is this why oversampling is not good

[bqc]

The bottom trace in the picture below shows the music that came out of the CD line out
of a Marrantz CD5000 which uses a TDA1549 with 4X oversampling.

The top trace came from patching into the I2S lines before
it gets to the TDA1549 and feeding them to the inputs of a
TDA 1543 Non OS DAC. The I/V conversion is done using an opamp per data sheet. Although the 2 traces are inverted with
respect to each other, one can see that the top trace is smoother
while the botton trace has jagged edges. Is this why oversampling sounds so harsh?
Unfortunately, I was not able to fully listen to the new NON OS sound since
my CD player starts messing up once I got everything hooked up.
It plays for a seconds, and stop for a couple seconds with static
noises and starts again for a seconds and stops again for a few
seconds with statics and over and over. I unhook all the patch
cables and put the CD player back to the way it was and it
still messed up. Anybody have any idea what could have gone
wrong? Is patching into the I2S caused the master clock to
become totally messed up? Do I need to build a new master clock?

Also when I build the I/V conversion using opamp per
spec, the sound is so soft. I had to replace the spec 1.2K feedback resistor with 66K to get any decent volume.
The Rbias is 3.3K. ANybody knows why the sound is so soft
using spec feedback resistor? Did the datasheet intend for the output
of the I/V opamp to be fed through another opamp for additional
voltage gain?

[haldor]

Looking at those waveforms I'd say the top traces indicate some serious high pass roll off going on. Either problems in the I-V or a really poor layout.

I think the differences between a competently implemented non-os and os DAC is not going to be enough to be grossly visible on your scope. Got access to a spectrum analyser? Hook it up and I expect you will see that the top traces have a very impared high frequency response. That may sound good for some kinds of music, but I prefer to use tone controls that I can adjust.

Phil

[fedde]

If you feed the I2S lines of your CD-5000 to a TDA1543, you'll not have a non-os DAC! The oversampling is in the controller, not in the TDA1549...

Fedde

[ojg]

Originally posted by bqc
Also when I build the I/V conversion using opamp per
spec, the sound is so soft. I had to replace the spec 1.2K feedback resistor with 66K to get any decent volume.

Do you also use a capacitor in parallell with the I/V feedback resistor? If so then you also need to reduce the value of this capacitor (by the same amount) otherwise you will get HF rolloff.

A 66k feedback resistor is way too much!!! The 1.2k as in the datasheet should be fine as this would give you 2.3mA*1.2kohm=2.76Vpeak output, I think the problem is somewhere else in your system...

[bqc]

I did parrallel the feedback resistors with caps per data sheet.
You are right that there must be something else wrong in my
system. I think the low level is caused by the fact that
the data in the I2S lines are 4X oversampled as Fedde pointed
out .

[rbroer]

BQC,

Read following thread completely:

Hint: I2S versus Jap. input format.