Audio buffer output

[Michael]

I have a multipart question for purposes of a self-tutorial.

Some of this "esoteric" audio information can drive you nuts and often
times, is contradictory.....hence, my questions.

Ok....so I have this little USB DAC chip that I've breadboarded......the
PCM2902 from TI.

It's really a neat device, taking my computer's WAV files through it's USB
port and ultimately outputing a stereo analog signal....all in one chip.

Works great.....and for now I just have the output feeding (2) 47uf
capacitors and some bleeder resistors right into my amplifier.

First of all, the chip shows a Vout peak to peak of .6v and an center
voltage of .5v and an impedance of 10k ohms.

(Q) Does that mean the output swings between .2v and .8v? I'm not clear on
that.



Secondly, I'd like to learn more about the "importance" of a buffer
stage.....("unity gain" or otherwise).

Many of the circuits using this chip are taking the output into high quality
opamps (analog devices and the like)....but I'm not clear on exactly "why".
(This chip sounds damn good with just the caps...right into the amp).
(and "maybe", less is better?)

(Q) Based on the .6v p-p figure (low?) and the 10k output Z -- would that
mandate a buffer stage? Is there some loading going on that is affecting
performance...(that I may not hear?)


(Q) Or are the opamps used to "tailor" the sound of the chip...using the
various opamps available and coming up with your own preferred "recipe"?


(Q) Or I suppose, a combination of the two?


Thanks kindly for any education on this....I really appreciate it.

 

[John Popelish]

Michael wrote:
8<

http://focus.ti.com/lit/ds/symlink/pcm2902.pdf

First of all, the chip shows a Vout peak to peak of .6v and an center
voltage of .5v and an impedance of 10k ohms.

(Q) Does that mean the output swings between .2v and .8v? I'm not clear on
that.

It is a bit confusing, but there are the letters, VCCCI below the
numbers, so I think that the .5 refers to half of whatever VCCCI is.
Likewise .6 ,means 60% of VCCCI.

So what it VCCCI?
According to page 2 (Maximums) it can't be more than 4 volts.
Near the bottom of page 5, we see it refers to the output of an
internal regulator that produces a nominal 3.35 VDC.
So, evidently, the DAC produces a signal based on this voltage, and
using half that voltage as a zero signal value. And that makes the
zero signal voltage a nominal 1.68 volts, and the peak to peak (full
scale) swing a nominal 2.01 volts.

The data sheet doesn't say what the output impedance is (I hate Texas
Instruments data sheets for stuff like this), but only that he specs
given apply with a load impedance of 10k attached. I would assume
that thew output impedance is something quite a bit lower than this
(but possibly not well controlled), so that if you keep the load at or
above 10k, it will work as speced. But they don't say this, explicitly.

Secondly, I'd like to learn more about the "importance" of a buffer
stage.....("unity gain" or otherwise).

I think this has to do with three things. One is the relatively high
output load impedance spec (compared to cable capacitive impedance at
the highest audio frequencies). If this thing acts strangely with
loads below 10k ohms, a long cable could do that. Any ordinary buffer
amplifier is likely to be able to drive a load with a lower impedance
than 10k.

The second thing is that there is not a very sharp low pass filter in
the chip to remove the digital hash from the DAC output, to keep the
device as general as possible. If you add some low pass filter to
your buffer, you can get a cleaner signal. So, you decide what
bandwidth you need, and include filtration that removes stuff higher
than that and include it in your buffer.

And then there is the issue of levels. If you require a signal with
peaks higher than +-1 volt, you will need to add voltage gain to your
buffer.

Many of the circuits using this chip are taking the output into high quality
opamps (analog devices and the like)....but I'm not clear on exactly "why".
(This chip sounds damn good with just the caps...right into the amp).
(and "maybe", less is better?)

(Q) Based on the .6v p-p figure (low?) and the 10k output Z -- would that
mandate a buffer stage? Is there some loading going on that is affecting
performance...(that I may not hear?)


(Q) Or are the opamps used to "tailor" the sound of the chip...using the
various opamps available and coming up with your own preferred "recipe"?

(Q) Or I suppose, a combination of the two?


Thanks kindly for any education on this....I really appreciate it.

 

[Michael]

So, evidently, the DAC produces a signal based on this voltage, and using
half that voltage as a zero signal value. And that makes the zero signal
voltage a nominal 1.68 volts, and the peak to peak (full scale) swing a
nominal 2.01 volts.

I'll remember that and for kicks I'll look at it on my scope.
I thought anything below a volt was too low. I need to make a chart with
datasheet "abbreviations" like VCCCI.....(I gave up when no one could make
their mind up on Vss and Vcc.)

The data sheet doesn't say what the output impedance is (I hate Texas
Instruments data sheets for stuff like this), but only that he specs given
apply with a load impedance of 10k attached. I would assume that thew
output impedance is something quite a bit lower than this (but possibly
not well controlled), so that if you keep the load at or above 10k, it
will work as speced.

Most amps inputs are usually higher so that's good.

The second thing is that there is not a very sharp low pass filter in the
chip to remove the digital hash from the DAC output, to keep the device as
general as possible. If you add some low pass filter to your buffer, you
can get a cleaner signal. So, you decide what bandwidth you need, and
include filtration that removes stuff higher than that and include it in
your buffer.

I think this was the concern I saw when it came to these opamps on the
output. I see the "stop band" attenuation is only -43db....pretty lousy, I
guess.

I'll just have fun playing with it......it sure sounds good to me....best
audio I've heard really, I'm hearing detail in some songs that I've never
heard before.....and I don't really "hear" any high freq digital noise
whatsoever.

So many audio nuts seem to design for test equipment rather than the
(limitations?) of the human ear...so maybe just raw caps will sound better
to me and that (should be) the bottom line.

I think when it comes to audio....it's all about how your brain processes
it.

John......thanks very much.

 

[JeffM]

(I gave up when no one could make their mind up on Vss and Vcc.)
Michael

Vss is the Source supply for a Field-Effect Transister (FET).
Vcc is the Collector supply for a Bipolar Junction Transister (BJT).
These specify not only the terminal,
but the technology involved as well.

Are you not finding this to be the case?

 

[Michael]

Vss is the Source supply for a Field-Effect Transister (FET).
Vcc is the Collector supply for a Bipolar Junction Transister (BJT).
These specify not only the terminal,
but the technology involved as well.

Are you not finding this to be the case?

Ok thanks.

I've run into documentation in the past that had IC's marked with Vss pins
as well as Vcc and Vdd used and even seen some newsposts as to what it all
means. (Of course, maybe the IC actually had an FET in it, so they
designated Vss?).

I've even seen an IC marked with Vss that was really a
ground pin.

Like I say, I'm going to make a list of what they "should be".

 

[JeffM]

I've even seen an IC marked with Vss that was really a ground pin.
Michael

That is quite normal. Look at some more spec sheets.

Having Vdd (Drain supply) and ground called out is more common,
but it's possible to have:
Vdd and Vss (both being non-zero potentials)
Vdd and ground
or
a negative-value Vss and ground.