Audio path gain, was: Re: Bluetooth module

Al Johnson openmoko at mazikeen.demon.co.uk
Fri Feb 4 14:46:33 CET 2011


On Sunday 23 January 2011, Hrabosh wrote:
> Al Johnson píše v Pá 21. 01. 2011 v 10:34 +0000:
> > On Friday 21 January 2011, Hrabosh wrote:
> > > Hi guys,
> > > 
> > > yesterday I opened my freerunner to check whether mine FR is affected
> > > by R3004 issue [1].
> > > 
> > > It was, I put a piece of wire there instead of those two caps and
> > > volume got better!
> > > 
> > > I have also connected sine wave from generator to the WD8753 mic input
> > > and cheched it on the WD8753's output with the oscilloscope. It seemes
> > > that the audio codec chip gets into trouble (distortion, oscillations)
> > > if the overall audio path has higher gain than 1. I'm going to write
> > > more about this soon, with the scope pictures included (taken by my
> > > camera, I don't have DSO at home :-/ )
> > 
> > That would be interesting to see, but your summary seems wrong or perhaps
> > oversimplified. Distortion is inevitable if you overload an amp, and
> > oscillation isn't unusual immediately after overload. It isn't a function
> > of the overall path gain though. The wolfson has a chain of amps and
> > volume controls, and it only needs overload at one of them to introduce
> > distortion. Stick the gain too high on the first stage (mic input) and
> > the distortion won't go away no matter how far you turn down the gain on
> > later stages. This has been a long standing problem with the default
> > volume settings.
> 
> Yes, it is oversimplified. It is obvious that you can't fix distortion
> by lowering the gain of next-in-chain amplifier. You are just
> controlling the volume of already distorted signal.

If it was that obvious we wouldn't have had default settings guaranteed to 
cause distortion earlier than necessary in the first place. It seems it's only 
obvious to those of us with a grounding in analogue electronics, or those who 
spend time with guitar amps and similar.

> I put that picture from my scope with a short description to my web
> page, here:
> 
> http://hnet.endofinternet.net/elektro/Freerunner/ALSA/sound_setting.html
> 
> I was very surprised how LOW I had to set the audio controls to get rid
> of distortion:
> 
> Mic2: 0
> Sidetone: 1
> 
> Strange ...

How did you pick 0.5mV as a suitable input amplitude? Measurement of peaks 
when talking?




More information about the community mailing list