Pandaboard AOSP Audio input

[Sebbbi]

Hi, 

I found this thread where JBQ tells that audio input in AOSP is not working. This was in March.

https://groups.google.com/forum/?fromgroups=#!topic/android-building/kJ2eWcXTwZ0

I have exactly this issue with my 4.0.4 checkout.

I didn't try a newer checkout until now because for my needs 4.0.4 was all good. But now I'm faced with the fact that there is no audio input. 

Is this fixed in newer builds like the current AOSP master branch?

If not, are there any recommendations for a build which gives me working Audio I/O, HW Accel, and BT? (This are the basic featured needed for my application.) The last time I was looking around all the Linaro builds had also issues when at least one of these.

Any help is appreciated.

[kozak | Subbu]

Hi Sebbi,

On Wednesday, 5 December 2012 03:43:04 UTC-6, Sebbbi wrote:

Hi, 

I found this thread where JBQ tells that audio input in AOSP is not working. This was in March.

https://groups.google.com/forum/?fromgroups=#!topic/android-building/kJ2eWcXTwZ0

I have exactly this issue with my 4.0.4 checkout.

I didn't try a newer checkout until now because for my needs 4.0.4 was all good. But now I'm faced with the fact that there is no audio input.

I think this is something to do with the audio hal (audio_hwc.c), under device/ti/panda/. Can you please post (pastebin) your logcat on bootup?  I might have something working in a couple of days. Will share that if everything goes well :).

[Sebbbi]

Hi,

I did another checkout and compile yesterday, so I'm on 4.2.1 atm. Didn't try audio in until now, will check that today. I will switch back to 4.0.4 and give that logcat to you asap.

Thank you

[Sebbbi]

OK, here is the complete dmesg and logcat output:

http://pastebin.com/PUYzzfGi

http://pastebin.com/gi6D57FN

At the end i started Tape-a-Talk Pro and tried to record some audio. Dmesg output was

[  193.746582]  PDM-UL1: dsp: backend prepare failed -22

[  193.794342] twl6040-codec twl6040-codec: capture mode is not supported at 17640000Hz

[  193.803649] asoc: codec DAI prepare error

The interesting thing is, only every now and then when I hit the record button, this error appears on dmesg. When it appears I get to hear a short (approx. half second long) "noisy" beep on the Audio out.

I will jump to 4.2.1 now again, but if you need some more output let me know, I'll dd back that image.

 

Am Mittwoch, 5. Dezember 2012 19:40:10 UTC+1 schrieb kozak | Subbu:

[Sebbbi]

Hi,

thanks for your answer.

Yes, all my boards are rev B1. I just checked the schematic. It's definitely a Line IN on the B1, as the mic jack is routed to AFML/AFMR which is a Line In according to the twl6040 datasheet.

I'm not familiar with patching, how can insert that patch into my kernel?

Am Freitag, 7. Dezember 2012 08:21:40 UTC+1 schrieb Misael Lopez:

Hi Sebbi,

This error occurs due to an invalid constraint in the twl6040 kernel driver, this constraint basically limited 17.64MHz sysclk to headset playback only, but hardware wise this constraint doesn't exist, so it was removed from the driver. Please take a look at [1].

BTW, what Panda rev are you using? IIRC, there is a known problem with the headset mic in Rev EA1 and A1/2/3, since the differential input pins (HMICP and HMICN) are connected to separate channels in the microphone jack, instead of having HMICN connected to ground. As per schematics of Rev B1 [2], the audio input is more a Line In as no bias voltage is supplied, but should work fine.

Regards,

-Misa

[1] http://git.omapzoom.org/?p=kernel/omap.git;a=commitdiff;h=a3fd6da674306bb589376a48144adbca4b2bae7f

[2] http://pandaboard.org/node/223/#versions

[Sebbbi]

Hi,

ok, successfully patched the kernel, this worked so far. (No more errors on dmesg.)

Now my problem is that I don't get any signal recorded. Input is @ -100dB all the time. My suggestion is that android is catching the signal of the HMIC input (which is grounded). What do I have to change to make android listen to the  AFML/R inputs?

Soldering a mic to the primary microphone inputs is no option as the twl6040 is a BGA.

Am Freitag, 7. Dezember 2012 08:21:40 UTC+1 schrieb Misael Lopez:

Hi Sebbi,

This error occurs due to an invalid constraint in the twl6040 kernel driver, this constraint basically limited 17.64MHz sysclk to headset playback only, but hardware wise this constraint doesn't exist, so it was removed from the driver. Please take a look at [1].

BTW, what Panda rev are you using? IIRC, there is a known problem with the headset mic in Rev EA1 and A1/2/3, since the differential input pins (HMICP and HMICN) are connected to separate channels in the microphone jack, instead of having HMICN connected to ground. As per schematics of Rev B1 [2], the audio input is more a Line In as no bias voltage is supplied, but should work fine.

Regards,

-Misa

[1] http://git.omapzoom.org/?p=kernel/omap.git;a=commitdiff;h=a3fd6da674306bb589376a48144adbca4b2bae7f

[2] http://pandaboard.org/node/223/#versions

On Thursday, December 6, 2012 3:19:33 AM UTC-6, Sebbbi wrote: