[Peripherals] [Its.an.education.project] Ivan's latest blog entry on OLPC

[Richard A. Smith]

Edward Cherlin wrote:

> On Sat, May 17, 2008 at 8:32 PM,  <forster at ozonline.com.au> wrote:
>> Wad
>> 
>>> To the person using car batteries for storage --- bad idea, due
>>> to cost over lifetime
>> That's me.
> 
> It is my understanding that car batteries are widely used regardless 
> of that issue, because they are readily available. It is not enough
> to have a better solution. We have to be able to deliver the better 
> solution.

I've got field reports from Cambodia that indeed car batteries are used
to provide power for lighting at night.  They haul them down to a local
who has a diesel generator and get them charged for a fee.  I don't have
any info to what extent and what other things they are used for.  I'll ask.

If they are using car batteries then they must not be discharging them
very deeply.  Everything I've read says if you deep cycle your typical 
car battery more than about 10 times its toast.

> We need a variety of figures, or better still a spreadsheet that will
>  allow us to put in the performance measures we want and tell us how 
> many batteries we will need to do it.

I've messed with several I've found on the Internet and a few things I 
created myself.  If you search for "solar batteries" or things like 
"sizing solar system" you get lots of web sites and spreadsheet that 
will let you calculate things.

But none of those are going to help you with car batteries. They all
assume proper deep cycle cells.  However, some of the more detailed 
sites on LA battery life describe the parameters used to determine life 
expectancy.  If we can get a various sampling of those parameters for 
car batteries then you can make a good stab.

The rule of thumb for deep cycle LA systems seems to be to plan your
depth of discharge (DoD) to 50% so you need 2x the battery
capacity of your load.  I suspect to make that feasible with car
batteries you will need to limit your DoD to the 80%-90% range.

> I estimate that 1 KW would be enough to run a school server and 100 
> XOs in full operation (100% computation, backlight on, peak power 8.7
>  W). I take that to define the base unit that we need. So 8KWh should
>  cover a school day, and somewhat less should cover evening use.

9W is the peak draw I've measured while running from battery.  If the 
battery needs charging then that number jumps up to a max of 17 Watts 
(depending on how low the battery is).  So you will have to size for 
peak draw as well.  A full battery recharge will consume 30 Watt hours 
at the laptop DC input.  So the exact amount of energy required will 
depend on the efficiency of your supply chain upstream of the laptop(s).

> http://wiki.laptop.org/go/Multi_channel_battery_charger
> 
> Charge 15 XOs simultaneously in 2 hours at 150 W-DC. This comes to .3

I've tried to update the stuff on the wiki a bit so it better matches 
reality.  The 1st stage DC/DC converter is rated at 150 W and I don't 
yet know its efficiency (I don't have a prototype for it yet) but it 
should be at least 85%.  The 2nd stage DC/DC converter (the charging 
circuitry) is coming in at around 82%.

The 150W refers to the output power of stage 1. So to reach that you 
will need at least 176W of input.  Out of that 150W 123W will be 
available to the batteries which will charge 8 batteries over the 2 hour 
period.

> usage. I am told that car batteries run 30 Wh/kg (108 kJ/kg), so we 
> would be talking about a few kg of deep cycle batteries per system to

Battery capacity is a tricky measurement since it depends on your power 
draw.  The harder you pull on a battery the less efficient it is.  But 
the real kicker is the DoD% that I was talking about earlier.  It can 
quickly double or triple the required capacity of your battery bank.

More expensive LA battery technologies such as valve regulated lead acid 
(VRLA) or different chemistries like NiMH or LiFePO4 are able to do DoDs 
down to 10% or below.  Across an extended period they seem to be much 
better options.  Just higher initial cost and more limited availability.

> These specs look quite inefficient.

Right now it works out to around 70% overall.  You can do a lot better 
but it increases the cost of the charging circuitry by triple or more 
and since one of those costs is per battery channel it makes a 
considerable impact on the overall price.

>> Can you provide costings for the charger and tech specs?

Estimated cost right now is about $280 - $300 (AC only) the DC input 
option adds another $12 or $15.

Hopefully this covers the tech specs you were interested in.  If not 
I'll be happy to answer any more questions.

Its a bit busy around here this week with country meeting so my response 
may have a few days of lag.

-- 
Richard Smith  <richard at laptop.org>
One Laptop Per Child