[Power] Haiti power questions
Richard A. Smith
richard at laptop.org
Thu Dec 2 03:22:57 EST 2010
The following is from a thread that Tim Falconer started with me on one
of his Haiti trips. He's trying to establish a template for Haiti
power. I've pulled it onto power@ because that's what I created this
list for. :)
>>> Where are the best numbers for Wattage (power consumption) of the
>>> XO-1: 1) after the battery is full and green?
>> Assuming AC side of the adapter.
>>
>> According to Google, Haiti is 120Vac 60Hz. If your generator is
>> using a different configuration then I'll need to adjust the
>> numbers.
>>
>> After the battery has been charged the XO draws approx 4-8 Watts
>> but it can peak up to 9W. When you reflect that load back to the
>> AC side of the adapter and compensate for efficiency you get an AC
>> line draw of 6-11W. I usually use the XO avg draw as 5W.
>
>> Here's a chart:
>>
>> XO AC
>> -- --
>> 4 W 6 W
>> 5 W 7.5 W
>> 8 W 11 W
>> 9 W 12.4 W
> Okay, so for the situation where the battery is already charged and
> it's plugged into a 120Vac 60Hz outlet, I should use 7.5 W as the
> average draw, correct?
Thats my average assuming that the XO can go idle (not suspend) since I
happen to know that you use etoys extensively I ran some logs with a few
machines running etoys.
if etoys is running the (battery) power draw is between 6 and 7W
depending if its in design mode or if its ticking. That's going to
translates to between 9 to 10 W at the Wall.
Updated Chart.
XO AC
-- --
4 W 6 W
5 W 7.5 W
6 W 9 W
7 W 10 W
8 W 11 W
9 W 12.4 W
17 W 23.5 W
> If the XO is not plugged in, I should use 5 W as the draw, correct?
I don't think you really care when its not plugged in. You are running
from the battery which is stored charge. The requirements of that
stored charge should be included in the over all AC power draw budget.
(But just FYI as above in etoys its 6-7. XO sitting there untouched idle
is about 4W)
> What would the average numbers be if the laptop is also charging
> (assume dead battery at the start of each class, which is basically
> what happens with kids bringing the laptops in from home)?
You can't really do an average here. The reason is that when you are
charging the battery the power draw is not constant. At the moment you
plug up the empty battery you are going to peak at 23.5 W but then
within the first 3 or 4 minutes that draw will begin to drop. Thats
assuming XO is not powered up, if its powered up then the the 23.5 draw
will stay for a lot longer. As the power draw decreases the efficiency
of the power adapter changes. So the power draw during battery charging
is iterative value that you have to compute across the charging time.
If you are going to factor in battery charges then the only way to do it
without going crazy is to fip over into total energy required to charge
a battery from empty to full. Based on my measurements the amount of
juice it takes at the AC plug to charge a battery is about 50-60
Watt-hours.
Lets take the high end number as worst case. So if you have 50 XOs and
you want all of them to get charged then you need a minimum of 60kWh *
50 XO = 3 kilowatt-hours of juice. How you distribute that juice
depends on your setup.
If the XO is powered up it will be more. Looking through all my data I
see that I don't really have a good measurement of that. But using the
estimated draw of a XO running etoys from above I think its pretty safe
to add an extra 20 Wh on the estimate. So say 80 Wh if the XO is on and
60 Wh if its off.
> Are these numbers appropriate for both power-grid systems and solar
> to marine battery systems?
If you are on a solar power set then there is one more set of efficiency
numbers you have to factor in. Thats the eff% of the AC inverter.
Typically those are 90% or more. That means that you should take all
the numbers above and divide them by .9 if you are on solar. Thats the
juice that your batteries will have to provide.
> What would the comparable average draw be for a generator powered
> setup? (Two numbers ... battery starts as charged, battery starts as
> empty) Assume your 5 W average.
There won't be any difference in the power draw on a generator compared
to a grid. You need:
~60 Watt-hours to charge the battery
~10 Watt-hours for running etoys.
Full battery is easy. 10W * hours run.
Empty battery is more difficult to estimate at least 60Wh and if powered
on laptop is going to consume 10W/hour and charging will take at least 2
hours then you have at least 80 Wh.
The concern with generators is the power factor. Generators are
specified at a power factor of .8 . That's a good number to use if you
are powering a house.
Switching power supplies however don't behave nicely. They are
non-linear in their operation so they don't map well on to the
"standard" power factor estimate.
AC circuits are not steady state. In a 60Hz the voltage goes from max
positive to max negative 60 times a second. So on the wire there's a
lot of things happening. Without getting in too deep what power factor
tells you is how much extra stuff is going on on the wire. A perfect
system has a pf of 1.0 (its a ratio number) .8 is a good all around
estimate. A block of XO power supplies however can present a power
factor of .5 or lower.
Many generators are rated in watts only but there is also a limit to the
amount of current they can provide. If you run over that limit for an
extended period the windings in the motor get hot and it will burn up.
Power factor is a measure of how large that dynamic current is for a
given load.
So if you take a 3kW generator rated for a pf of .8 and run it at 3kW on
a system with a pf of .5 your generator is going to fail pre-maturely.
You have to oversize the generator to compensate for the lower power
factor. This is the VA rating.
3kW / 120 Vac = 25Amps pf = 1. At pf of .8 the maximum dynamic current
gets adjusted by the pf. so that 25A turns into a peak of 31.25A
at a pf of .5 its 50 Amps. So for brief periods in the cycle your
generator is having to supply 1.6x more current than it was designed to
handle.
So event though your physical real power is 3kW you need a generator
that can deal with 6kVA if you want to drive a load with a .5 pf.
Most of my measurements show that usually the pf of a bank of XOs is in
the .6 to .7 range. But there are cases when it can be down in that .5
range. This usually happens on 240 Vac when the XO power supply is only
very lightly loaded, say during suspend. So in that case even though
the power factor is very low the overall power draw is also very low.
How a generator deals with low power draw and low power factor is a
question for the manufacturer of the generator.
> AC grid-system, battery starts charged .. AC grid-system, battery
> starts empty ..
>
> solar+battery, battery starts charged .. solar+battery, battery
> starts empty ..
>
> generator system, battery starts charged .. generator system, battery
> starts empty ..
I think this should all be covered above but here's the summary.
Grid: Batt charged = 10W/hour;
Batt empty = 60Wh + ( 10 * (# hours XO is powered up))
Solar: Take the grid numbers and / .9
Generator: Same as grid but peak dynamic power must be considered or
generator may fail pre-maturely. For 50 XOs you need a generator that
can handle a reactive load of at least 2kVA. 23.5 W * 50 XOs / .6 pf =
1.95kVA. You should ask the mfg what happens if its operated with a
lower load but a pf of .5 .
> What number do you use as an average for battery life with XO-1s?
2.5 - 3 hours
> What number do you use for average battery charge time for XO-1s in
> AC-grid systems, with no laptop use?
2 hours. The LED will often stay lit for 2.5 - 3 hours but thats cell
balancing. After 2 hours of charge very little additional charge is
acquired by the battery.
> What number for average battery charge with AC-grid, with average
> laptop use simultaneously? (i'm sure there's some easy math here :)
When I talk to groups about laptop power I have a quote I use...
"If you ask a power usage question an you get back a single answer with
no qualifiers its wrong."
Sadly there is no easy math. Its all so situation dependent.
We don't have any good models for average use. That's why powerd on the
new builds has logging so we can begin to get numbers on that.
On your next trip you should probably purchase something like a
Kill-a-Watt meter or a WattsUP. That way you can take some in-the-field
measurements and close the loop on verifying my measurements and
assumptions.
http://www.p3international.com/products/special/P4400/P4400-CE.html
https://www.wattsupmeters.com/secure/products.php?pn=0
I use a WattsUP .net model because it supposed to be much more accurate
at measuring the power factor and it has a USB connection to the
computer where I log the readings every 10 seconds. It has memory so
you could also sample readings without a compute an then download them
later.
The kill-a-watt is much cheaper but it has no logging and no output to
the computer. You would need to periodically sample it for readings.
It does have a kWh meter but its only displays down to .01 kWh so you
need a fair bit of power draw to get a good average.
Both of them are rated to 15A so on a 120Vac setup thats 1800W. Thats
enough if you only have 50 XO's or so on that circuit but more than 65
XO's would start to get close to the limit.
--
Richard A. Smith <richard at laptop.org>
One Laptop per Child
More information about the Power
mailing list