See Missing/Duplicate Samples!

[Ulas]

When the clock extracting of bits from the CD is not synchronized with the clock stuffing bits into the DAC chip, the result is duplicate or missing samples.

Here is a 1KHz sine wave reproduced with synchronized clocks.

Here is the same wave when the CD clock is faster than the CD clock. The result is missing samples.

Here is the same wave when the CD clock is slower than the CD clock. The result is duplicate samples.

The clock speed difference shown is about 8%, which is greater than would be expected of two clocks with the same nominal clock rate, but it illustrates the point. Regardless of how well the two clocks are matched, if they are not synchronized, one will get ahead of the other resulting in missing or duplicate samples.

Of course, not everyone can hear distortion of that magnitude. Fans of the DDDAC don’t hear it. But that’s not surprising: The DDDAC outputs a 1KHz sine wave that looks something like this.

[SY]

The DDDAC waveform would seem to have a lot of third and fifth harmonic.

Out of curiosity, do you have spectra of the first three sine waves?

[ezkcdude]

Of course, not everyone can hear distortion of that magnitude. Fans of the DDDAC don’t hear it. But that’s not surprising: The DDDAC outputs a 1KHz sine wave that looks something like this. [/B]

How do speaker cones deal with such signals, not to mention amplifiers? Those waveforms have a lot of road to travel (and air) before they hit your ears. Have you modeled all that, too?

[Ulas]

Originally posted by SY
Out of curiosity, do you have spectra of the first three sine waves?

Here is a 1KHz sine wave reproduced with synchronized clocks.

Here is the same wave when the CD clock is faster than the CD clock. The result is missing samples.

Here is the same wave when the CD clock is slower than the CD clock. The result is duplicate samples.

The DDDAC outputs a 1KHz sine wave that looks something like this.

With all the extraneous garbage coming out of the TDA1543s it’s easy to understand why nobody hears the duplicate or missing samples.

Originally posted by ezkcdude
Those waveforms have a lot of road to travel (and air) before they hit your ears. Have you modeled all that, too?

What is there to model? Compare the first and last spectral plots. It’s simply a matter of GIGO, unless you expect the amplifier and speakers to some how remove the noise and leave the pure signal.

[ezkcdude]

Originally posted by Ulas
What is there to model? Compare the first and last spectral plots. It’s simply a matter of GIGO, unless you expect the amplifier and speakers to some how remove the noise and leave the pure signal. [/B]

Nope, you definitely got me there.

Although, in my defense, I didn't think the noise (distortion?) would be that great. In other words, if the noise of the DAC distortion were below that of the amp/speaker combo, then it wouldn't matter so much, right? Heck, I'm probably wrong about that, too.

[singa]

Ulas,
This is interesting,so you are saying synchonous clock/reclocking is anytime better than asynchonous reclocking and that low noise with spectrally clean circuits are required for best audio reproduction?(assumed)Given the current state of audio,do you have better solution?Please enlighten us,I'm asking for myself and not mocking you.

[Bernhard]

Ulas,

your cleanest plot shows a noise flloor of -70dB, right ?

Is that a 0dB sinewave ?

With any of my DACs I have a noise floor that is more or less determined by that of my spetrum analyzer, which is -94dB.

Although I think the 1543 was intended by Philips to be used in housewife players, the distortion looks a little bit unreal.

[macboy]

When the clock extracting of bits from the CD is not synchronized with the clock stuffing bits into the DAC chip, the result is duplicate or missing samples.

Obviously.
What's your point?
In other words, how or when would this ever happen? Clocks in any CD player are all synchronous to each other. The engineer would have to be severely mentally impaired to use separate clock sources for different parts of the circuit.

[singa]

Ulas,
Thank you for the reply,at least now I know where to devote my energy with regards to this digital problem.

[ezkcdude]

Ulas, I knew I had heard something similar to what you're saying. I remembered where just now, the data sheet for AD1896 (p. 18):

Asynchronous sample rate conversion is converting data from
one clock source at some sample rate to another clock source at
the same or a different sample rate. The simplest approach to an
asynchronous sample rate conversion is the use of a zero-order
hold between the two samplers shown in Figure 4. In an asynchronous
system, T2 is never equal to T1 nor is the ratio between
T2 and T1 rational. As a result, samples at fS_OUT will be repeated
or dropped producing an error in the resampling process.

Is that kind of what you are talking about?