Document with answers to most popular battery-related questions is ready

Wolfgang Spraul wolfgang at qi-hardware.com
Sun Aug 2 21:15:31 CEST 2009


Paul,
one more thing...

> To me it seems that CC readings are almost unused except for
> presenting the user with a bit more accurate capacity data. And when
> someone is developing something lowlevel he could as well connect and
> external ampmeter, much more reliable and flexible approach.

This is where I'm not so sure. Also see Michal Brzozowki's mail.
I think real power consumption improvements in software can only be made if
it's throughout the whole stack, from kernel to middleware to apps.
I don't think you can optimize real system power consumption in the kernel
alone, where the kernel knows or correctly predicts all sorts of middleware or
app behavior.

So of course you first want the kernel situation, and charging, to be robust
and cover all cases. But then it should go on.
And I don't think we should assume every software developer has an ampmeter.
There are lots of application developers that could do valuable work if they
would get speedy and accurate readings on how much energy was consumed in the
last 5 minutes, 20 minutes, 2 hours, etc.

Now, if an accurate way to measure the current in the device solves 80% of that
as well, then the (big) extra cost of supporting CC batteries is not worth it.
So for me it all comes down to precision. Is it possible to build a cheap and
accurate way to measure current in the device?

For the NanoNotes, I will look into this first. If it's possible, this may be
all it needs to support application developers doing power consumption work.
If it's not possible, I think CC is good - I don't want to make the assumption
that you have to have an ampmeter to be able to do power consumption work.

Feedback very appreciated.
Wolfgang

On Sun, Aug 02, 2009 at 10:02:42PM +0400, Paul Fertser wrote:
> Wolfgang Spraul <wolfgang at qi-hardware.com> writes:
> > this is an excellent document, obviously I cannot spot anything
> > wrong!
> 
> Thanks for you kind words. :)
> 
> > I didn't know that some external chargers would not charge gta01 or gta02
> > batteries because of the thermistor check you are writing about.
> 
> I'm not sure i saw any charger like that myself or seen reports
> (except on our wiki). But i suspect the probability of that for nokia
> brand charges is quite high.
> 
> > All the cheap external chargers we bought in Taiwan or China for testing can
> > charge both gta01 or gta02 batteries without a problem.
> 
> Proves my point :D
> 
> > I have a question for all:
> > We will ship our first NanoNote with a BL-4C compatible battery, without
> > Coulomb counter (middle pin unused) [1].
> 
> Hm, can BL-5C fit there? Because it'd be much nicer as modern 5Cs have
> quite some capacity.
> 
> > How are people really using the Coulomb counter in gta02?
> > Theoretically I would think that it provides far superior power measurement
> > options for actual software development, just as you write.
> > For example when playing with power saving codes, whether in the kernel,
> > middleware or applications, I would think over the course of several hours or
> > days the Coulomb counter data is the primary means for efficient
> > development.
> 
> To me it seems that CC readings are almost unused except for
> presenting the user with a bit more accurate capacity data. And when
> someone is developing something lowlevel he could as well connect and
> external ampmeter, much more reliable and flexible approach.
> 
> > So we are considering shipping the next version of the NanoNote with Coulomb
> > counter batteries same as the Neo FreeRunner.
> > But if nobody is actually using the data from the Coulomb counter, then it's a
> > wasted effort.
> 
> I'm not sure you'll get much for using CC.
> 
> The only problem so far with using a dumb battery on FR is that
> there's no way to know the current from inside the device. On gta01
> there's a resistor shunt supposed to be used to measure the current
> but readings are too noisy to be useable. So if you have a good way to
> measure current already i'd not go for CC.
> 
> -- 
> Be free, use free (http://www.gnu.org/philosophy/free-sw.html) software!
> mailto:fercerpav at gmail.com



More information about the community mailing list