[sound]: capacitor that act like a high-pass filter and so removes the bass on the headphones jack

Andy Green andy at openmoko.com
Thu Nov 27 00:32:50 CET 2008


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Somebody in the thread at some point said:

| I have no clue why the capacitor is needed (or why it's needed at
all), but
| for me it works great without it. Maybe it is handled already inside
the IC?
| Is this possible?

This is basically the "DC coupling" scheme supported by the Wolfson
chip, but in their version they issue a reference voltage that is half
of full scale to be used instead of 0V.  This is done so that minimal
current is taken for silence or quiet sound... in 16-bit signed the
digital 0 level is "in the middle", so 50% of DC range.

When you have a capacitor in series, it blocks this constant DC and
"just passes on the changes", so no current flows at all through the
transducer when the thing outputs "digital 0 sample level".

Again if you use the 50% level as the reference no current flows when
the output is at rest at the 50% level either, both sides of the
transducer are at the same voltage.

But the way you're doing it the transducer is constantly passing a fair
amount of static power since at "silence" it is continuously ~1.5V on
one side of the 32R transducer and 0V at the other.  Only when we drive
- -32768 sample do we pass the minimum current, but probably still
something.  It's possible to disable the outputs in the driver, but it's
not simple figuring out when they're really not in use since the
analogue paths can drive them too for the telephony action.

| It would be great if there were no objections. Then everybody could do the
| HW fix for the sound issue, even without soldering skills and without
| risking damage.

It "works" but with the drawback explained above.

- -Andy
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org

iEYEARECAAYFAkkt3KIACgkQOjLpvpq7dMofUwCbBfWa61Bb6+eLCxZp4pIK+Ar5
JcwAn2PL7yLbuKzJmN5B+9/hqr/nk9uZ
=S87u
-----END PGP SIGNATURE-----



More information about the hardware mailing list