#1735 BLOC Trial-2: Audio clicking on resume.

[Zarro Boogs per Child]

#1735: Audio clicking on resume.
----------------------+-----------------------------------------------------
  Reporter:  jg       |       Owner:  dilinger    
      Type:  defect   |      Status:  new         
  Priority:  blocker  |   Milestone:  Trial-2     
 Component:  kernel   |     Version:  Build 406.14
Resolution:           |    Keywords:  power       
  Verified:  0        |  
----------------------+-----------------------------------------------------
Comment (by cjb):

 Mitch gave me the beginner's introduction to our amp on IRC, in response
 to some basic questions:
 ===

 I would be tempted to scope the waveforms on pins 4 and 5 of U44 (the amp)
 at the instant of amp power-on.

 The voltage at the CODEC output is going to be about 1.5V with no audio
 signal, because the CODEC power supply is from 0 to 3.3V.  In order for
 the audio signal to swing both positive and negative, it has to be biased
 somewhere near the middle of the power supply range.

 In the steady-state case, the DC voltages on either side of the coupling
 capacitors C362 and C363 will equalize to some nominal value, but when the
 amp is powered off, the input bias voltage may be 0, so there would be a
 transient current through the capacitors when the input bias suddenly
 changes to something nonzero.

 If we can determine the amp input bias voltage in the case where the amp
 is powered off, we might be able to minimize the pop by making the CODEC
 send out close to that value, then immediately after EAPD on, ramp the
 CODEC output to match the ramp on the amp input.

 It will be complicated by the fact that the amp input is differential, so
 the ramp on C402 will need to be compensated.

 The other fly in the ointment is that I have asked Quanta to reduce the
 values of those coupling capacitors by a substantial factor, in order to
 roll off the useless low-frequency signals presented to the speakers.

 That will change the time constant of the pop waveform, and may, if we are
 lucky, reduce the pop energy in and of itself.

-- 
Ticket URL: <http://dev.laptop.org/ticket/1735#comment:31>
One Laptop Per Child <http://laptop.org/>