#436 HIGH Trial-3: Battery charge estimation non-linear.

[Zarro Boogs per Child]

#436: Battery charge estimation non-linear.
-----------------------+----------------------------------------------------
  Reporter:  dwmw2     |       Owner:  David.Lin
      Type:  defect    |      Status:  assigned 
  Priority:  high      |   Milestone:  Trial-3  
 Component:  hardware  |     Version:           
Resolution:            |    Keywords:  power    
  Verified:  0         |  
-----------------------+----------------------------------------------------
Comment (by David.Lin):

 In EC PQ2C20, I have put new GP algorithm to increase the

 preciseness of the charge estimation.



 The *capacity_percentage* comes from SOC counter,

 and note that the SOC estimation will be the SOC estimation.

 it's nature the bias will grow with cycle times and be sensitive to
 ambient temperature.



 I think that's not easy to indicate much more accurately in short term,

 because of the deisgn architecture and the native chemical attributes:

 1)We only have half-function *HARDWARE* gauge IC on battery like most
 conventional laptops,

   which doesn't record the characteristic charge pattern for battery cell,

   thus it becomes tough on SOC counter to have a general pattern to fit
 different kind of cell.

 2)EC does a calibration only when battery's full charged(ex:85% for Ni-MH)
 or

   completely discharged(ex:7% for Ni-MH).So if the bias exists in SOC
 estimation, the user space find that

   until EC calibrate when reaching the high or low limits. That's why we
 see the jumping, because a calibration occured.

-- 
Ticket URL: <https://dev.laptop.org/ticket/436#comment:22>
One Laptop Per Child <http://laptop.org/>