Audio Basis - articles about audio
NOS DAC (non-oversampling DAC) is a digital-to-analog converter without an oversampler. The converter may not contain an analog filter sometimes (filterless). Read why NOS DAC is used, its types, advantages and disadvantages, oversampling vs non-oversampling DAC.
A digital filter causes ringing audio and passband magnitude and, sometimes, phase deviation. A non-oversampling digital-analog converter is used to rid oversampling artifacts, that the filter makes. It is the usual DAC without an oversampler and digital filter module.
Let's look at the picture.
NOS DAC. Non-oversampled digital-analog converter
There are different converter types: two PCM, and one DSD DACs.
The sample rate of NOS DAC is defined according to the analog filter stopband.
But the analog filter has no exact border between bands. Because it is gradual.
In the picture, we can see the difference between digital and analog filters.
Digital versus analog filter
NOS DAC has no digital filter. Thus for non-oversampling DAC sample rate should be high enough, to provide more effective suppression of the first (by frequency) group of aliases by the analog filter.
NOS DAC. Sample rate and analog filter
Therefore, recording should be done on the DAC's high sampling rate.
Some NOS DAC may not contain an analog filter. It allows us to rid of the analog filter's frequency and phase distortions. Analog filter may be active (contains operational amplifiers, transistors, tubes). It means that an active analog filter can also cause non-linear distortions.
But "no analog filter" means that well-known "stairs" are present at the output. Let's consider the spectrum of the "stairs".
DAC output without analog filter
(it is a spectrum of the "stairs")
Alias spectrum is a copy of the audio signal. The signal is all in the frequency range between 0 and [sample rate]/2.
The aliases are repeatable along the frequency axis from [sample rate]/2 to infinity. Also, part of them is flipped by frequency.
Could we listen to such signal without issues? Yes. Under certain conditions.
If half of the sample rate is above 20 kHz, all aliases are located in the ultrasound range. And we can't hear it. Some people say that ear works as an analog filter. And it is so.
But there are no absolutely linear audio systems. It causes intermodulation distortions of ultrasound components. And intermodulation products by ultrasound components appear in the audible frequency range. And these products correlate with the useful audio signal. It is not good.
The level of the intermodulation products depends on non-linearity. Aliases have a level same to the useful audio signal. The more level of ultrasound, the more level of its intermodulation products.
Also need to remember that analog parts of apparatus have dynamic-range limitations. And the maximal level of signal, transmitted thru the analog parts, is defined by summary spectrum energy of the audio signal and its aliases.
I.e. aliases consume part of the dynamic range. In other words, aliases decrease useful dynamic range.
Does dynamic range decreasing give something useful? No. It reduces the maximal signal/noise ratio. And it's not good for lower-level signals especially.
To compensate alias impact on dynamic range, it should be increased.
Fortunately, analog parts of audio devices have band limitations due to capacitance and inductance of conductors and electronic components. It limits alias spread along the frequency axis via decreasing alias levels with frequency growth.
Resume: NOS without analog filter allow to rid its distortions. But such DAC may cause:
Practically, we have a limited range of available resolutions (sample rates) of recordings that are downloaded from online stores or ripped from a CD. To prepare music downloads for NOS DAC, these audio files should be pre-upsampled to a sample rate compatible with the converter.
Upsampling on a personal computer (even a laptop) may give audio developers more resources for precise digital signal processing calculations. But audio quality is defined not only by computing resources. Because design of the resampling filters may be done in many ways.
Upsampling issue, considered in this part of the article, may be applied to both PCM and DSD DACs.
The preliminary upsampling may be provided:
|No need pre-convert (upsample) records||Records should be pre-upsampled before playback|
|The required computing power of high-quality algorithms may be limited in real time. Due to a lack of power, the audio stream may be interrupted. Issues are probable with older computers||No real-time - almost no computing power limits|
|For each playback, additional electric energy is consumed for resampling||Electric energy for conversion is consumed once|
|Inline upsampled by the player audio stream should be sent to DAC without any processing (bit-perfect playback)||Offline upsampled content of the audio file should be sent to DAC without any processing (bit-perfect playback)|
In the Windows operation system, to avoid processing between the player and DAC (bit-perfect playback), ASIO and WASAPI drivers are used.
Oversampling and/or filtering may be turned on and off, it gives a DAC "NOS mode" (non-oversampling and/or non-filtering).
We must remember, that oversampling may use a digital filter in addition to the analog one.
|Non-oversampling DAC||Oversampling DAC|
|There are ringing artifacts and other digital filter distortions|
|External audio file pre-upsampling may need||Oversampling / upsampling and filtering in the digital domain before conversion to analog form inside DAC|
|R2R, sigma-delta modulated||R2R, sigma-delta modulated|
It is impossible to guarantee, that NOS DAC gives sound advantages.
As a device with simpler inner works (lesser amount of internal elements), the DAC gives potential abilities for better sound (lesser distortions).
But there are other factors, that impact to work of any kind of digital-analog converter:
For non-oversampling DAC, quality of external upsampling is very important.
If high-resolution audio records are used to playback at NOS DAC, products of ultrasound intermodulation may be more intensive in the audible frequency range due to non-linear distortions and "softer" filtering without a digital filter.
Resume: sound quality of NOS DAC vs other types depends on device implementation.
Warning: Here are no recommendations. It is not the full list. It is not advertising or offer. It is NOS DAC example list only. The list can contain mistakes and irrelevant information. Obligatorily ask DAC vendor/manufacturer about actual NOS ability.
|Manufacturer||Model||Sample rate, bit depth||Remark|
|DigiMaster||192kHz / 24 bit||R2R, USB/SPDIF|
|S5||up to 384 kHz||[Manual]|
|DAC 5 Special|
|dB Audio Labs|
|Tranquility USB DAC|
|Solid State DAC|
|Model 4705 Gemini|
|TDA1543 NOS DAC-V2.6D|
|TDA1543 NOS DAC-V3.1D|
|TDA1543 NOS DAC-V4.0|
|TDA1543 NOS DAC-V4.1D|
|TDA1543 NOS DAC-V4.5|
|LIO DSD/PCM DAC (with LIO DAC 2.0)|
DAC converts a digital signal to analog form (numbers to electrical oscillations). Analog signal is converted to acoustic waves.
Read more about:
DAC may significantly impact sound quality. Read more...
NOS DAC is non-versampling DAC. Read more...
Oversampling DAC increases sampling rate (oversample) of input digital audio signal to better filtering with an analog filter.
Oversampling uses analog and, sometimes, digital filters. Analog filters cause frequency and phase distortions. Digital filters cause also ringing distortions.
Non-oversampling is a DAC feature. It's the absence of oversampling before output analog filter.
Non-oversampling DAC is a digital-to-analog converter without oversampling at the device output.
NOS in DAC is a non-oversampling ability. I.e., such DAC converts a digital signal to analog one, but oversampling before the analog filter isn't used. Also, the analog filter may be absent.
NOS R2R DAC is a digital-to-analog converter based on resistor matrix (ladder) principle. R2R is a subtype of the matrix DACs, that is designed on resistors with 2 values: R and 2R.
NOS with analog filtering may work the same way, as the oversampling, at high sampling rates.
It's a matter of design of using DAC and sampling rate.
For low sampling rates 44.1 and 48 kHz, oversampling is required for proper analog filtering.
DAC can't improve sound quality of a played-back recording. It can convert the recording to analog with higher or lesser distortions.
DAC is important and its "quality share" may be audible. The author would place DAC's importance after a loudspeaker with amplifier.
Read about sound file optimization for your DAC...
Yes. The difference between DACs may be audible. Also, it's necessary to account a recording quality and other equipment in your audio setup.
In AV receiver or a home-theater unit, a built-in DAC is installed. It may be better or worse than stand-alone digital-to-analog converter. Such analysis requires special technical skills.
In general case, an audio comparison of the AV receiver with and without the DAC may be recommended.
Proper testing is described here...
But, it's too sophisticated and usual home comparison is enough to choose the decision.
Potentially, oversampling allows distortion reduction. However, it does not mean that any oversampled DAC sounds better than any NOS one. Because, oversampling is not a single factor.
Oversampling causes distortions (noise, ringing). However, for accurate algorithms, it's almost transparent for use.
Theoretically, ringing may become a matter for audio stuff with sharp level altering (piano, guitar, etc.). At the time of the article writing, the author knows nothing about ringing-listening impact studies from safe sources.
DAC applies low-frequency filtering of aliases. The aliases are excessive high-frequency products, that rise after conversion of digital signal to analog.
Oversampling in DAC fits analog signal for better filtering. It allows for eliminating intermodulation distortions in the audible range.
For higher sampling rate, quantization noise energy is distributed in a wider band. Noise level in the audible band is lower.
However, oversampling can't improve signal-to-noise ratio of original signal.
Also, DAC's electrical noise in the output circuits may be upper than the quantization noise of digital signal.
And we don't get noise reduction at the DAC output.
ADC (analog-to-digital converter) captures an analog electrical signal and transforms it into a digital one.
Quantization noise is combined with the electrical noise of the input circuits of the ADC. It's an input noise.
As rule, the capturing is provided at a sampling rate several times above captured range (oversampling).
It allows the distribution of the input noise energy in a wider band.
After it, in the digital domain, we can cut all frequencies above the useful captured signal band.
Part of the total noise energy will be cut too.
All these things improve the signal-to-noise ratio in the useful captured band. In instance, in the audible band.
R2R DACs have issues with non-linear distortions.
You can use pre-amp that built-in DAC or external amplifier. The result is individual for different device combinations.
You can use an external DAC, if you find it sounds better than built-in one.
If you are satisfied with your DAC, you can use an amplifier for headphones or speakers connected to the DAC.
DAC defines sound quality. YouTube will sound better with a more qualitative DAC.
As the author knows, Shiit calls R2R DACs "true multibit". Here you can find the answer about Schiit Multibit R2R DAC...
Yes. Schiit Bifrost is based on TI DAC8812 which is multiplying digital-to-analog converter.
Read more R2R DACS...
Check out who makes R2R DAC...
Oversampling requires a digital filter application. The filter causes ringing audio.
However, ringing is not critical in many cases.
Author: Yuri Korzunov,
Audiophile Inventory's founder, developer.
December 10, 2022 updated | since March 11, 2018