Compact disc player sonic differences by various signal procesing devices before DAC

[Ken Newton]

-3.2dB at 20kHz is typical for NOS - this is the sinc function's response. It can be corrected in analog or digital form - the original SAA7220 has a smaller correction (for 4X and a bessel post-filter) built in to the digital filter. I agree with you pieter that its barely noticeable subjectively - perhaps my ears are getting too old

This is an interesting implementation detail of NOS DAC design. I'm 50 years old and find that I do notice an important difference when the -3.16dB @ 20kHz zero'th order hold (ZOH) response droop of CD rate audio is EQ'd flat, just as Thorsten has long recommended. Where I most immediately perceive the benefit of ZOH-EQ is in the metallic quality (in the positive sense) of cymbals and bells. Cymbals, bells, and brass instruments in general have a gestalt which seems exceedingly difficult for digital to convincingly reproduce.

I find soundstage 'air' not as affected by ZOH-EQ. Such 'air' seems more a function of midrange clarity, which NOS already tends to present in relative abundance.

 

[pieter t]

I am not sure what you have tested and remember, but as we have pretty good representation in the Benelux Area it should be to go and audition one of the products I designed, which include such a filter (optional). In fact, it may be fun to hear how you rate the Metrum DAC against them, but that is another story...

Well, the Metrum DAC has been reviewed in the US (6moons) and Great Britain (Martin Colloms) being able to stand up against DAC's 4-5 times the price, so when you point at the AMR DP777, yes that might be interesting.

 

[ThorstenL]

Hi,

Well, the Metrum DAC has been reviewed in the US (6moons) and Great Britain (Martin Colloms) being able to stand up against DAC's 4-5 times the price, so when you point at the AMR DP777, yes that might be interesting.

Try the CD-77 instead, the DP-777 is pretty good, but it is the "premier series", why aim low?

Ciao T

 

[abraxalito]

I'm 50 years old and find that I do notice an important difference when the -3.16dB @ 20kHz zero'th order hold (ZOH) response droop of CD is EQ'd flat, just as Thorsten has long recommended.

When you listened was the EQ applied in a digital filter or an analog one? Curious to know the implementation details.

Where I most immediately perceive the benefit of ZOH-EQ is in the metallic quality (in the positive sense) of cymbals and bells. Cymbals, bells, and brass instruments in general have a gestalt which seems exceedingly difficult for digital to convincingly reproduce.

Brass (particularly massed brass - such as found in the opening of Tchaik 4) is the biggest gain in accuracy I found when moving from SD to multibit. I simply can't find SD's rendition of brass convincing any longer

 

[Ken Newton]

When you listened was the EQ applied in a digital filter or an analog one? Curious to know the implementation details.

EQ circuit was an analog LCR tank in series with a passive resistor I/V, which is a topology that Thorsten has before advocated. The DAC chip was an AD1865. I trimmed the EQ network component values and achieved a surprisingly (to me) flat net response of within +/- 0.2dB up to 20kHz. As measured by FFT via an high-resolution PC soundcard with a stepped digital sinewave test signal.

 

[Ken Newton]

When you look at such impulse responses please look at them in spectral view. You´ll see that this pre and post ringing only is happening above the cutoff frequency of the lowpass-filter. Most likely no one can hear this.

I too, once supposed that to be so. I then began experimenting with a programmable digital filter which enabled me to cut-off the recording and mastering anti-alias SINC filter 'ringing' frequency energies, ala' Craven. This energy seems to exists on CD's in a narrow band, between about 19.5kHz and 20.5kHz in my experiments. Wonders, much of the annoying character of red book digital disappeared. The beneficial filter cut-off band was immediately obvious from listening. Cut-off above this narrow band, typical CD sound. Cut-off below that band, much greater ease, not unlike what one experiences with NOS (no digital-filter). I should add that while greatly beneficial, this experiment did not totally eliminate CD artifacts, just a majority of them, I'd say.

My hypothesis (more semi-educated guessing, at this point) is that the filter ringing energy, some of it ultrasonic but much of it not, is producing some unusual form of audible dynamic intermodulation products which don't necessarily show on static IMD tests. This notion is bolstered by the fact that I experienced the benefits of cutting-off the recording and mastering anti-alias filter ringing while using a common half-band linear-phase SINC function anti-image filter, itself having pre and post ringing. Not the minimum-phase, post ringing only, filter function advocated by Craven. Which may indicate that the root of the problem problem lie not so much in the ringing of any single filter, but in the ringng frequency interaction of multiple overlapping SINC filters.