[Gta04-owner] Sound devices in the GTA04

Sat Dec 31 10:25:06 CET 2011

On Sat, 31 Dec 2011 09:36:56 +0100 "Dr. H. Nikolaus Schaller"
<hns at goldelico.com> wrote:

> Hi Neil,
> 
> Am 31.12.2011 um 05:34 schrieb NeilBrown:
> 
> > 
> > 
> > I often get confused by the myriad options that "alsamixer" present to me for
> > the sound card in the Freerunner and the Phoenux.
> > 
> > So having read through lots of docs and figured most of it out for the
> > Phoenux I thought I would write down what I know in the hope that 
> > a/ it will be useful to others, and
> > b/ if I am wrong about something, someone will tell me.
> > c/ it will force me to confirm that my understanding is complete.
> > 
> > 
> > There are 4 sound devices in the Pheonux - each wired directly to the OMAP
> > SOC (each to a different McBSP port - multi-channel buffered serial port).
> > 
> > Three of them are very simple.
> >   gta04voice - the sound to and from the GSM chip
> >   gta04headset - sound to and from the bluetooth chip
> >   gta04fm - sound to/from the FM transceiver (not in 3.2 yet)
> > 
> > They don't have any mixer controls - they just send and receive data in a
> > fixed format at a fixed rate (or maybe a small selection of rates).
> > 
> > The other is more complex:  "gta04" is provided by the TWL4030.  It has
> > some functionality that is not connected in the GTA04 so we should try to
> > hide it from ALSA so I don't keep getting confused.
> > 
> > The best starting point is the diagram on page 692 of the doco for
> > TPS65950 aka TWL4030 referenced in table 5.1 of the GTA04 manual.
> > It is figure 14-6.  It displays "Option 1".  "Option 2" doesn't make sense for
> > the GTA04A3 due to the connections available.
> > 
> > 
> > There are 2 input channels, 'left' and 'right'.
> > 
> > 'left' can be connected to:
> >  - on-board microphone                                 "Analog Left Main Mic"
> >  - headset microphone                                  "Analog Left Headset Mic"
> >  - AUX input line on alternate cable in headset jack   "Analog Left AUXL"
> > 
> > 'right' can be connected to
> >  - AUX input line on alternate cable in headset jack   "Analog Right AUXR"
> > 
> > Both channels have adjustable gain:  "Analog"
> > 
> > (For these to be accessable "TX1" and "TX2" must be set to "Analog" - we
> > should try to impose that in the kernel and hide those settings).
> > 
> > The output side is a bit more complex.  There are 4 digital channels allowing
> > quadraphonic sound!  Left and Right, Front and Back.  They can be mixed in
> > various ways with each other and with the amplified input streams and sent to
> > various output devices.
> > 
> > The 'front' channels are labelled '1' and the 'back' channels '2', so
> > 'Right1' is Front/Right.  To avoid confusion I'll stick with '1' and '2' as
> > those are the labels used in the ALSA device - just remember that '1' is the
> > primary (front) pair and '2' is the secondary (back) pair.
> > 
> > Some mixing and amplification happens digitally and then 4 digital signals
> > are fed into 4 DACs - DAC1 and DAC2 each with left and right channels - to
> > produce analogue signals.  Each of these has a fine-grained and a
> > course-grained digital amplifier before the DAC.  So a digital signal goes
> > through:
> > 
> > "DACn Digital Fine" then "DACn Digital Coarse" then becomes analogue and
> >    goes through "DACn Analog".
> > 
> > Each of these are stereo and with n={1,2} so we have 4 paths.
> > 
> > The hardware allows these 4 paths to be fed from various sources however
> > the current driver doesn't support switching and the obvious defaults are
> > used.
> > The hardware allows the '1' (front) paths to be fed from either 'front' or
> > 'back' of the appropriate side, or an L+R mix of either front or back (so 4
> > options).
> > The '2' (back) paths can be fed from the 'back' of the appropriate side
> > or a mix of L+R from the back (so 2 options).
> > 
> > The lack of driver support means that e.g. you cannot send a stereo signal to
> > the audio device and have the channels mixed and sent out the single speaker
> > of the GTA02 case.  So I'll probably have to fix that short-coming.
> > 
> > The 4 amplified analogue signals are optionally mixed with either the left or
> > right input channel (to provide a Voice Side-tone) in the main amplifier
> > stage.  These are controlled by the switches
> >   "Left1 Analog Loopback", "Right1 Analog Loopback"
> >   "Left2 Analog Loopback", "Right2 Analog Loopback".
> > 
> > There is then a 5th analogue amplifier which just takes the Voice Loopback
> > (sidetone) signal and amplifies it separately to the digitally provided
> > signals.
> > 
> > These 5 signals can be mixed or multiplexed into various output devices:
> > (The "hands free" speakers can just select one signal, the other outputs can
> > choose a mix of signals).
> > 
> > The "headset" (plugged in to 2.5mm jack) has a Left and a Right.
> > Each can receive a mix of 
> >  Voice,  DAC1,  DAC2
> > (where Voice is the side-tone loopback - a single channel - and DAC1 and
> > DAC2 are either the Left or Right as appropriate).
> > 
> > Then there are the speakers which is where I get confused again.
> > There can be 2 "HandsFree" speakers, and 1 "Ear" speaker - or maybe Earphone.
> > 
> > The HandsFree seem to be the 2 speakers in the GTA01 case, or just the Left
> > HandsFree is the single speaker in the GTA02 case.
> > 
> > The gta04 Schematic shows the Ear connections wired to something labelled
> > "CASEEAR" (Case Ear) but I cannot see where that appears on the board layout.
> > I would assume that this is just a future-expansion port, except that the
> > "sound-demo" script provided sets the volume for "Ear" but does not set the
> > input for "HandsFree".  so - confused.
> 
> This is the earpiece speaker. It is just two contact points on the PCB and is sitting
> in the front cover. So it is not a component to be soldered. Therefore, it is designated
> as a part of the case.

Oh, I get it - there are actually 2 speakers.  A little ear-piece speaker at
the 'top' and one (or 2) louder speaker nearer the bottom.  I never realised
that before!

Thanks.

> > So that is how it works in GTA04A3.  I've heard rumours that there is a
> > change for the A4 so I'll say something about that...
> 
> Yes, the direct voice path has been added. It is important to
> tri-state that or it will try to drive the PCM link but the GTM601 is the
> master.

If the GTM601 is the master, it presumably provides the clock - so it
probably drifts slight w.r.t the main system clock.  Maybe that explains the
'clicks' that Radek reports when using 'sox' to copy data.

> 
> So it is a three-way connection between the GTM601, the TPS voice
> channel and the McBSP.

So you either have the GTM601 talking directly to the McBSP, or have it
talking to the TPS voice, and have McBSP listening so you can record etc?

I guess that could work - I haven't delved quite that deeply into the various
interfaces yet.

> 
> 
> > 
> > The TWL4030 (aka TPS65950) allows digital connections directly to the GSM
> > Modem and the bluetooth controller.  I believe GSM is connected in GTA04A4 -
> > don't know about bluetooth.
> 
> Bluetooth is not connected directly since it shares the PCM clock with the voice
> channel.
> 
> > This allows the TWL4030 to be configured (Option 2) to forward audio between
> > the mic/speakers or the bluetooth controller  and the GSM Modem.  That way
> > the CPU can just set up the audio channels and let the phone call progress
> > without being involved - it could even go to sleep to save power.
> 
> Yes!
> 
> > Also the FM radio *could* be wired to the TWL4030 as the "AUX In" so you
> > could listen to the radio without needing the CPU to copy data around.
> 
> I am not sure how that could work since the Si4721 has even 6 lines for PCM
> (separate lock & frame sync signals).

But it can be configured for analogue in/out leaving the TPS to digitise if
necessary, or just route directly to the headset/speaker.

> 
> Unfortunately it is very difficult to do experiments on such a miniaturized
> device because you can't easily solder 4 or 6 wires... It needs to produce
> a new PCB variant.
> 
> > 
> > So I'm curious as to why bluetooth and GSM aren't wired directly to the
> > TWL4030 in the A3, and exactly what is wired in the A4.   Nikolaus??
> 
> The reason is simple: we did not understand what you have described
> very well when we designed the A3 board.

:-)

Thanks,
NeilBrown

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 828 bytes
Desc: not available
URL: <http://lists.goldelico.com/pipermail/gta04-owner/attachments/20111231/aa222a7a/attachment.bin>