, but this thread is about the possibility of feeding exactly 3.3V on the combined line. We do not have the electrical chapters of the datasheet to check, but this seems plausible, and we have at least one report of success.

User avatar
joan
Posts: 14274
Joined: Thu Jul 05, 2012 5:09 pm
Location: UK

Re: Running Raspberry on 3.3 V

Sat Nov 02, 2013 4:55 pm

Why can't someone with a 3.3V supply just try? I tried from a USB PSU but only had partial success as the current I could supply was tiny (so couldn't be sure if the PSU or the Pi failed).

User avatar
daveake
Posts: 184
Joined: Thu Jul 12, 2012 12:07 am

Re: Running Raspberry on 3.3 V

Wed Nov 06, 2013 9:52 pm

mahjongg wrote:according to Dave Akerman 5V for the SoC is only needed for
A battery sense pin on the BCM2835
That came from Eben, though there's every possibility that what I thought he said wasn't what he actually said :-)

I've run many model A's from 3.3V applied to the 3.3 and 5V lines - several balloon trackers with Pi Cam + GPS + radio, a chase car computer running Pi Cam + GPS + WiFi, and a video streamer again with GPS and WiFi. I've met no problems whatsoever.

Dave

boyoh
Posts: 1330
Joined: Fri Nov 23, 2012 3:30 pm
Location: Selby. North Yorkshire .UK

Re: Running Raspberry on 3.3 V

Wed Nov 06, 2013 10:50 pm

FiddlerJones wrote:I know that is possible to power the raspberry via GPIO, connecting 5V to pin 2 and GND to pin 6. Here's my question now: is it possible to power the raspberry connecting 3.3V to pin 1 and GND to pin 6?
Of course I won't be use anything that runs on 5V!



For what reason do you want to back feed 3.3v to
The 3.3v output pin. What is your project ,for doing
this connection ? Knowing this might help a lot
BoyOh
BoyOh ( Selby, North Yorkshire.UK)
Some Times Right Some Times Wrong

User avatar
FiddlerJones
Posts: 17
Joined: Wed Sep 18, 2013 5:44 pm
Location: Italy

Re: Running Raspberry on 3.3 V

Wed Nov 06, 2013 11:22 pm

boyoh wrote:For what reason do you want to back feed 3.3v to
The 3.3v output pin. What is your project ,for doing
this connection ? Knowing this might help a lot
BoyOh
I need the longest run time with the lightest battery option, and the LDO on the raspberry is an enormus waste of energy, so I was thinking about using a switching regulator like this:http://uk.farnell.com/tracopower/tsr-1- ... =lookahead, bypassing entirely the LDO on the RPi. I seem to understand that is entirely possible as long I don't use anything that run on voltage higher than 3.3V.
My first thought was to de-solder the LDO and replace it with the Traco Power one, but apparently that's not necessary, it should be possible to combine the 3.3V and 5V lines together and feed exactly 3.3V, leaving the LDO right where it is, without any problem, am I correct?

Anyway, my project is a high-altitude weather balloon.

User avatar
daveake
Posts: 184
Joined: Thu Jul 12, 2012 12:07 am

Re: Running Raspberry on 3.3 V

Thu Nov 07, 2013 3:22 pm

Yes you can do that. Personally I remove the LDO anyway.

A switcher can be more efficient than an LDO but it weighs more and will likely have a higher dropout voltage than the LDO that I use, so you'll need more batteries. I suggest you do the maths to see what's best for you (see my article linked to above). I can get >5 hours run time from 4 AAAs, which is enough for a normal flight, and that solution weighs less than a switcher solution will.

User avatar
FiddlerJones
Posts: 17
Joined: Wed Sep 18, 2013 5:44 pm
Location: Italy

Re: Running Raspberry on 3.3 V

Thu Nov 07, 2013 4:05 pm

With my configuration it should run for almost 7 hour. (6 AAA, 85% efficiency, 400mA used by the RPi)

The current measured by the multimeter was about 300mA while powering the GPS and the pressure/temperature sensors, and 650mA with the Camera Board taking Full HD video or photo, so, considering taking 1 pic every 30 seconds and a video once in a while, I'm guessing the average is about 400mA.

7 hours = 5 hours flight, 1 hour of intense beeping to find the load and 1 hour just to be safe in case of bad calculations, or bad luck!

while I'm here may I ask you another question? Should I do something to protect the batteries from the low temperature and the Raspberry from overheating (consequence of < 1% atmosphere)?

User avatar
daveake
Posts: 184
Joined: Thu Jul 12, 2012 12:07 am

Re: Running Raspberry on 3.3 V

Wed Nov 13, 2013 4:15 pm

By far and away the best option for stopping any overheating in a weather balloon payload is to not generate the heat in the first place, and for a Pi tracker that means using a model A. Average consumption will then be around the 200mA mark for the entire tracker, except for during any video recording when it will be more than double that.

When I did fly model Bs (before the A came out) I put heatsinks on the USB/ethernet chip and the processor. The latter isn't really needed but the former probably is. The only hot item I had was an external linear regulator on my first flight (I was waiting for a switcher to arrive), and that got hot enough to deform some foam in the case!

With a model A, run time on 4 AAA lithiums and a *good* LDO is 6 hours, enough for a "normal" flight (2-3 hours) and the launch and recovery times. With a model B you'll have to use AAs instead. Whichever you choose, only use Energizer Ultimate Lithium batteries as nothing else will work when cold. Also, alkaline cells really aren't very good in high discharge applications, so you don't get the rated energy out of them.

Dave

boyoh
Posts: 1330
Joined: Fri Nov 23, 2012 3:30 pm
Location: Selby. North Yorkshire .UK

Re: Running Raspberry on 3.3 V

Wed Nov 13, 2013 5:42 pm

If you are a novice on electronics,
Don't take any advice on using
A soldering Iron to the Pi board
Even a skilled technician would
Would have to take care doing It'

Use the correct voltage the Pi
Was designed for.
BoyOh ( Selby, North Yorkshire.UK)
Some Times Right Some Times Wrong

User avatar
jojopi
Posts: 3085
Joined: Tue Oct 11, 2011 8:38 pm

Re: Running Raspberry on 3.3 V

Thu Nov 14, 2013 11:24 am

boyoh wrote: Use the correct voltage the Pi
Was designed for.
Your lovely poetic advice,
Does seem a little misplaced
In a section for users advanced.

When the Pi goes up in a balloon
There's a risk it won't safely return;
It isn't a priceless heirloom.

boyoh
Posts: 1330
Joined: Fri Nov 23, 2012 3:30 pm
Location: Selby. North Yorkshire .UK

Re: Running Raspberry on 3.3 V

Thu Nov 14, 2013 11:38 pm

jojopi wrote:
boyoh wrote: Use the correct voltage the Pi
Was designed for.
Your lovely poetic advice,
Does seem a little misplaced
In a section for users advanced.

When the Pi goes up in a balloon
There's a risk it won't safely return;
It isn't a priceless heirloom.
BoyOh ( Selby, North Yorkshire.UK)
Some Times Right Some Times Wrong

boyoh
Posts: 1330
Joined: Fri Nov 23, 2012 3:30 pm
Location: Selby. North Yorkshire .UK

Re: Running Raspberry on 3.3 V

Thu Nov 14, 2013 11:40 pm

boyoh wrote:
jojopi wrote:
boyoh wrote: Use the correct voltage the Pi
Was designed for.
Your lovely poetic advice,
Does seem a little misplaced
In a section for users advanced.

When the Pi goes up in a balloon
There's a risk it won't safely return;
It isn't a priceless heirloom.

As a advanced user you don't seem
Very confident of your project working
You seem to be asking questions
You all ready know the answers to
I hope the radio transmitter will
Work ok

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

User avatar
FiddlerJones
Posts: 17
Joined: Wed Sep 18, 2013 5:44 pm
Location: Italy

Re: Running Raspberry on 3.3 V

Fri Nov 15, 2013 12:57 am

daveake wrote:By far and away the best option for stopping any overheating in a weather balloon payload is to not generate the heat in the first place, and for a Pi tracker that means using a model A. Average consumption will then be around the 200mA mark for the entire tracker, except for during any video recording when it will be more than double that.

When I did fly model Bs (before the A came out) I put heatsinks on the USB/ethernet chip and the processor. The latter isn't really needed but the former probably is. The only hot item I had was an external linear regulator on my first flight (I was waiting for a switcher to arrive), and that got hot enough to deform some foam in the case!

With a model A, run time on 4 AAA lithiums and a *good* LDO is 6 hours, enough for a "normal" flight (2-3 hours) and the launch and recovery times. With a model B you'll have to use AAs instead. Whichever you choose, only use Energizer Ultimate Lithium batteries as nothing else will work when cold. Also, alkaline cells really aren't very good in high discharge applications, so you don't get the rated energy out of them.

Dave
Thank you, these are some really good advice! Of course I'm using a model A, but for some reason it still drains a lot of current, maybe I need to do something via software, I'm gonna try using Arch and disable the audio/video exits, disconnect keyboard/mouse and do some more tests.
boyoh wrote: As a advanced user you don't seem
Very confident of your project working
You seem to be asking questions
You all ready know the answers to
I hope the radio transmitter will
Work ok

Regards BoyOh
There's never to be absolutely confident that a project like this will work. The balloon might not explode, the RPi might overheat, the batteries might die before the playload return to the ground, the playload may fall into the sea, etc.
Besides this is my first attempt, so yeah, I'm absolutely not confident of my project working.
Before this topic I did not know that it was possible to run the RPi using just 3.3V, now I do and I can hope to get a little bit higher.
Also, with other people trying to do something that the RPi is not supposed to do (like run on 3.3V) we may find other consequences, weather bad or good.

I don't think that asking question, in order to use something NOT for the purpose it was built is a bad thing. I don't think asking question is ever a bad thing!

User avatar
daveake
Posts: 184
Joined: Thu Jul 12, 2012 12:07 am

Re: Running Raspberry on 3.3 V

Sun Nov 17, 2013 1:28 pm

If your model A Pi is drawing 400mA, something is wrong. Perhaps a bad DMM. It should be around 115mA.

I tried a few things to reduce that, and the only noticeable power reduction was to switch off the video output.

On top of that, you'll need around 80mA for the radio and GPS, so total for the tracker should be around 200mA.

The LDO comes off easily provided you use a hot soldering iron with a bit large enough to make contact with most of the tab. Put some solder on the bit first to aid conduction. The risk of damaging the Pi is minimal and you certainly don't want to risk losing your payload because the batteries died before you got to it.

Dave

karlkiste
Posts: 184
Joined: Tue Jan 22, 2013 8:50 am
Location: berlin, germany

Re: Running Raspberry on 3.3 V

Sun Nov 17, 2013 3:31 pm

daveake wrote:The risk of damaging the Pi is minimal and you certainly don't want to risk losing your payload because the batteries died before you got to it.
The batteries should not be affected. If 5V and 3.3V rails are wired to 3.3V both, the LDO shouldn't consume more than a few milliamperes (max. 10 mA says the datasheet). Those can be saved by just cutting both the little wires on the LDO. Desoldering then has only impact on the weight, not on the current draw. The weight is not very high - I would estimate that shaving the GPIO pins will have a bigger effect.

But, I agree, with a good, powerful, regulated soldering iron with a short, wide, hot tip, desoldering shouldn't be a problem. Components tend to die when soldered for too long, not too hot. And, the LDO is broken anyway. It's the surrounding components that count.

boyoh
Posts: 1330
Joined: Fri Nov 23, 2012 3:30 pm
Location: Selby. North Yorkshire .UK

Re: Running Raspberry on 3.3 V

Sun Nov 17, 2013 4:12 pm

jojopi wrote:
boyoh wrote: Use the correct voltage the Pi
Was designed for.
Your lovely poetic advice,
Does seem a little misplaced
In a section for users advanced.

When the Pi goes up in a balloon
There's a risk it won't safely return;
It isn't a priceless heirloom.
That is a very honest answered, and not misplaced
In the advanced users section.
I'm 82 and still learning, I wish you great success
In your project, and your balloon don't get full of
Hot air, Please post on your coming success

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

User avatar
FiddlerJones
Posts: 17
Joined: Wed Sep 18, 2013 5:44 pm
Location: Italy

Re: Running Raspberry on 3.3 V

Sun Nov 17, 2013 4:38 pm

daveake wrote:If your model A Pi is drawing 400mA, something is wrong. Perhaps a bad DMM. It should be around 115mA.

I tried a few things to reduce that, and the only noticeable power reduction was to switch off the video output.

On top of that, you'll need around 80mA for the radio and GPS, so total for the tracker should be around 200mA.

The LDO comes off easily provided you use a hot soldering iron with a bit large enough to make contact with most of the tab. Put some solder on the bit first to aid conduction. The risk of damaging the Pi is minimal and you certainly don't want to risk losing your payload because the batteries died before you got to it.

Dave
It does not draw 400mA, it does draw ~300mA (around 285mA to be more Precise), i did the test while using the Radio, the GPS and a Pressure/Temperature sensor altogether, and while using a keyboard and a mouse too, without switching any output off, so I guess that 285mA it's pretty much right. I calculated 400mA as an average because with the camera board the drawing goes up to 650mA. I've written the software to take pictures and data every 30 seconds, and video at certain altitudes. In 5 hour flight, I calculate to use the camera for one hour, so 20% of the time. (0.8*300+0.2*650) = 370mA as an average use. I'll do the math again when the hardware/software part is over, switching the video off, disconnecting the keyboard/mouse and with an optimized software to use the camera at minimum.

About losing the playload, I was thinking to add an indipendent GSM/GPS tracker, like this: http://alinpopescu.com/2013/06/25/how-g ... tem-works/ so even if the batteries dies before the landing I still have a chance to find the playload. I'll settle for lower altitudes, but after all this is a trial flight.

User avatar
daveake
Posts: 184
Joined: Thu Jul 12, 2012 12:07 am

Re: Running Raspberry on 3.3 V

Tue Nov 19, 2013 8:13 pm

OK, well obviously the keyboard and mouse won't be flying, so test again without those! Also, the GPS will draw a lot more current before it acquires a lock than when it does after.

I don't fly the GPS/GSM trackers anymore. I did use one twice as backup, and it worked one of those 2 flights. Briefly. I think they're a waste of money unless you're flying an expensive payload.

The camera only draws high current briefly, so the average will be low.

Best bet as always is to test the whole thing running the same code you're going to fly, with the same type of batteries. See how long it lasts till it dies.

If the run time is a bit tight, consider only running the camera when (say) above 200 metres altitude. Also, definitely, switch off the HDMI/TV outputs. Finally, the GPS may have some power-saving modes. Or, more simply, add batteries!

Dave

User avatar
FiddlerJones
Posts: 17
Joined: Wed Sep 18, 2013 5:44 pm
Location: Italy

Re: Running Raspberry on 3.3 V

Sat Nov 23, 2013 8:56 pm

I did the test again, no keyboard, no mouse, locked GPS (thanks for the heads up), video outputs off, and I've optimized the software to turn off the temperature/pressure sensor between the data collection and to use the camera taking video only where I need (for a final ~ 3 min videoclip). So I've also reduced the usage time of the camera to 4-5%. The average drain should be now around 250 mA, and that's very good! I'll do some run tests to see how long will it last with various options!

I already required the permission for a flight this spring (here in the winter the winds can be very annoying), I'll report success or failure here! Finger crossed!

Thanks for the help!

P.s. About the topic subject: I'm running the Rpi feeding 3.3V on the 5V and 3V pins and it's working. I've encountered no problem so far!

Return to “Advanced users”