[Trac #696] Power supply failure in a system.

[Zarro Boogs per Child]

#696: Power supply failure in a system.
----------------------+-----------------------------------------------------
 Reporter:  jg        |        Owner:  mfoster
     Type:  defect    |       Status:  closed 
 Priority:  high      |    Milestone:  BTest-2
Component:  hardware  |   Resolution:  fixed  
 Keywords:            |  
----------------------+-----------------------------------------------------
Changes (by mfoster):

  * status:  new => closed
  * resolution:  => fixed

Comment:

 Hi, Folks,

 To be complete, we took this failure very seriously, and assigned it the
 highest priority.  In addition to the battery removal tests noted above,
 voltage logging was also done on ten units for six cycles each during all
 phases of operation, including battery removal and insertion, power
 up/down, etc.  The logging results allowed us to completely characterize
 operation in all modes.  From this data, we were able to ascertain how far
 out of spec the capacitor was actually being used.  The voltage was found
 to be too high by approximately 20% for brief periods of time when the
 battery was unplugged while charging, with voltages of ~12V versus a 10V
 spec.

 To understand exactly how the degree of overvoltage found would impact the
 capacitor, we then collected an assortment of 30 different samples of the
 same capacitor, attempting to select them from different production
 batches.  These capacitors were then subjected to an intentional extreme
 overvoltage condition to measure the response to this far more serious
 stress condition.  For each capacitor, the voltage on the bench was
 increased until the cap failed; the failure voltage ranged from
 approximately 26V to ~31V, at which point the caps would overheat and/or
 burn.  This far higher voltage range strongly suggested that the initial
 failing unit contained a defective capacitor - subsequent sample return
 and consultation with the capacitor manufacturer confirmed that while the
 cap was operated slightly out of spec, the failed part itself had suffered
 from a material defect in manufacturing.

 As a consequence of this root cause determination, several changes have
 been made to the design of the charging electronics for the B2 build,
 including specifying higher voltage ratings for the critical capacitor, as
 well as redesign of the charging electronics to remove the voltage peak
 when the battery is removed while charging is occurring (this peak was in
 fact caused by following the charging IC manufacturer's reference design -
 that manufacturer has since accepted our proposed changes to the charging
 circuitry which eliminate the peak).

 We have high confidence that the revised circuitry design will eliminate
 any potential for overvoltage, and the increased design margins will
 further reduce the probability of any similar problems.

 Best Regards,
 MarkF

-- 
Ticket URL: <http://dev.laptop.org/ticket/696#comment:2>
One Laptop Per Child <http://laptop.org/>