I have just read the NOS threads and frankly dont undertand it.
I am considering buying an already modded CD player with NOS.
Clearly manufacturers put in oversampling for a reason, for the same reason there must be a down side to not having it, so in laymens terms what is that down side ?
(my concern is buying a NOS player and finding it wont play all my CD's)
I am considering buying an already modded CD player with NOS.
Clearly manufacturers put in oversampling for a reason, for the same reason there must be a down side to not having it, so in laymens terms what is that down side ?
(my concern is buying a NOS player and finding it wont play all my CD's)
Oversampling is in theory the best way to do it. Most of the criticism of it is based on misunderstanding (e.g. about the effect of the anti-alias filter on a square wave input to the ADC). Oversampling allows better filters to be used after the DAC.
NOS (with little or no filtering) is theoretically wrong, but some people prefer it. I suspect that the reason (apart from the above misunderstanding) is that NOS allows ultrasonic images above 22.05kHz to mimic the effect of transients which were necessarily removed by the anti-alias filter before the ADC.
NOS (with little or no filtering) is theoretically wrong, but some people prefer it. I suspect that the reason (apart from the above misunderstanding) is that NOS allows ultrasonic images above 22.05kHz to mimic the effect of transients which were necessarily removed by the anti-alias filter before the ADC.
I'm a laymen when it comes to the technical background of oversampling, but from my understanding it's quite simple: No oversampling means one stage less that potentially alters the signal. (Again: In laymens terms!)
Oversampling, done right, can solve certain issues in the digital route of the signal. But without it, you might get a signal that is perhaps "less perfect" (from a technical viewpoint), but "more real" (from a musical viewpoint). I've tested both, and since I got hooked on NOS-DACs there is no way back for me. Every DAC with oversampling I heard in the last few years just sounds less musical but more clinical than a well made NOS-DAC.
One idea: Ask the seller of the CD-player how the nos-mod is done. Pehaps it's just a wire bridge that can be desoldered to have oversampling again. This would give you the opportunity to listen to both versions and decide by yourself what you prefer.
Martin
Oversampling, done right, can solve certain issues in the digital route of the signal. But without it, you might get a signal that is perhaps "less perfect" (from a technical viewpoint), but "more real" (from a musical viewpoint). I've tested both, and since I got hooked on NOS-DACs there is no way back for me. Every DAC with oversampling I heard in the last few years just sounds less musical but more clinical than a well made NOS-DAC.
One idea: Ask the seller of the CD-player how the nos-mod is done. Pehaps it's just a wire bridge that can be desoldered to have oversampling again. This would give you the opportunity to listen to both versions and decide by yourself what you prefer.
Martin
NO. NOS menas one stage less that you need to REALLY reconstruct the signal corectly. The signal is NOT real, is distorted by lack of a clean brickwall analog filter. OS is the only way you can filter corectly that signal with a resonable analog filter.
All NOS modded players (by any DYI) have at the output tons of alias/image components that are NOT present in original recoding. Some people consider that an "improved" sound, call it "musical" - but it is just pain old distortion.
Of course, not all OS is equal. Some is done poorly, but that doesn't mean that OS is "bad". Just some people know how to do it right and some don't.
On the other side - NOS is bad always.
All NOS modded players (by any DYI) have at the output tons of alias/image components that are NOT present in original recoding. Some people consider that an "improved" sound, call it "musical" - but it is just pain old distortion.
Of course, not all OS is equal. Some is done poorly, but that doesn't mean that OS is "bad". Just some people know how to do it right and some don't.
On the other side - NOS is bad always.
Last edited:
As soon as someone says that he prefers "musical" to "clinical" I assume that he is not interested in reconstructing the signal immediately after the anti-alias filter, but merely hearing a pleasing sound.
Oversampling, done right, solves the problem of brick-wall filters which are difficult to do in analogue. OS lets you have a digital filter (which can be much more complex) followed by a simple analogue filter. The result is a closer approach to the original sound.
You will see NOS fans show pictures of ringing on a square-wave output from an OS system. What they usually don't tell you is that this ringing has not been added to the signal by the OS, but is the result of subtraction (of higher frequencies) before the ADC in the recording system- exactly what the anti-alias filter is supposed to do! This signal is actually on the CD, and OS reconstructs it faithfully. NOS smears the edges, because is has a sinc-type HF frequency droop and at the same time adds images above 22.05kHz which were never there in the original signal.
Oversampling, done right, solves the problem of brick-wall filters which are difficult to do in analogue. OS lets you have a digital filter (which can be much more complex) followed by a simple analogue filter. The result is a closer approach to the original sound.
You will see NOS fans show pictures of ringing on a square-wave output from an OS system. What they usually don't tell you is that this ringing has not been added to the signal by the OS, but is the result of subtraction (of higher frequencies) before the ADC in the recording system- exactly what the anti-alias filter is supposed to do! This signal is actually on the CD, and OS reconstructs it faithfully. NOS smears the edges, because is has a sinc-type HF frequency droop and at the same time adds images above 22.05kHz which were never there in the original signal.
DF96:
Could a linear phase (Bessel-type) analog filter be applied before the A/D conversion at the recording side? Then there were no ringing square wave signal on the CD, just a band limited signal with rounded edges.
Also, I edited a test CD wav file in a hex editor, that contains alternating 0000s and FFFFs, that is a perfect square wave signal. Playing back the test CD on a NOS DAC I get nice square wave, bot on an OS DAC I get ringing. Obviously, perfect square wave can not go to a CD through ADC, because it contains harmonics beyond 22.05 kHz.
Could a linear phase (Bessel-type) analog filter be applied before the A/D conversion at the recording side? Then there were no ringing square wave signal on the CD, just a band limited signal with rounded edges.
Also, I edited a test CD wav file in a hex editor, that contains alternating 0000s and FFFFs, that is a perfect square wave signal. Playing back the test CD on a NOS DAC I get nice square wave, bot on an OS DAC I get ringing. Obviously, perfect square wave can not go to a CD through ADC, because it contains harmonics beyond 22.05 kHz.
I think putting a square wave through a DAC is an unfair test - or at least points out that NOS DAC in question is not outputting the correct signal (iow - if its a perfect square wave then it's probably not being filtered).
Anyway, my take on it. I think NOS is easier to get good results - especially for a DIYer (and skimp on the output filter). Because the clock frequencies used are lower, the circuit is somewhat less susceptible to jitter - and the digital filter/oversampling chip is missing (which can add jitter and inject PS noise).
If done properly, I think OS is much better. In theory, and from experience.
Anyway, my take on it. I think NOS is easier to get good results - especially for a DIYer (and skimp on the output filter). Because the clock frequencies used are lower, the circuit is somewhat less susceptible to jitter - and the digital filter/oversampling chip is missing (which can add jitter and inject PS noise).
If done properly, I think OS is much better. In theory, and from experience.
Of course there is filtering, decimation and noise shaping on the mastering side of a CD. You won't have a square singnal on a real-life CD due to the format limitations.
On the other hand on a NOS system playing regular sinewave in audio domain will give you some "extra" sinewave components at the output generated by alias with no filtering at the output. That components have a bigger amplitude and sonic impact than the smal "ringing" that you tried to create artificialy.
BTW, you could buy SACD's and you will have NOS, with filtering at 50kHz audio signal bandwidth. Best of both worlds.
The recording side of a CD needs to pass signals up to as close to 22.05kHz as possible, while completely blocking all signals from 22.05kHz and upwards. A linear phase filter can't do this; it would need to start at a much lower frequency so you would lose audio HF. So you have to choose: throw away much HF signals, or keep them and suffer ringing on square waves? There is no way round this. OS at the recording end lets you get closer to the theoretical limits because digital filters can be smarter than analogue filters, but even OS can't defy mathematics.
Square waves (e.g. 000, FFF signals on the CD) are artificial so only create confusion.
Square waves (e.g. 000, FFF signals on the CD) are artificial so only create confusion.
Just as a point of interest, it would be quite interesting to get an explanation of why the OS filtering creates(evidently) the "ringing" in the digitally synth'd squarewave. Not something I have personally researched, so I don't know, myself.
I do believe proper OS is superior to NOS, but I have heard an NOS dac sound awfully damn good. On one that I had in my shop, a cheap & ack-dac, I had a 1kHz @ 0VU & -40VU that I'd previously burned, and checked out the waveform. Did not run it through a distortion analyser, but the waveform on the scope, even the -40VU tone, looked strangely clean. I'm sure a high frequency tone would have been a different story, but no such cd on hand at the time.
I do believe proper OS is superior to NOS, but I have heard an NOS dac sound awfully damn good. On one that I had in my shop, a cheap & ack-dac, I had a 1kHz @ 0VU & -40VU that I'd previously burned, and checked out the waveform. Did not run it through a distortion analyser, but the waveform on the scope, even the -40VU tone, looked strangely clean. I'm sure a high frequency tone would have been a different story, but no such cd on hand at the time.
A square wave consists of the fundamental frequency plus all odd harmonics, at a level equal to the inverse of their order. So it is, say, 1kHz at 1V, 3kHz at 0.33V, 5kHz at 0.2V etc. You can programme a spreadsheet to add up the series. If all harmonics are present you get a pure square wave. Now just use harmonics up to 20kHz (actually 19kHz, as they have to be odd). You will see ringing. This is not caused by adding anything, but because all the harmonics above 20kHz are missing. The filtering creates this ringing by removing the the harmonics which would normally cancel it i.e the ringing was already there in the signal before you filtered it, all you have done is make it visible.
This is quite different from the ringing you get from, say, an HF resonance even though they look superficially similar.
This is quite different from the ringing you get from, say, an HF resonance even though they look superficially similar.
OS needs two stages while NOS needs only one.
So NOS needs one stage less.
As you said it.
Yes. You can't reconstruct the signal before the anti-alias filter as you don't have enough information.
In case of confusion, the anti-alias filter is a brick-wall filter in the recording system which ensures that the signal going in to the ADC is band-limited - hence preventing aliases. People sometimes confuse this with the reconstruction filter in the CD player - this filters out images above 22.05kHz. It is easy to confuse aliases and images, as they are related issues.
In case of confusion, the anti-alias filter is a brick-wall filter in the recording system which ensures that the signal going in to the ADC is band-limited - hence preventing aliases. People sometimes confuse this with the reconstruction filter in the CD player - this filters out images above 22.05kHz. It is easy to confuse aliases and images, as they are related issues.
I really don't know what you are talking about.
How and at which point do you want to immediately reconstruct the signal after the adc filter ?
I guess you want to say, people do not care to filter after the DAC, so they hear a subjective pleasant sound.
What I am saying is that in the playback device it is not possible to recover (by reconstruction) the signal which originally appeared in the recording device just before the anti-alias filter. You seemed to be questioning this. Maybe I misunderstood your comment:
It is possible to reconstruct the signal which appeared after the anti-alias filter in the recorder. This what OS in the player does. NOS does not do this, as it omits the reconstruction filter.
It is possible to reconstruct the signal which appeared after the anti-alias filter in the recorder. This what OS in the player does. NOS does not do this, as it omits the reconstruction filter.
That OS ringing shown by NOS fans is preringing, precisely.
NOS has post ringing only.
Wether or not NOS has a sinc-type HF frequency droop or not, depends on the designers choice to compensate for it or not.
OS does have it too, but it is compensated in the OS filter, which means another tinkering on the signal with arithmetic operations that have finite resolution.
If images are not filtered, again this is the designers problem, not that of NOS.
By the way, most recent custom os filters are claimed to not show preringing and to sound more analog.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.