[Milkymist-devel] J3 redesign for rc4

Werner Almesberger werner at almesberger.net
Tue Jan 10 15:20:05 PST 2012


We're contemplating the fate of J3 for rc4. It's not terribly useful
in its present state and at its current location it also gets in the
way of improving J21. I've discussed our options with Joerg on
#qi-hardware:

- can we use LNLVLOUT{L,R} as audio output ?
  Wolfson documentation is remarkably vague on these pins, but it
  appears that they're just another line output, exactly like
  LINEOUT{L,R}. So the answer should be "yes".

- J3 reduction: use AUX* or CD* ?
  CD L/GND/R is suitable for a differential configuration, making it
  a little more flexible than AUX.

- given that we don't know what (if anything) will be connected to
  that audio input and thus can't optimize for any specific
  configuration (impedance, etc.), we just consider it "low-fi"

- Joerg suggests to add footprints for clamping Zeners on CD*

- R14 through R16 don't make sense, particularly considering that we
  may have a differential configuration. Instead, there should be
  small caps (~47 pF) for RF filtering.

- Wolfson recommend output DC blocks (C19, C21, C22) of 10 uF instead
  of the 1 uF we have. Joerg says 1 uF may be tolerable, but it might
  produce some small distortions in the codec's amplifier.

- Joerg also found L1 irritating. (Not for the first time :-)

My suggestions regarding J3 and the signals it connects to:

- eliminate J3 and add

  - a new Jx with HP OUT L, AUDIO_AGND, HP OUT R. Placement where
    it's convenient. Also rename HP OUT to something that doesn't
    suggest a strong amplifier. Maybe LINE OUT ?

  - a new Jy with CD L, CD GND, CD R. Placement where it's
    convenient.

  Jx and Jy would be unpopulated footprints for 1x3 100 mil (2.54 mm)
  headers.

- change R14 through R16 to small RF-filtering capacitors (~47 pF)
  May also become DNP (see below).

- change C23 and C25 to unpolarized, 10 uF

- maybe remove C24, which only adds confusion

- consider scrapping the AUX* and VIDEO* input paths. That's R2
  through R9, and C4 through C8. Maybe keep the DC block footprints
  (C4 through C8), so that one could solder something external
  (just in case).

- consider scrapping MIC2 (C14, R22, R25) and MONO OUT (C22, R19,
  C17) as well. Idem for keeping the DC block footprints.

- regarding the Zeners on CD* and considering that Wolfgang doesn't
  think anyone will actually make use of Jx and Jy, we may leave
  the task for adding suitable protection to the user of Jy.

Other suggestions for audio improvements, unrelated to J3:

- change R23 and R24 to small RF-filtering capacitors (~47 pF)

- consider increasing C19, C21, C22 towards 10 uF, if easy

- consider removing the small caps on output (C15, C16, C26 and C27)

- speaking of protection, perhaps we should add them to LINEIN* ?
  Not sure if Zeners with suitable capacitance could also act as
  RF filters, removing the need for separate small caps ?

- replace L1 with a "wire"

Not sure if we want to populate the input/output path to Jx and Jy.
Not sure what to do with SPDIF_{ENABLE,OUT}, if anything.

I hope I got everything. Joerg, thanks a lot for your help !

- Werner


More information about the Devel mailing list