[Power] Haiti power questions

[Richard A. Smith]

On 12/02/2010 04:22 AM, forster at ozonline.com.au wrote:
> Thanks Richard for a detailed analysis
>
> The power factor and harmonics of the switching regulators are a
> concern.
>
> The generator should be protected against power factor PF =
> watts/(Vrms*Irms) because it will have a circuit breaker that
> triggers on Irms

That's what I would have expected but we have received reports back of
XO's burning out generators.  I have very little more data than "we ran
it on a generator and it burned up."

I don't reallyt have the necessary tools to properly measure whats going 
on with a bank of XO's connected to a generator and good data on the web 
for operation of a generator outside of the .8 zone is scarce.

> but it will trip earlier than you would predict from the wattage. The
> harmonics could cause generator heating even though the breaker
> thinks the current Irms is OK. Derating by 50% may be
> overconservative or maybe not.

I don't have much experience in this area.  I've just been reading lots 
of app note and manufacturers recommendations.

Some real world tests in this area would be very welcome.

> Another concern is waveform distortion and poor voltage regulation
> and its effects on the electronics. Both harmonics and power factor
> could be a problem. Harmonics could create a spiky waveform that
> still is 110Vrms. Power factor can mess voltage regulation, I have
> seen inductive pf lower voltage and capacitive pf raise voltage. I
> have burnt out MOV's (surge protectors) in electronics when feeding a
> strongly capacitive load.

Interesting.  The power in most of the part of the world is really 
crappy.  We had the mfg of the XO-1 AC adapter try to make a front end 
that was resistive to crappy input.  Almost all of the failures of the 
XO-1 AC adapter that have been reported are physical mechanical failures 
rather than electrical so so they seem to have done an ok job.

OLPC Nepal actually cuts open failed supplies and fixes them and I think 
on some units they have reported evidence of excessive current inside 
the adapter.  I suspect that its from a brown-out condition.  But the 
vast majority of adapter failures so far are cable breakage.

Again. Experiments here with real data are welcome.  But most people 
don't have access to the necessary isolated instruments that you would 
need for proper measurements.

> You can either take a chance or to be conservative include some
> resistive load like incandescent lighting.

Yeah. But thats not what a generator powered deployment wants to hear. 
That means they have to add power draw to the system.  Also in most of 
the world with no power grid you don't get incandescent lighting.  Its 
all LED or CFL.  In many parts of the world you can't purchase 
incandescent lighting any more.

http://en.wikipedia.org/wiki/Phase-out_of_incandescent_light_bulbs

Just FYI.  OLPC is actively working on tying to make the power factor 
issue better.  For the 1.5 power supply (the brick supply not the wall) 
we have been going round, and round, and round with our power supply 
vendors.  The problem is that a cheap, efficient _and_ power factor 
corrected supply is almost an impossible situation.

The net result is that the upcoming 1.5 brick power supply overall is 
better but not by a whole lot.  We have a new possibility in the works 
right now.  Hopefully we can get it in place for 1.75.  All the power 
supplies are backwards compatible so once the 1.5 and 1.75 supplies are 
available you can use them on XO 1.

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