DC to DC Converter

Perhaps 30k, aye, that would still work for the companies.

From 30k to the 300k which Farnell and RS are selling just on preorders alone is an order of magnitude rise in volume, and "plummet" is the right word for the BOM cost under such change of numbers.  Plus, the educational market beckons, and that could quite easily result in yet another order of magnitude in the UK alone (with help of government), and certainly across Europe.  Worldwide ... the numbers are staggering.

So no, don't believe the "“we won't make money" statement, that's just part of the company PR.  The PR person in the report was just doing his job, and it seems very effectively. :-)

Farnell's quarterly results come out Thursday, so maybe we will get

some indication of profitability.

Hi,

I use one of these:

http://www.ebay.com/sch/i.html?_sacat=0&_nkw=lm2596&rt=nc&LH_BIN=1

I currently use a 30cm cable made from normal wiring cables. These are MUCH thicker than what you normally find in micro USB cables. I'm currently using a micro USB plug that I bought at farnell for this. But next time I'm going to plug in on the GPIO connector and just provide 5V unfused.

I've adjusted the module to provide 5.01V unloaded. Now under load it provides 4.99V. Then I have a 0.05V drop in my cable, so I end up with 4.94V on my 'pi. Next the fuse drops 0.15V so the rest of the 'pi only gets 4.79V.

If you only power your 'pi, the LM2596 will need only 1V of "dropout". So 6V at the input will be enough. If you're powering other stuff as well, at 2 or 3A, the LM2596 will require up to two volts of headroom, or at least 7V of input.

So... powering the 'pi from two lithium cells should be possible. Make sure you cut off by the time the voltage drops to 6V. Three cells gives you more margin, but probably slightly more losses.

4 NiMH cells is cutting it too close. Use 6. 12V lead acid should work too.

Above some people are saying things about ventilation of the lead-acid batteries. IMHO, if you treat them well, they won't vent. That's the thing about sealed lead-acid batteries: they don't vent. The trick is that a catalist in the battery lets the oxigen and hydrogen recombine producing heat and not an explosion. The "treat them well" means you don't charge anywhere NEAR the maximum conversion capacity of the catalyst. The "treat them well" also means you don't apply enough power to significantly heat up the battery when the catalist is recombining the oxigen and hydrogen to produce water and heat.

So: Sealed lead acid batteries are nice. Stick to the charging rules and they have a long lifetime and do not pose an explosion danger.

The day before you asked this question we started designing our "pipower" module. This module takes 5V input (e.g. from a LM2596 module as linked) and allows software to decide when to power the 'pi. For example the pi might "ask": "Wake me up in an hour". So the 'pi might wake up, find there is not much too do and go to sleep again (i.e. powerdown).

Configuration 1:

You get 5V from somewhere. The module powers the 'pi whenever it feels like it. In the case of the LM2596 module, that will cost you the 7.5mA "no load current" as a battery drain, limiting the "low duty cycle" stamina of the battery powered pi system.

Configuration 2:

You power the module separately. The pipower module switches the 12V to the LM2596 module. A simple non-switching regulator is used to power the pipower module. With a bit of luck the linear regulator providing power to our AVR processor on the pipower module will use less than the 7.5mA that the LM2596 uses in the no-load situation.

Your post has made me realize we haven't designed a battery power monitor yet (i.e. measure the battery, cut off the 'pi when it gets too low), nor a "power up now" button. Those will be included.

Also the software will be able to "watchdog" functions. We can powercycle the 'pi if the raspberry doesn't contact us within an agreed-upon timespan.

Do you have any other requirements? If the current design is not able to do what YOU want, it is not good enough. I want to make it "good enough". :-)

Andrew if you buy a rpi_serial board and one of the expansion boards, I'll throw in a beta "pipower" board for the field-test. :-)

Morgaine,

  the education market may have taken a bit of a hit after Eben's quote today:

"There's nothing specifically educational about this device. The only reason it's applicable is it's cheap."

http://www.pcpro.co.uk/news/education/375106/founder-no-raspberry-pi-for-every-student

Haha, good ol' Eben.

See, you can tell he's an engineer, he's being accurate rather than strategic.  PR is basically about lying, no matter how generously you try to slant it, and engineers are not good at lying.  In an important sense, it conflicts with the engineer's "religion" or core principles.  This is why engineers should never do PR.  The lying is best left to those who lack the engineer's respect for honesty.

That said, he could have been accurate while still promoting Pi for education, simply by following his "There's nothing specifically educational about this device," with "except in the sense that it is mostly open source and hence inherently educational."  That's still accurate.

And, if his audience showed signs of wanting more, he could have added "However, what does have a strong educational element is the motivation that brought this device into existence, because Raspberry Pi was developed for the specific purpose of providing a widely accessible platform for education in computing."

That might have been a bit more encouraging, without descending into manipulating reality for PR.

Morgaine.

When I wrote "This is why engineers should never do PR", I did of course mean "if you want to preserve the current social order".  In contrast, if what you want is a better world, then only engineers should ever do PR.

- "These shoes were made by slave labour with the cheapest materials we could find.  They have a year's guarantee by law, but we really don't expect them to last that long.  And you really shouldn't be wearing shoes with these ridiculous heals, they're bad for your health and my employers should be ashamed."

Andrew,

Check Roger's solution! This is a better solution, as switched regulators are much more efficient. I didn't point it out, because designing one is not easy, but since we have them ready at ebay...

This converter is a step-down converter, so you must have an input voltage well above the 5V. Somewhere above 6V should work, as Roger pointed out.

As for the uninterruptable charging, the solar minty boost might shed some light. It uses only one cell and a step-up converter, so you might have to change it.

As for the endurance, let me explain how you calculate it:

1) The R-Pi uses 5V at 700mA, leading to a 3.5W power consumption

2) All your batteries offer 11.1V, with 6450mAh total charge (in parallel), which goes to 11.1*0.645=71.595Wh available energy

3) Dividing the available energy by the power consumption: 71.595/3.5=20.45h. But this implies in 100% efficience.

4) The converter has a minimum efficience of 85%, so the final endurance must be: 20.45*85% = 17.39h or 17h 23m 15s.

I think it might work for your application...

Your calculations miss that the 'pi uses only 440mA on average. So there is only 2.2W of power used by the 'pi. (on the other hand, a "worst case" calculation is also worth doing).

The 700mA minimum is including the 2x 100mA for the downstream USB ports. If you're not using anything there, the maximum drops to about 500mA. You'll be a bit closer to that than my 440mA measurement if you are running full-HD video decoding on the GPU..... which you probably won't as I don't see you powering an HDMI screen from that battery......

On the charging front, the currently suggested solar panels are too small. They would provide almost enough to power the 'pi in ideal conditions. But no margin to charge the batteries during daytime so that you can survive the night. You need about 3x more solar power to survive in summer and 5-10x (wild guess) more in winter.

I usually do the worst scenario, since I don't want to estimate 27h (using 440mA) or 24h (using 500mA) and be surprised by a low battery after 17h of usage, because a wifi adapter and a webcam went nuts!

Anything beyond 17h is profit! ;-)

BTW, I think that, if Andrew is using it with a webcam and a wireless link, he will need a powered hub! Which lead us to even more consumption...

Morgaine,

   Sometimes engineers can be very effective at marketting because

they are believed even when they should be challenged, just like a discount

store can be very profitable because shoppers tend to mistakenly assume that

all their prices are equally discounted.

   Eben is asked what the greatest challenge currently is, and says its ramping

up volume.  This may be true, but if unchallenged, tends to lead one to believe

that he thinks the design is solid and all that's necessary is more production. 

But I think this theory ought to be challenged.