On 4/22/06, Carlos Rodrigues <[EMAIL PROTECTED]> wrote: > Right now megatec just calculates a linear approximation. It doesn't > give accurate values, but it is enough to have a fuzzy idea of the > charge left (like saying "it's almost noon" instead of "it's 12:45"), > and I'm not seeing how to plot a more accurate charge that fits all > the models that megatec is supposed to support (especially given I'm > no electrical engineer).
Don't get me wrong - I spent less than five minutes with the battery charge code for tripplite_usb.c (shamelessly ported from tripplite.c). I should have just said that a battery is more complicated than a linear "bucket of charge", and left it at that :-) I think that the solution that APC and MGE apparently use (probably others as well) on their PDC HID UPSes is better - do the charge calculation in an on-board microcontroller, and calibrate it as part of the final QA stage. If the manufacturer can't afford a couple extra cents for a microcontroller that can do that math, the components will probably drift over time anyway (making PC-side charge calculations futile). > Now, I've introduced these values to megatec, and I've commited it to > SVN. Can you (not Charles) check if they are acceptable now? > > On 4/22/06, Charles Lepple <[EMAIL PROTECTED]> wrote: > > On 4/22/06, lonely wolf <[EMAIL PROTECTED]> wrote: > > > The voltage doesn't look to have a linear dependency with the charge ... > > > > Here's the charge curve used by the Tripp-Lite driver: > > > > http://boxster.ghz.cc/projects/nut/browser/trunk/drivers/tripplite.c#L456 > > > > Generally speaking, measuring the voltage to find charge is a bit > > problematic unless you take into account the amount of current flowing > > in or out of the battery as well (a real-world battery has non-zero > > Thevenin resistance, so it is not a pure voltage source). The best you > > can do is plot voltages over time (as reported by the UPS), and try to > > fit the curve. > > > > -- > > - Charles Lepple > > > > _______________________________________________ > > Nut-upsuser mailing list > > [email protected] > > http://lists.alioth.debian.org/mailman/listinfo/nut-upsuser > > > > > -- > Carlos Rodrigues > -- - Charles Lepple _______________________________________________ Nut-upsuser mailing list [email protected] http://lists.alioth.debian.org/mailman/listinfo/nut-upsuser
