Talk:Neo Freerunner audio subsystem

From Openmoko

Revision as of 18:03, 19 March 2010 by JOERG (Talk | contribs)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

User:RuiSeabra said:

    • State: GSM <-> Built-in Handset (file gsmhandset.state)
      • Mic volume:

...

      • Control 63: 'Mic Sidetone Mux' MANDATORY VALUE "Mic 2"!!!
        • User:RuiSeabra's GTA02v5 with buzz fix has horrible static with this value, but using either "Left PGA" or "Right PGA" works very well.


That's very interesting, but - as elaborated to Rui in IRC - doesn't mean much as buzz isn't related to using PGA. Rather using PGA "detunes" the overall setup of mic path, by introducing several new parameters (PGA controls in digital domain of mixer), which Rui hasn't looked at, taken into account, or even finally listed all the relevant settings' values here. Furthermore there's nothing (except AGC, which is considered contraproductive for calypso's NR&AEC, and very unlikely Rui had a working setup for AGC by mere incident) you could implement by using a routing via PGA, rather than just adjusting #5 (#48, #12. In that precedence) to compensate for pathgain changes done by PGA.

Finally buzz is basically completely unrelated to PGA (if not any mixer volume settings) at all

2009-12-01 jOERG--Openmoko HW development and synergy

it's not buzz fix, is being clearly heard on the other side

This is not buzz related (my phone has the buzz fix).

  1. With Mic2:
    1. Calling myself from a Nokia phone has loud static, and background noise can easily make it hard for remote end to hear me correctly
  2. With Left or Right PGA
    1. Calling myself from a Nokia phone has almos no static at all, and background noise is dampened enough so that even at a noisy restaurant the remote end could hear me about as normally as with any other phone

With no electricity or sound engineering background, I can't explain this other than in these terms:

  1. Mic2 == awfull sound on remote end
  2. Left|Right PGA == normal sound on remote end

There may be better solutions, but while they are not found, can we at least provide a working sound setting?

You're 'fixing' clipping in the mono mixer

The loud static is the result of an overloaded amp stage somewhere in the chain. Routing through the PGA will reduce the signal slightly, reducing the distortion or removing it. The problem is probably occurring in the mono mixer as the default setting (7) has +6dB gain. Reducing it to 5 should help as it will then have no gain.

Empirical Data for Mic Settings

Wouldn't it be better to insert previous settings into the table, instead of just saying "it's better now"? --Fabian Schölzel 16:55, 17 March 2010 (UTC)

Never mind, thats what "Microphone Settings in Use" is for. --Fabian Schölzel 16:58, 17 March 2010 (UTC)


I'm so tired to read reports following the scheme "Car's motor didn't start. After fiddling with almost any of the 90 adjustment dials in carburator and ignition and 7 other domains, I finaly found it started when I used a grip to turn the key". Two days later we have to deal with user B, C and D wining "I tried the 'turn key' method. Doesn't work for my car :-(( (even took a hammer to adjust some random dials)". And probably 2 app devels nagging kernel devels about "I implemented proven good turn-key sound improvement method. User said fails for him. FIX THAT!"

Once again: left/right PGA *might* work (as well), but it's introducing 10 unknown variables to the whole process, which are set "correctly" by mere incidence *if* it works for you, and there's absolutely NOTHING that couldn't be done _the right way_ as well, rather than by random switching of dials

*SIGH*

2010-03-19 17:49 jOERG

Personal tools

User:RuiSeabra said:

    • State: GSM <-> Built-in Handset (file gsmhandset.state)
      • Mic volume:

...

      • Control 63: 'Mic Sidetone Mux' MANDATORY VALUE "Mic 2"!!!
        • User:RuiSeabra's GTA02v5 with buzz fix has horrible static with this value, but using either "Left PGA" or "Right PGA" works very well.


That's very interesting, but - as elaborated to Rui in IRC - doesn't mean much as buzz isn't related to using PGA. Rather using PGA "detunes" the overall setup of mic path, by introducing several new parameters (PGA controls in digital domain of mixer), which Rui hasn't looked at, taken into account, or even finally listed all the relevant settings' values here. Furthermore there's nothing (except AGC, which is considered contraproductive for calypso's NR&AEC, and very unlikely Rui had a working setup for AGC by mere incident) you could implement by using a routing via PGA, rather than just adjusting #5 (#48, #12. In that precedence) to compensate for pathgain changes done by PGA.

Finally buzz is basically completely unrelated to PGA (if not any mixer volume settings) at all

2009-12-01 jOERG--Openmoko HW development and synergy

it's not buzz fix, is being clearly heard on the other side

This is not buzz related (my phone has the buzz fix).

  1. With Mic2:
    1. Calling myself from a Nokia phone has loud static, and background noise can easily make it hard for remote end to hear me correctly
  2. With Left or Right PGA
    1. Calling myself from a Nokia phone has almos no static at all, and background noise is dampened enough so that even at a noisy restaurant the remote end could hear me about as normally as with any other phone

With no electricity or sound engineering background, I can't explain this other than in these terms:

  1. Mic2 == awfull sound on remote end
  2. Left|Right PGA == normal sound on remote end

There may be better solutions, but while they are not found, can we at least provide a working sound setting?

You're 'fixing' clipping in the mono mixer

The loud static is the result of an overloaded amp stage somewhere in the chain. Routing through the PGA will reduce the signal slightly, reducing the distortion or removing it. The problem is probably occurring in the mono mixer as the default setting (7) has +6dB gain. Reducing it to 5 should help as it will then have no gain.

Empirical Data for Mic Settings

Wouldn't it be better to insert previous settings into the table, instead of just saying "it's better now"? --Fabian Schölzel 16:55, 17 March 2010 (UTC)

Never mind, thats what "Microphone Settings in Use" is for. --Fabian Schölzel 16:58, 17 March 2010 (UTC)


I'm so tired to read reports following the scheme "Car's motor didn't start. After fiddling with almost any of the 90 adjustment dials in carburator and ignition and 7 other domains, I finaly found it started when I used a grip to turn the key". Two days later we have to deal with user B, C and D wining "I tried the 'turn key' method. Doesn't work for my car :-(( (even took a hammer to adjust some random dials)". And probably 2 app devels nagging kernel devels about "I implemented proven good turn-key sound improvement method. User said fails for him. FIX THAT!"

Once again: left/right PGA *might* work (as well), but it's introducing 10 unknown variables to the whole process, which are set "correctly" by mere incidence *if* it works for you, and there's absolutely NOTHING that couldn't be done _the right way_ as well, rather than by random switching of dials

*SIGH*

2010-03-19 17:49 jOERG