I think 192 may well be pointless, see Dan Lavry's white papers on his site.
And then the difference between 96k and 44.1 or 48k is a fair bit smaller than the difference between 16 and 24bit.
With DACs money thrown at it is not necessarily a deciding factor.
There used to be a post (now sadly gone) on another site where somebody posted otherwise identical recordings but one file was recorded through a cheap ART, one through a £350 Echo Audio AF12 while the third through a £8000 Apogee Rosetta.
While everybody of the dozens of people who tried to identify the three could easily identify the ART as the cheapest/worst the differences between the Echo Audio and the Apogee were way to close to tell. There were very, very subtle differences but slight majority preferred the Echo unit in the end if I remember correctly. Either way not enough to draw any conclusions about quality or accuracy between the two.
To my ears the differences between convertors is smaller than those between different amplifiers which leaves speakers/drivers as THE major determinant of sound quality.
I don't have a problem including the "cost no object approaches" ... just the inference that it would take stacks of cash to go toe to toe with a Passive system. That's absurd to me. (if Jack wasn't implying that, then I stand corrected)
However, getting the most out of a DSP is more than twisting a couple knobs. One can fail to produce satisfactory results with a cheap unit just as well as with the best on Earth if you don't spend some time with it. (thinking that there's more of this going on in this threads' posted opinions than anything else)
What I like about the active approach is that you aren't limited in driver selection. Combining drivers, that one wouldn't normally care to make the effort for, can be done with satisfactory results.
The guys with cradle to grave control over their components and/or intending to use low order xo's will, most likely, choose passive and have outstanding results. The additional complexity (manufacturer) or expense (avg. diy guy/*manufacturer wanting a simple audio solution) isn't worth it to these folks.
Getting the drivers to work together is the important part ... the rest is "cAke DeCoration". Sky's the limit for sure. .. but the users decision to use active/passive when assembling their system shouldn't include the notion of a second mortgage.
Yes, and sometimes active is cheaper... Particularly when instead of needing 1 ultra-high quality 300W/channel amplifier ($$$$) you can get away with one cheap, budget 150W/channel amp for the bass driver, and one homebrew 10-20W class A amp´for the highs, also taking advantage of no more power lost at the (passive) crossovers.
Really good passive crossovers are very expensive, btw. A DIY class-A discrete crossover wouldn't be so expensive to make, comparatively.
Well, because the analog part can be more expensive than the digital part, if it is to be realized correctly.
For example there's a lot (LOT) of things written on the net about the "horrible brick-wall filters", but when correctly realized they can sound allright. And to correctly realize one, it is an expensive affair (lots of precision inductors). The first digital multitracks had non-oversampling DACs coupled to classic brickwall (i.e. 9th order) filters and top mastering engineers of the day compared them favorably to their golden analog multitracks...
The key -as always- is on the implementation.
Last edited:
We're steering off-topic but since i've spent the last 3 years learning everything related to vinyl playback, let me answer shortly and as simple as i can:
On the inner grooves there is less linear groove speed available. Thus, the highest frequencies get more difficult to read by a conventional stylus. In consequence, most mastering engineers intentionally reduce the HF energy (i.e. by equalization) at the inner grooves, to prevent this. Thus, limited HF extension available. Or they choose to not do this, and the high frequencies at the inner grooves may be a bit harsh if they're not read with the narrowest elliptical (.2x.7 mil) *
Another choice is to significantly reduce the cutting level at the inner grooves.
Most of these choices will make the audio on the inner grooves different from the outer grooves, unless the record has been mastered with great care and your cartridge has a narrow-side-radius stylus, and your TT system is perfectly aligned.
* Harmonic distortion skyrockets at the inner grooves because of increased groove modulation curvature, which may reduce so much that the stylus does not correctly fit in the groove but ride up and down (pinch effect). This increases distortion.
Last edited:
Why off-topics, why needles and grooves? Because the topic was not well defined.
Let's still start from definitions.
1.) Active: design, build, fine tune the system, enjoy the music.
2.) Passive: Discuss on forums endlessly classifications of systems and of their parts, which one is better and why.
Conclusion: I prefer active.
Let's still start from definitions.
1.) Active: design, build, fine tune the system, enjoy the music.
2.) Passive: Discuss on forums endlessly classifications of systems and of their parts, which one is better and why.
Conclusion: I prefer active.
Well said Wavebourn. With my properly implemented DEQX, multi-amp system I have finally achieved EXACTLY what I wanted after years of randomly trying off the shelf passively crossed speakers. I am sure there are people out there who accidentally or technically hit on a perfect amp-speaker-room combination. I never did and all I ever wanted to do was listen to recorded music as close as possible to a real live event. I now have that and I have been EXTREMELY fussy in my pursuit
I think its flaming calculated rubbish to get you going again
1) and 2) are only different depending on who says it
I lost a rather lengthy post on this due to some network screwup, so this may be a bit brief.
A little known but pervasive limitation of parallel quantized PCM is the instantaneous waveform reconstruction accuracy which at 0.8 Nyquist is about 5%. This waveform reconstruction error phenomenon also does not reveal itself to continuous waveform distortion measurements of any competent digital audio system design. This is a major concern with A/D conversions in systems that perform waveform envelope detection in the digital domain where errors over a tenth of a percent or so are not acceptable. To a slightly lesser extent, this factor drives the oversampling ratio design of quality digital oscilloscopes.
Also, this error only decreases as a first order function of the Nyquist ratio, so to obtain 'traditional' 'digital distortion' figures of under 0.01%, the Nyquist ratio would have to be under 0.2%, or for a 20khz bandwidth system, about 40 hz. This error is independent of sampling word length and is purely a result of the actual sampling conversion rate.
I believe this is a contributing factor that has caused major transient, imaging stability, HF, and even midrange problems in redbook audio and, from the above description, even 192Ks/S parallel schemes would not really come close to eliminating it. Perhaps this is a reason that massively oversampled (at the actual A/D conversion) and 'one bit' digital audio has usually appeared to have an advantage in some areas such as 'smoothness'.
A little known but pervasive limitation of parallel quantized PCM is the instantaneous waveform reconstruction accuracy which at 0.8 Nyquist is about 5%. This waveform reconstruction error phenomenon also does not reveal itself to continuous waveform distortion measurements of any competent digital audio system design. This is a major concern with A/D conversions in systems that perform waveform envelope detection in the digital domain where errors over a tenth of a percent or so are not acceptable. To a slightly lesser extent, this factor drives the oversampling ratio design of quality digital oscilloscopes.
Also, this error only decreases as a first order function of the Nyquist ratio, so to obtain 'traditional' 'digital distortion' figures of under 0.01%, the Nyquist ratio would have to be under 0.2%, or for a 20khz bandwidth system, about 40 hz. This error is independent of sampling word length and is purely a result of the actual sampling conversion rate.
I believe this is a contributing factor that has caused major transient, imaging stability, HF, and even midrange problems in redbook audio and, from the above description, even 192Ks/S parallel schemes would not really come close to eliminating it. Perhaps this is a reason that massively oversampled (at the actual A/D conversion) and 'one bit' digital audio has usually appeared to have an advantage in some areas such as 'smoothness'.
Last edited:
VERY interesting info Thoriated. Im surprised that the conversion is at sub nyquist levels. Id imagine with my limited knowledge (comms from uni), that that would cause gross errors. 5% doesnt sound brilliant to me, even though i understand BERs of 0.1% are acceptable for telephony, i wouldnt want my hifi chain at that level.
Last edited:
is copper top testing our ears now
Yes, sort of. Back in the vinyl days, I was acutely aware of the inferiority of tracks on the inside tracks of LPs as they slow down; the words "harsh" and "grainy" spring to mind.
Yet High End people claim to be able to hear night and day differences between sample rates, or one DAC and another, yet never mention the fact that there is huge sonic degradation towards harshness and graininess as they listen to a side of an LP - just before they have to get up to turn it over.
If they can't hear it, how can they hear the difference between 192 and 44.1? If they can hear it, but it still sounds OK to them, what does that say for their judgement when recommending equipment to the rest of us?
Fair enough but do you really need to adjust to better than 0.010 milliseconds which is about 1 sample period at 96k?
It is passive approach, it is endless listening to judgments. Active approach is to try for yourself, and end the quest on your own decision, what sounds better, and what is different in cost of ownership only.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.