If Bits are Bits

[Galliardist]

GO DIRECTLY TO JAIL.
DO NOT PASS GO.
DO NOT COLLECT $200.

You're late.
It's already pretty much worked out that the OP's specific CD player is better than his specific streamer, and that the difference may even be audible (though no evidence/formal blind test) as he is using them. No jail time needed on this occasion.

 

[voodooless]

No, it is totally correct, albeit a bit misleading in representation.

I wasn’t reacting to you though. I was reacting to this:

They are just plotting sample value over time using bar chart (can also call it histogram). So, the presentation is correct. It is supposed to be a straight line because the value does not change during that period.

Which is just wrong.

EDIT: As for the "unknown" thing, that isn't correct either, the analog waveform between sample points is known when proper sinc() filtering was used in the analog-to-digital conversion and thus can faithfully be reconstructed within the system bandwidth.

Yes, I mean unknown as in context of digital values.

EDIT2: "Therefore we cannot use staircases to represent digital audio samples." Again: the staircase plot is a perfectly valid graphic representation of a digital sample stream. It does not represent the corresponding proper analog waveform, though.

I have to disagree strongly here. The staircase representation leads to so many confusions and wrong conclusions that it should be avoided at all cost.

 

[Inner Space]

The staircase representation leads to so many confusions and wrong conclusions that it should be avoided at all cost.

The staircase representation is the last refuge of a scoundrel.

 

[JaMaSt]

it will show tiny steps or variations like zagged lines.

The map is not the territory.

 

[Galliardist]

Sorry Folks, i am late to attend to page 8, I am hoping to acquire soon a bryston bp 17 with the DAC included, then a streamer with digital only outputs. Then i can compare the CD player and a streamer using the dac input of the bp17. Other than Lindemann Limetree Bridge II streamer, what do you folks recommend.

Your choice is largely about what interface/control software you want to use. The Bridge II is outputting digital and should be OK. I'm sure some here would argue against it on price, but if you like the control app, why not stick with it.... beyond the device not being totally incompetent, the interface software is the most important thing here.

 

[DonH56]

Most any DAC holds its output value between samples, so the output does look like a stairstep, at the actual DAC output. The output anti-imaging filter is responsible for removing the HF content beyond Nyquist to generate a smooth waveform. For a DAC-in-a-box, you never see the stairsteps, because the output filter is inside the box before the output terminals. At that point, assuming proper filtering, the output is integrated and is one continuous (smooth) waveform with no stairsteps to be found. Zoom in all you want.

Mathematically the digital input to the DAC is better represented as a series of weighted impulses (or Dirac pulses), which in signal processing theory get applied to a ZOH (zero-order hold), and then filtered.

I think some confusion, at least mine, arises because what audiophiles consider a DAC is really much more than that to me. The block diagram of the DAC (box, component) you buy has a lot of components inside, only one of which is the actual digital-to-analog conversion part. Digital input capture/conversion and buffering, clock/data recovery and retiming, digital signal processing, the actual DAC, then output conversion to voltage (if needed) followed by filtering and analog output buffers. Plus power and control circuitry.

 

[antcollinet]

Most any DAC holds its output value between samples, so the output does look like a stairstep, at the actual DAC output. The output anti-imaging filter is responsible for removing the HF content beyond Nyquist to generate a smooth waveform. For a DAC-in-a-box, you never see the stairsteps, because the output filter is inside the box before the output terminals. At that point, assuming proper filtering, the output is integrated and is one continuous (smooth) waveform with no stairsteps to be found. Zoom in all you want.

Mathematically the digital input to the DAC is better represented as a series of weighted impulses (or Dirac pulses), which in signal processing theory get applied to a ZOH (zero-order hold), and then filtered.

I think some confusion, at least mine, arises because what audiophiles consider a DAC is really much more than that to me. The block diagram of the DAC (box, component) you buy has a lot of components inside, only one of which is the actual digital-to-analog conversion part. Digital input capture/conversion and buffering, clock/data recovery and retiming, digital signal processing, the actual DAC, then output conversion to voltage (if needed) followed by filtering and analog output buffers. Plus power and control circuitry.

I think you are talking about the dac chip.

A dac on the other hand is a complete analogue to digital converter. That has to include the reconstruction filter - else you don't have a complete conversion.

 

[DonH56]

I think you are talking about the dac chip.

A dac on the other hand is a complete analogue to digital converter. That has to include the reconstruction filter - else you don't have a complete conversion.

Read my last paragraph. I have designed DAC chips (among other things) so yes that is what it is to me, the thing that actually does the conversion process from digital to analog. BTW, a DAC is digital-to-analog(ue); analog(ue)-to-digital is an ADC.

 

[pjug]

Most any DAC holds its output value between samples, so the output does look like a stairstep, at the actual DAC output. The output anti-imaging filter is responsible for removing the HF content beyond Nyquist to generate a smooth waveform.

To be clear though, you should say that the stairsteps in a modern DAC will be much higher frequency than the music file sample rate (so not like shown on misguiding illustrations). And the heavy lifting on the filtering is done digitally so that the analog filter on the stairsteps can be trivial.

 

[DonH56]

To be clear though, you should say that the stairsteps in a modern DAC will be much higher frequency than the music file sample rate (so not like shown on misguiding illustrations).

That's what "HF content beyond Nyquist" means. The stairsteps usually show very fast edges and a ZOH so are accurate enough as shown at the DAC's output (the actual conversion, discrete or IC); but those edges (should) never reach the outside world.

And the heavy lifting on the filtering is done digitally so that the analog filter on the stairsteps can be trivial.

Only for oversampled architectures, like delta-sigma designs, which are the majority used today in audio. My world encompasses more than audio so I tend to think more broadly, though define a DAC more narrowly.

Sorry guys, I seem to have stepped into something more than my little pea brain can handle, y'all enjoy. - Don

 

[pjug]

Sorry guys, I seem to have stepped into something more than my little pea brain can handle, y'all enjoy. - Don

Ha ha, hardly. Or if you have a pea brain then mine is a grain of sand. Also nice pun on "stepped".

 

[Palladium]

You're late.
It's already pretty much worked out that the OP's specific CD player is better than his specific streamer, and that the difference may even be audible (though no evidence/formal blind test) as he is using them. No jail time needed on this occasion.

If bits are bits, then why do CD players all have different analog output stages?

Checkmate objectivists. /sarc

 

[solderdude]

Just in case someone takes this serious...

Because all manufacturers do things their own way for many reasons including not want to simply 'copy' someone else or like to add a twist to the recommendations from the DAC chip manufacturer or to be 'different' just to be different than others (sales brochure argument).
Also a 4x oversampling CDP needs a different post filtering than a 8x oversampling CDP.
Also a CDP only needs to filter 22.05kHz bandwidth (fixed) as opposed to an external DAC which is more complex to make properly.

In the end one should end up with a smooth sinewave on the output connectors. When this is not the case the manufacturer decided to do something different.

 

[KSTR]

I have to disagree strongly here. The staircase representation leads to so many confusions and wrong conclusions that it should be avoided at all cost.

You're free to contact the authors and ask them to replace it with a more correct pin plot, for the sake of mankind ;-)

 

[pkane]

You're free to contact the authors and ask them to replace it with a more correct pin plot, for the sake of mankind ;-)

This is the only correct way to show the actual waveform reproduced from its PCM representation

 

[KSTR]

This is the only correct way to show the actual waveform reproduced from its PCM representation

Sure (all competent software use proper sinc interpolation for the resulting waveform display), but the plot so many got offended by never was a plot of a reconstructed waveform, rather it was to represent the sample sequence in graphical form with the original (or reproduced) analog waveform in the same plot (note the arrow pointing to it), to recap:

"This diagram shows how an analogue sound wave can be represented with 24-bit 176,400 samples per second PCM encoding. The sample rate being higher than CD audio above allows for a greater representation across the X axis of this graph, whereas the higher bit-depth allows for the exact amplitude of the wave to be more accurately represented with each sample – the Y axis."

Bold mine, again.
I completely fail to see how anybody in the world could misinterpret the stair-case plot of sample values as showing an analog waveform... unless you completely miss what the verb to represent stands for, that is.

 

[Blumlein 88]

Sure (all competent software use proper sinc interpolation for the resulting waveform display), but the plot so many got offended by never was a plot of a reconstructed waveform, rather it was to represent the sample sequence in graphical form with the original (or reproduced) analog waveform in the same plot (note the arrow pointing to it), to recap:

Bold mine, again.
I completely fail to see how anybody in the world could misinterpret the stair-case plot of sample values as showing an analog waveform... unless you completely miss what the verb to represent stands for, that is.

And yet many, many people do misinterpret the stair case plots. In fact, almost everyone who doesn't understand what it represents seems to easily get hung up on it and it stays in their mind misleading them for years.

 

[pkane]

Bold mine, again.
I completely fail to see how anybody in the world could misinterpret the stair-case plot of sample values as showing an analog waveform... unless you completely miss what the verb to represent stands for, that is.

You apparently rarely visit the land of true audiophiles. This is misinterpreted all the time, along with the accompanying intuition that you need to sample more points in between the steps to smooth out the staircase, and, obviously, that’s why we need high sampling rate PCM

 

[Rednaxela]

You apparently rarely visit the land of true audiophiles. This is misinterpreted all the time, along with the accompanying intuition that you need to sample more points in between the steps to smooth out the staircase, and, obviously, that’s why we need high sampling rate PCM

Just like megapixels in digital photography.

 

[thewas]

You apparently rarely visit the land of true audiophiles. This is misinterpreted all the time, along with the accompanying intuition that you need to sample more points in between the steps to smooth out the staircase, and, obviously, that’s why we need high sampling rate PCM

Every time audiophiles misinterpret such Shannon's and Nyquist's dead bodies shortly switch from rotational energy state "0" to "1".