pulse audio

[Len Ovens]

On Mon, December 19, 2011 5:33 pm, David Henningsson wrote:
> (Please don't respond in private if anyone can benefit from the answer)

I didn't realise I had, Sorry. I just hit reply without thinking.


> You did not give me your alsa info, so I'm assuming it looks like mine.
>
> The signal chain looks like this:
>
> Physical Mic Jack => Mic Boost control => Capture control => ADC
>
> So far I've seen only amplification on "Mic Boost", and most "Capture"
> controls being able to do both amplification and attenuation. On my
> computer, my Mic Boost goes from 0 dB to +30 dB (in 10 dB steps), and
> Capture goes from -16.5 dB to +30 dB.

Looks close enough, at least what alsa sees. I would suggest though, That
the physical layout is more like:

mic -> boost amp -> preamp (constant gain) -> capture (atten) -> ADC

Its just so much easier to do with any accuracy. Mixer boards are designed
this way... well at least they were in the seventies when I was getting
trained in such things. Digital mixers are of coarse all math.

>From the noise I hear in mine it seems that the boost amp is bypassed at 0
and may switch through 3 different amps for 10 20 and 30 db boost as the
noise signature (spectrum wise) seems different for each level. Mine also
has a mic monitor which seems to be prefade. The only way I could hear the
difference was by recording and playback. It is also possible that the
input impedance of the preamp is much lower than the mic boost as most
small cheap mics are high impedance and most line devices are low
impedance.

The devices I see with xubuntu oneric are not consistent with what I see
with Ubuntustudio 12.04 Alpha. There are two capture channels showing in
the 11.10 version and only one in 12.04a. Interesting. I will have to
install some recording software on the 11.10 disk and play with that.

> For mic inputs, you almost always want more than 0 dB in total. There is
> no attenuation. PulseAudio first tries to adjust "Mic Boost" to get at
> close as possible, then does the fine tuning in "Capture".

Yes, that is what it looks like. I think that the same thing could be
obtained by using the boost at the high end rather than the low end and
get a cleaner signal. Pretty easy to test... use pulse to set +40db, boost
will be 30 and capture at 10. Then try capture at 30 and boost at 10. You
may have to do this adjustment with alsamixer in a terminal while you are
recording. Then listen to the playback. On my machine the boost on 10 and
capture on 30 sounds way better with a much lower noise floor.

Right now, pulse moves the capture level up till one notch shy of 10db up
from -16.5 (-8.5?) and then hits boost 10 and goes back to capture -16.5.
I think it would be better to work from the top. Start with +60 (b30, c30)
and drop capture till (b30, c21), drop boost (b20, c30), then (b20, c28.5)
till (b20, c21)... then boost (b10, c30) so that by the time we get to +30
boost is all gone. The preamp after that is there all the time, but seems
relatively quiet. I don't expect to get the quality I get with mackie pres
and the delta66 (and there are better... but I have family to feed once in
a while), but It would be nice to be able to do some on the road audio
note taking without carrying all the gear and with the best quality what I
have can do.

The only problem I can see with having no boost for most of the range
might be matching with a high z mic. Maybe mine is a lower z mic than
some. Maybe high z and low z need some numbers :-) Professionals use
600ohm stuff except most mics are 150ohm (really low). Most consumer gear
has low z means 10Kohm. High z mics used to be 1Mohm, but many of the
electrets are lower. Electric guitars are 45K it seems to me. Quite a
range.

Noise is mostly voltage. High z inputs sense voltage which is both signal
and noise. Low z inputs expect the signal to be current and the noise
doesn't have much if any current. Balanced lines help a lot too (look how
well the old telephones do with miles of wire and no shielding).

I don't know everything by any means, but my training was in broadcast
electronics so audio should be one of my better areas... but I've been
wrong before...

-- 
Len Ovens
www.OvenWerks.net