24/7 Solar Powered Pi in the UK

[Evan Davey]

Hi All

I am investigating the feasibility of running a Raspberry Pi (let's say Model B) 24/7 in a remote site in the UK.  The budget for panel + battery would be no more than £200 pounds.  

Firstly, I haven't done much work with solar before, so would greatly appreciate an opinion on the calcs below which could be very wrong (which suggest summer will be fine, but no chance in winter).  

Secondly, thinking about the problem in a different way, is there a way to work out the maximum Wh I could power for that budget (which would give me a constraint to optimise in the search for more efficient components)?

Thirdly, budget constraints aside, what size panel / battery could achieve the desired results with a pi?

Any help much appreciated.

Cheers,

Evan

Calcs

The data sheet suggests the model B consumes 700ma @ 5V (unclear if this is an average or max) which is 3.5W.  Running it constantly this is 24h*3.5W = 84Wh.

Quick research suggests (very roughly) £200 will get about a 50W panel and maybe a 80Ah battery.  

Putting this into the stand alone PV calculator at http://re.jrc.ec.europa.eu/pvgis/apps4/pvest.php (and using the default 40% cut-off and a location around London) suggests there is no chance in winter.

Month Ed Ff Fe Jan 40.00 1 85 Feb 69.00 6 49 Mar 89.00 49 3 Apr 84.00 87 0 May 83.00 87 0 Jun 84.00 95 0 Jul 83.00 96 0 Aug 83.00 85 0 Sep 83.00 70 0 Oct 80.00 33 0 Nov 51.00 1 71 Dec 39.00 0 93 Year 73.25

Ed: Average energy production per day (Wh/day)

Ff: Percentage of days when battery became full (%)

Fe: Percentage of days when battery became empty (%)

[Jonty Wareing]

I'm going to assume that this isn't connected to the internet as you
have no power, which leads to the question: Is a raspberry pi the right
thing to be using?

What will it be doing out there in the wilderness? It might be a hell
of a lot easier to use something else without such high power
requirements.

I've personally looked into this before and gone "Nope", which is why
I ask if a pi is the most appropriate device.

--jonty

> E d : Average energy production per day (Wh/day)
> F f : Percentage of days when battery became full (%)
> F e : Percentage of days when battery became empty (%)
>
>
>
> -- You received this message because you are subscribed to the Google
> Groups "London Hackspace" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to london-hack-sp...@googlegroups.com.
> For more options, visit https://groups.google.com/groups/opt_out .

[Toby Catlin]

Is combing solar and wind feasible?
I setup a remote WiFi node that ran off a battery charged by large solar panel and small wind turbine. The turbine was from a boat shop and it ran all year, mostly. You could try building one from a stepper motor and a bunch of diodes.

A lower power device is a great idea. There isn't much a Rpi can do that can't be done better by something else.

T

[Evan Davey]

Yeah, agreed.  I need something 'off the shelf' that is also CE certified (which rules out many of the development boards).  Can maybe put together a microcontroller based solution (eg: something building on Arduino) but was leaning towards embedded linux since the device needs to act as a wireless base station (which of course, will increase power needs further).

Also looking at hacking a TP Link TL-WR703N which looks to have much lower power consumption (under 1W max).  

There is power around, just a desire not to use it for ease of installation.  That might have to give...

Any thoughts on how to take a dollar budget and turn that into a maximum Wh?

[Evan Davey]

Wind could be an option.  This is a prototype that could turn into something commercial so there is a trade-off in terms of reliability etc.  Thinking wind would be more prone to failure etc.

Ideally, I'd do a custom build and look at full power optimisation eg. sleep modes etc. but limited on time and budget.  Raspberry Pi (or equivalent) seems like an easy answer due to possibility of plug in wifi adapters etc. (I know, not without further issues).  If not PI, any thoughts on another linux platform that's lower powered and CE certified?

I've seen the odd solar powered sign etc. around the UK - do these have a mains backup or are they just low powered enough (and probably don't need 24/7)?

[wyan std]

A Pi with a WiFi dongle will easily draw up to 1A current or worse when transmitting, which would require very aggressive power optimisation or a bigger battery (I think the solar powered signs used in the roads usually have huge batteries).

There is apparently an Arduino shield with WiFi on the works, and the WiFi module runs an embedded linux. Perhaps this will be low powered enough?

http://www.kickstarter.com/projects/sparkdevices/spark-core-wi-fi-for-everything-arduino-compatible

[wyan std]

This one looks pretty neat as well:

http://www.kickstarter.com/projects/1286098094/wirelessly-connect-all-the-things-with-sapphire

[Metz]

The figures being quoted in this thread for the Pi consumption as maximum values. I have measured a Pi, with a Wifi dongle connected, at less than 450mA, max. This was a model A, but that would clearly be best for this anyway. The dongle was a £9.99 from Maplin, unbranded. Granted, the Wifi probably didn't have to ramp up, as it was line-of-sight to the base station about 15ft away when I ran the tests. It was running a network file copy at the time.

[Nick Leaton]

http://jeelabs.org/

It's arduino based, but there is a lot of stuff on his website about low power and controllers. 

--
Nick

[Martin Klang]

On 23 May 2013, at 07:49, wyan std wrote:

> A Pi with a WiFi dongle will easily draw up to 1A current or worse when transmitting, which would require very aggressive power optimisation or a bigger battery (I think the solar powered signs used in the roads usually have huge batteries).
>
> There is apparently an Arduino shield with WiFi on the works, and the WiFi module runs an embedded linux. Perhaps this will be low powered enough?
>
> http://www.kickstarter.com/projects/sparkdevices/spark-core-wi-fi-for-everything-arduino-compatible

I was going to suggest this - I'm already getting one myself.
3.3v and
• 50mA typical current consumption
• 300mA peak current consumption (during transmit events)

Note that it's got an ARM Cortex M3 mcu so it's a LOT more capable than a regular 'duino.

Again, it comes down to what you plan to do with it. Can you power the circuit down for most of the time, or does it need to run constantly?
How often and for how long do you need to be on the network?

If you can design it so that the device sleeps for most of the time then you will be able to reduce power consumption to a fraction. You can do this using the Watchdog timer built in to most MCU's, including Arduino/atmegas and ARMs. It's like setting an alarm; when it goes off the device wakes up and does what you've told it, e.g. make a measurement and send it off then go back to sleep.

hth,

Martin

[Toby Catlin]

Might be useful information for you

http://www.raspberrypi-spy.co.uk/2013/02/running-a-raspberry-pi-from-6-aa-batteries/

[Peter "Sci" Turpin]

I note that you can get the Pi running on a lot less power if you're
willing to remove the linear regulator and replace it with a switch-mode
regulator. The quoted power saving has been about 25%.

While power draw varies massively depending on what you're doing (200mA
or less at idle), having a peak consumption around 525mA might make a
lot of difference.

On 22/05/2013 18:37, Evan Davey wrote:
> Hi All
>
> I am investigating the feasibility of running a Raspberry Pi (let's say
> Model B) 24/7 in a remote site in the UK. The budget for panel +
> battery would be no more than �200 pounds.
>
> Firstly, I haven't done much work with solar before, so would greatly
> appreciate an opinion on the calcs below which could be very wrong
> (which suggest summer will be fine, but no chance in winter).
>
> Secondly, thinking about the problem in a different way, is there a way
> to work out the maximum Wh I could power for that budget (which would
> give me a constraint to optimise in the search for more efficient
> components)?
>
> Thirdly, budget constraints aside, what size panel / battery could
> achieve the desired results with a pi?
>
> Any help much appreciated.
>
> Cheers,
> Evan
>

> *Calcs*

>
> The data sheet suggests the model B consumes 700ma @ 5V (unclear if this
> is an average or max) which is 3.5W. Running it constantly this is
> 24h*3.5W = 84Wh.
>
> Quick research suggests (very roughly) �200 will get about a 50W panel
> and maybe a 80Ah battery.
>
> Putting this into the stand alone PV calculator at
> http://re.jrc.ec.europa.eu/pvgis/apps4/pvest.php (and using the default
> 40% cut-off and a location around London) suggests there is no chance in
> winter.
>

> Month E_d F_f F_e

> Jan 40.00 1 85
> Feb 69.00 6 49
> Mar 89.00 49 3
> Apr 84.00 87 0
> May 83.00 87 0
> Jun 84.00 95 0
> Jul 83.00 96 0
> Aug 83.00 85 0
> Sep 83.00 70 0
> Oct 80.00 33 0
> Nov 51.00 1 71
> Dec 39.00 0 93

> *Year* *73.25*
>
> E_d : Average energy production per day (Wh/day)
> F_f : Percentage of days when battery became full (%)
> F_e : Percentage of days when battery became empty (%)

[ܔܢܜܔηιηℓιℓιzιܔܢܜܔ‎]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

a 50watt panel has little chance for an english summer unless its
beautifully dry and clear for months on end.

I think you need to shop around for panels. I recently picked up a
250watt for £150 from a place in london.
I really wouldnt go for less than 100watt to cover a pi... If you cant
find a decent 200watt for about £100. Your not looking hard enough.
Cant comment about dark winter days... But I use a 50watt to run a
similar power draw off a 20amp/h lipo, for about 4/5 hours a day.....
Its may, it only wins on the days its clear for at least 4/5 hours of
daylight. I'm using a rather expensive MPPT regulator for this
setup... Which is the component which will likely eat most your budget.

There are some mods you can make to the pi that will reduce its hunger.
It has onboard a really rubbish linear regulator that burns 30% of its
power as heat.... Thats your low hanging fruit. Will cost you about
£10 to replace it with something 99% efficient. But a better option
would be to remove the linear reg entirely... and build a switch mode
power supply to feed the pi 3.3v directly from your battery.

Regards,
/Nin lil'izi/

GPG Fingerprint: C510 909B 811E D6F5 0DFF 5D91 CF03 8FEA FD69 4622

On 22/05/13 18:37, Evan Davey wrote:
> Hi All
>
> I am investigating the feasibility of running a Raspberry Pi (let's
> say Model B) 24/7 in a remote site in the UK. The budget for panel
> + battery would be no more than £200 pounds.
>
> Firstly, I haven't done much work with solar before, so would
> greatly appreciate an opinion on the calcs below which could be
> very wrong (which suggest summer will be fine, but no chance in
> winter).
>
> Secondly, thinking about the problem in a different way, is there a
> way to work out the maximum Wh I could power for that budget (which
> would give me a constraint to optimise in the search for more
> efficient components)?
>
> Thirdly, budget constraints aside, what size panel / battery could
> achieve the desired results with a pi?
>
> Any help much appreciated.
>
> Cheers, Evan
>

> *Calcs*

>
> The data sheet suggests the model B consumes 700ma @ 5V (unclear if
> this is an average or max) which is 3.5W. Running it constantly
> this is 24h*3.5W = 84Wh.
>
> Quick research suggests (very roughly) £200 will get about a 50W
> panel and maybe a 80Ah battery.
>
> Putting this into the stand alone PV calculator at
> http://re.jrc.ec.europa.eu/pvgis/apps4/pvest.php (and using the
> default 40% cut-off and a location around London) suggests there is
> no chance in winter.
>

> Month E_d F_f F_e Jan 40.00 1 85 Feb 69.00 6 49 Mar 89.00 49 3

> Apr 84.00 87 0 May 83.00 87 0 Jun 84.00 95 0 Jul 83.00 96 0 Aug
> 83.00 85 0 Sep 83.00 70 0 Oct 80.00 33 0 Nov 51.00 1 71 Dec 39.00 0

> 93 *Year* *73.25*
>
> E_d : Average energy production per day (Wh/day) F_f : Percentage
> of days when battery became full (%) F_e : Percentage of days when
> battery became empty (%)
>
>

> -- You received this message because you are subscribed to the
> Google Groups "London Hackspace" group. To unsubscribe from this
> group and stop receiving emails from it, send an email to
> london-hack-sp...@googlegroups.com. For more options,
> visit https://groups.google.com/groups/opt_out.
>
>

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.19 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/
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=pq9C
-----END PGP SIGNATURE-----

[Jon Russell]

This has some interesting info in it too ...

https://www.sparkfun.com/news/1131

Jon.

[David Murphy]

Can you disclose any more details about the goals of the project?

Does it require a constant wifi connection or does it just need to be able to send/receive a small amount of data now and then?etc

[Evan Davey]

Hi All

Thanks very much for all the responses.  The plan now is to prototype using Arduino (where we can have more control over power consumption) and then re-evalute if the business case is there.  It's a trade-off between grabbing what's available off-the-shelf and spending the time on custom development.  I'll probably try to build something up with shields if I can but not adverse to designing something bespoke if needs be.

wyan - I have a ATMega128RFA1 sample I've been meaning to play with.  It looks like a great chip and I'm surprised it's not in wider use

marser - I've been following the spark closely as I've just been designing a break out for the CC3000 myself. If we go bespoke, it's probably what I'll use.

metz - that's an interesting experience and good to know.  I had a project working with enocean wireless and a pi and found power consumption was all over the place (and I result I have lost confidence in the pi as a long-term solution for anything).

nick - yeah, I subscribed to the jeelabs blog and follow that closely.  The RFM12B is a good cheap module but in this instance I need wifi.  We might be connecting a child and master node later so it could be good then.

jon - that's a good link, I've also looked at https://www.sparkfun.com/tutorials/309

toby - good link and some handy reference info there

nin - good to know. In the end, we've decided size of the panel is going to be an issue for us so we need to optimise power rather than going with a bigger panel

This came up as an option for an off-the-shelf http://www.libelium.com/products/waspmote/ solution but cost and lack and of control leant us towards a bespoke build if the Arduino proof-of-concept works.

David - can't disclose a massive amount.  The goal is to have a self-powered device that can act as a base station that allows mobiles (running an app) to detect and connect to for a small amount of data transfer.  Have also considered bluetooth over wifi.

The actual application will need wifi (or bluetooth), nfc, sms/gprs, and maybe xbee or rfm12b if there are master/slave client nodes - so we are starting to get relatively complex and power hungry.

To that end, any recommendations on good Arduino shields to meet the above would be greatly appreciated.  

Evan

[David Murphy]

just for something a little off the wall: if you've considered bluetooth have you considered using a cheap phone with bluetooth running an app as the base station as well?  

I'm pretty sure a lot of off the shelf phones could probably handle that while they also tend to be very power efficient. 

[Mark Steward]

Also, if you get a newer phone, you can use Bluetooth Low Energy, which is actually more like Zigbee.  Higher range than traditional Bluetooth and lower overall power consumption, especially for infrequent connections.

Mark

[Dirk-Willem van Gulik]

On 24 mei 2013, at 00:35, Evan Davey <evan.j...@gmail.com> wrote:

Thanks very much for all the responses.  The plan now is to prototype using Arduino (where we can have more control over power consumption) and then re-evalute if the business case is there.  It's a trade-off between grabbing what's available off-the-shelf and spending the time on custom development.  I'll probably try to build something up with shields if I can but not adverse to designing something bespoke if needs be.

I found RMF12B's quite workable for this type of scenario; low power ones with Arduino's/atmels/pics are commonly found - a good example may be

http://jeelabs.com/products/jeenode

as he is quite focused on low power. I've got some of his modules in temperature measuring mode - running now close to 2 years on just a penlite with a broadcast every 300 seconds.

Dw.

[Evan Davey]

Anyone managed to get an Arduino with wifi to act as an access point?  

Can't think why it wouldn't be possible but I haven't managed to find any examples.

--
You received this message because you are subscribed to a topic in the Google Groups "London Hackspace" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/london-hack-space/KUt1y9UzdUk/unsubscribe?hl=en-GB.
To unsubscribe from this group and all of its topics, send an email to london-hack-sp...@googlegroups.com.