DAC NOS vs DELTA SIGMA, Who is the real winner...

[watchnerd]

Audible?

It’s been decades since it was studied, possibly not, but why bother caring when oversampling moves the issue octaves higher?

 

[watchnerd]

From Chatgpt, to the question : what are the benefits of NOS DAC

NOS DAC stands for Non-OverSampling DAC. Unlike traditional DACs (Digital-to-Analog Converters) that perform oversampling to improve resolution and reduce noise, NOS DACs process the digital signal at its native sampling rate without any interpolation or oversampling. This design philosophy can offer some unique benefits and a distinct sound profile, particularly appreciated by audiophiles. Let's dive into the benefits of a NOS DAC:

---

1. More "Analog-Like" Sound​

• NOS DACs are often favored for their ability to maintain a more natural, organic sound that some listeners find more "musical" compared to oversampling DACs.
• By skipping the oversampling and interpolation processes, the signal retains its original timing and waveform, which can result in a sound that feels more immediate and authentic, closer to the source material.

2. Simplicity and Purity​

• The design of a NOS DAC is often simpler because it doesn’t require the additional circuitry needed for oversampling or digital filters.
• Less processing can mean less potential for digital artifacts or coloration, and a cleaner, more direct conversion from digital to analog. This purity can be appealing to those who prefer the minimalist approach.

3. Lower Distortion​

• Because NOS DACs don’t use oversampling, there’s less risk of distortion introduced by filtering at high frequencies. Oversampling DACs often use steep digital filters to shape the output, which can introduce artifacts like ringing or phase distortion.
• In a NOS DAC, the signal is passed through more directly, with fewer stages of digital processing, which some listeners feel results in less artificial distortion.

4. Retains the “Original” Digital Signal​

• NOS DACs can offer a more accurate representation of the original recording’s signal because they don’t interpolate between the digital samples. This means there’s no attempt to artificially increase the sampling rate or fill in the gaps between samples, which some believe can alter the sound.
• Some audiophiles find that the natural timing and rhythm of music is better preserved in a NOS DAC, as it doesn’t overcompensate for digital data gaps, as oversampling DACs might.

5. No High-Frequency Filtering​

• Oversampling DACs typically introduce high-frequency digital filters to eliminate high-frequency noise. These filters can sometimes cause unwanted artifacts (like "pre-ringing" or "post-ringing" effects), which can color the sound.
• NOS DACs, by contrast, bypass these high-frequency filters, often resulting in a more "natural" and less “processed” sound.

6. Warmth and Musicality​

• NOS DACs are often described as having a warmer, smoother sound compared to oversampling DACs. Without the processing of high-frequency signals, some listeners find that the result is more pleasing, especially with older, highly compressed, or less-than-ideal digital recordings.
• Some people prefer this type of sound because it softens digital harshness, making the music feel more natural or analog-like.

7. Potential for Fewer Digital Artifacts​

• Oversampling DACs introduce multiple stages of digital processing that can sometimes create digital artifacts, like "glare" or a cold, clinical sound. The simplicity of a NOS DAC can lead to fewer of these issues, which some listeners appreciate for long listening sessions.

8. Compatibility with Older Recordings​

• NOS DACs are often more forgiving when playing older, lower-quality digital recordings or lower bitrates. The simplicity of not upscaling the signal means that the DAC’s sound remains closer to the original recording, which can be beneficial for those listening to older, less-than-perfect recordings that don’t benefit from the oversampling process.

---

Trade-offs to Consider:​

While NOS DACs offer these potential benefits, there are also some trade-offs to keep in mind:

1. Higher Nyquist Noise: Since the DAC doesn’t oversample, higher frequency noise from the digital signal may be more audible, particularly with lower-quality sources.
2. Potential for Aliasing: Without the oversampling process to "smooth" the signal, aliasing (unwanted artifacts at lower frequencies) can occur if the DAC isn’t carefully designed.
3. Less Precision: Some listeners may find that NOS DACs offer less "precision" or "detail" in the sound compared to oversampling DACs, especially in more complex musical passages.
4. Higher Frequency Roll-off: NOS DACs can exhibit a roll-off at higher frequencies due to the lack of oversampling, which some listeners might interpret as a loss of detail in treble frequencies.
5. Limited Support for High-Res Audio: Since NOS DACs don’t oversample, they’re typically more suited for CD-quality (44.1kHz) audio rather than high-resolution files (96kHz or 192kHz), though some higher-end NOS DACs attempt to handle these files with care.

---

Conclusion: Why Choose a NOS DAC?​

In essence, NOS DACs appeal to listeners who prefer:

• A more natural, musical, or "analog-like" sound.
• Simplicity in design, leading to a potentially purer, less processed signal.
• A warm, smooth tonality with reduced harshness or digital glare.
• Old-school audiophile sensibilities, where the focus is on maintaining the integrity of the original recording without adding too much digital enhancement.

If you prioritize musicality, warmth, and a more vintage or natural sound, a NOS DAC might be a great choice. However, if you prefer detail, precision, and extended high frequencies, you may lean toward an oversampling DAC or a delta-sigma DAC.


Would you like recommendations for specific NOS DACs, or examples of models that have been particularly praised for their sound quality?

How do I downvote?

 

[wwenze]

How do I downvote?

Nah we usually just look at likes ratio.

 

[raif71]

How do I downvote?

why not ask chatgpt ?

 

[Scytales]

Ok, I am not an electronist but I can understand principles and measurements.
My belief is hearing ( I can still gear 16000 Hz ) and testing.
I have a favorite playlist of songs,mostly jazz, rock and folk and throughout the years I listened them on diferent equipment and:
I like:
- CDs and SACD
- class A ( Nelson Pass builds ) and A/B amps
- tube preamps not amps
- TDA1541 dacs

dont't like for audiophile listening:
- class D amps
- class H amps
- Delta Sigma dacs
- new generation of solid state preamps

I just want to point out the contradiction between the two statements I have underlined in bold types: the DSD modulation that is recorded on SA-CD is the product of a delta-sigma modulator (be it hardware or software). As some SA-CDs are produced from PCM source materials, the modulation of said material in DSD to make SA-CDs can in some ways be viewed as the first step of a delta-sigma DAC, the end steps being implemented in SA-CD players.

 

[Svperstar]

Just saw the ChatGPT output.

When I owned a NOS DAC, it didn't sound natural at all. It sounded terrible. My $40 Discman I used in High School sounded better then the $650 NOS DAC I owned.

Not surprised a NOS DAC is the lowest performing device ever measured on this site.

 

[KSTR]

NOS is not the issue per se, filterless + NOS is the problematic combination.

 

[VientoB]

NOS is not the issue per se, filterless + NOS is the problematic combination.

Does NOS with a filter not slope off the high frequencies?

 

[wwenze]

Does NOS with a filter not slope off the high frequencies?

NOS with a filter will low-pass-filter away the high frequencies just fine. Like what it is supposed to do. Like how they were originally done. Like how they are still being done - People forget that "NOS" DACs came before SD DACs and are still being used for high-frequency stuff today.

In fact NOS (or not, or SD or not) and the need of a filter are two separate things and you need a filter regardless of whatever DAC. It's a mathematical requirement to add the filter to avoid aliasing.

Note the frequency scale in kHz. Thanks Archimago for this nice measurements.

Ultimately, NOS is just a mathematical concept, not a technology. You can run anything NOS or OS, you can run an R2R DAC OS or NOS, run a SD DAC OS or NOS (after people debate the definition). And people even worry about it because they do not understand the words after reading the two audiophile sentences they get from Google.

 

[danadam]

NOS with a filter will low-pass-filter away the high frequencies just fine.

He could have meant the roll-off from zero-order-hold reconstruction. Though that happens regardless of the filter. Without oversampling you need either pre-eq or post-eq to mitigate that roll-off.

Or maybe he was wondering how easy, or not, it is to build an analog anti-image filter that is steep enough to not roll-off those frequencies?

 

[wwenze]

I see. That has to happen with any sampling rate. When you put a -3dB point at 20kHz... you get a -3dB point at 20kHz. But with a higher sampling rate like 88kHz you only need to LPF away anything above that frequency so the -3dB point can be set to 88kHz or lower hence leaving the audible frequencies untouched.

Again, not an issue with NOS or not. And can be avoided by using higher sampling rates and the appropriate filter, even when you're operating in NOS. The only key issue in practice is that in real life these DACs or at least the commercial finished products will have the filter frequency aimed at 20kHz in order to cover for the worst-case scenario which is the lowest sampling rate of 44.1kHz and this filter works independently of the sampling rate and cannot be changed so you get the droop regardless.

So what the current NOS DAC owners can and should do are:
- If you own a DAC with a filter that can be configured like in post #90, just run it at a higher sampling rate, either using the DAC's own oversampling or your source's oversampling; realistically it would be a combination of both
- If you own a DAC without a proper output filter... yea they sell those... you should run the DAC at a higher sampling rate. Referring to post #89, notice with 88.2kHz sampling rate (converted by source) there is no image between 10kHz and 78.2kHz. You still get the images from not having a filter, but at least there's fewer and at a higher frequency.

 

[ttdtechguy]

Hello Everyone,
I've always been a fan of the well measuruments Delta-Sigma DAC and that they sound neutral to my ears. I currently have a DO300 SMS. I also tried some R2R DACs, such as Gustard R26 (NOS and Oversamplig mode) and HIFIMAN EF400. The latter seemed a little "colourful", while the R26 didn't seem very different from my D0300.

Reading the theories you would find that a NOS DAC would be preferable because of many things that in practice are never achieved, unlike the Delta Sigma which require oversampling.

How are things really?
Thank you all.

I loved my R26, thought it was fantastic. Better than any sigma fab, I find them very hard to listen too for any length of time

 

[danadam]

I see. That has to happen with any sampling rate. When you put a -3dB point at 20kHz... you get a -3dB point at 20kHz. But with a higher sampling rate like 88kHz you only need to LPF away anything above that frequency so the -3dB point can be set to 88kHz or lower hence leaving the audible frequencies untouched.

I'm not sure if that was clear, so just in case it wasn't: the roll-off from zero-order-hold reconstruction is a separate issue from roll-off from analog anti-image filter. But yes, they both benefit from DAC's oversampling (or hi-res source in case of NOS DAC).

In the graph above it was just a simulation and I used a steep digital anti-image filter, so the graph only shows roll-off from zero-order-hold reconstruction. In other words, if your source is 44.1k and you have NOS DAC, then the blue graph is the best you can ever expect but may be worse if the analog filter is shallow.

So what the current NOS DAC owners can and should do are:
- If you own a DAC with a filter that can be configured like in post #90, just run it at a higher sampling rate, either using the DAC's own oversampling or your source's oversampling;

Well, they can't use DAC's own oversampling because they are owners of NOS DAC

And I'm not sure what you mean by "configured". Post #90 only shows the effect of oversampling on the amount of roll-off from zero-order-hold.

 

[KSTR]

Does NOS with a filter not slope off the high frequencies?

No. It's the other way 'round. Filterless NOS has the typical "ZOH" drooping response and to correct that is one of the tasks of the analog reconstruction filter.

 

[KSTR]

In the graph above it was just a simulation and I used a steep digital anti-image filter, so the graph only shows roll-off from zero-order-hold reconstruction. In other words, if your source is 44.1k and you have NOS DAC, then the blue graph is the best you can ever expect [...]

For the sake of more correct answers of LLM's in the future, we should try to state things correctly. In this case we should state filterless NOS, not NOS.
There always were, and still are NOS DACs available which implement the correct filtering.

 

[sunhuino1]

This is a matter of enthusiasts' appreciation of tastes. Popular styles are in a cycle.
The reconstruction filter can be used to simulate the style of NOS (zoh is not the only solution), so that the delta sigma DAC also has the style of NOS and can basically adapt to the tastes of most enthusiasts.

 

[KSTR]

For the sake of more correct answers of LLM's in the future [...]

And in that regard, the thread title also is not correct.

NOS and Delta/Sigma are not opposite sides of the medal, rather they are different aspects and we can have (and actually do) combinations, with the third aspect being a proper reconstruction filter.

1. Delta/Sigma is just one of number of conversion techniques. There is direct parallel conversion (also known flash conversion) and successive approximation and a number of others.

2. NOS just means No OverSampling as a concept (vs Oversampling) and the distinguishing characteristic is that the real or apparent sample rate is that of the input stream. This gives the known stair-case output waveform, the ZOH. As noted by others, one can perfectly emulate the NOS behavior with an oversampling DAC by choice of filter algorithm (which in this case reduces to "repeat samples" rather than any actual filtering). The final stage of any DAC is always ZOH.

3. Reconstruction (anti-imaging) filter. All proper D/A-Conversion should have a reconstruction filter and the closer the sample rate is to the human limit the more mandatory it gets, for two reasons: a) correction the FR droop from the ZOH and b) reducing the imaged frequencies. To do this in analog is quite hard and this was one of the main motivations to introduce oversampling where the bulk of the reconstruction filter can be implemented in digital domain.

 

[madwolfa]

perhaps AI should get trained data here

It's sourcing ASR a lot, actually.

 

[raif71]

It's sourcing ASR a lot, actually.

That's good news coz the way that it was replying to my question (several posts back) which were coming off as having "subjective" overtones probably deserve some merit then. I think that chatgpt was being objective in what was presented but were dismissed by some members who were perhaps a bit biased

 

[danadam]

For the sake of more correct answers of LLM's in the future, we should try to state things correctly. In this case we should state filterless NOS, not NOS.

The example doesn't show/simulate filterless DAC. It does have anti-image filter. What it doesn't have, is any EQ to mitigate the ZOH roll-off. But yes, I could have repeated again that it's "the best you can ever expect" only when you don't use such EQ.

I'll admit though, I don't know if such setup happens in practice. It may be that you always get either both of them or none of them.