[FredA]

And if I stay on the top table (not jumping S2), what could I expect from this combination?

​

if not NOS, then perhaps it's an easter egg that loops a MOD/MIDI track of the Chinese national anthem buried in the FPGA

God only knows!

Seriously, you will get another flavor of nos. How different from the other ones? It might just play the data without any processing/filtering.

 

[FredA]

Another benefit from my current configuration is i get no popping noise when switching from 188 to 192 (i only do 2x overampling on the 96/24 files).

 

[gk2013]

"OS" - S2 OFF - implementation has digital filtering and requires a "master clock' to operate.
Oversampling options are 1x, 2x 4x and 8x. (1x is Non Oversampling)

"NOS" -S2 ON- implementation has no digital filtering; no oversampling and asynchronous. The incoming data is clocked against the internal TCXO of the R2R 7.

A "Master 7" and "NOS 7" combined (Not exactly, the new digital board has a lot of advantages)

 

[FredA]

I use the Izotope upsampler built in audirvana with the sharpest slope, max phase linearity, max anti-aliasing, cut-off freq. at 1.02 Nyquist, and a 1000000 sample window to compute the output. The latter makes a significant improvement from the standard izotope settings. You get better definition in short.

With 4x oversampling by izotope in nos2, i think i have a sound i like more than the os mode. It's more accurate. And smoother. And more laid back. I am gonna keep this configuration for a while and see. So far i like it a lot.

Using the same recipe as above but with nos 3, i have a new winner. Better bass impact and attacks in general while retaining the laid back character and relaxed sound. The sound is warmer, but still retain excellent top end accuracy. So far so good.

It's like the os mode but with smoother top end, more laid back and warmer. A bit softer attacks. Perhaps bigger bass impact. Excellent match for my gears. For now ar least. I would say the presention is close to that of the Master-7, but with more accuracy and a more liquid sound. A bit warmer as well. Maybe too warm, i'll see.

Kingwa had recommended i use this conguration given my comments at some point. So far so good. Makes sense that in the end, if timing is more accurate, i.e. faithful the the source, i will end up obtaining a similar sound to what i had with my old dac. It seems to me that timing has more influence on the sound than anything else with this level of equipement quality. The singxer f-1 well fed has a sound signature i just love (i2s out). I presume results will vary depending on the source you use.

 

[FredA]

And now, same settings with upsampling pushed to 352/384: fabulous top end. Smoothest and most resolved treble ever heard in this room. I am quite shocked.

Compared to the r2r 7 os mode, it's not as dynamic but it's so fluid... So easy on the ears. This treble is something really exceptional. As if suddenly i was using a 20K dac. So clean, the separation is exceptional as well. Bring me more highs, they are so awesome!

 

[HoloSpice]

"As if suddenly i was using a 20K dac"

 

[Currawong]

PS: @Currawong: when reviewing the R2R-7, I recommend to try to get a source like the Magna Mano or Sotm sMS-200 ultra to loan, just to experience the effect. It is really remarkable what impact the digital source has on these NOS dacs...

I have an iUSB 3.0 feeding an F-1, which has been a good combination.

I agree that NOS is more sensitive to upstream jitter and noise than oversampling. That was the main reason oversampling was invented.

Incorrect. Without oversampling, you get quite extreme levels of aliasing inside the audio band.

Here are the NOS modes from an email I got from Kingwa when I asked about them:

Mode 1 is simply process .
Mode 2 is FIFO mode process.
Mode 3 is ultra high speed process.
​

I gather that Mode 3 is technically the best, but the most transport-dependent.

Edit: I've pulled out my bag of spare connectors and bits and found a switch that would fit in the air vents in the top plate, so now I can switch in NOS mode whenever I like. I'm trying it again with the "accurate" firmware and it's much nicer than I remember it being before. It seems to hit a nice spot between precision and the "I don't want the music to stop" listenability I first experienced.

 

[Articnoise]

Incorrect. Without oversampling, you get quite extreme levels of aliasing inside the audio band.

I have an iUSB 3.0 feeding an F-1, which has been a good combination.

Deleted, see next post.

 

[Articnoise]

I have an iUSB 3.0 feeding an F-1, which has been a good combination.



Incorrect. Without oversampling, you get quite extreme levels of aliasing inside the audio band.

Here are the NOS modes from an email I got from Kingwa when I asked about them:

Mode 1 is simply process .
Mode 2 is FIFO mode process.
Mode 3 is ultra high speed process.
​

I gather that Mode 3 is technically the best, but the most transport-dependent.

Edit: I've pulled out my bag of spare connectors and bits and found a switch that would fit in the air vents in the top plate, so now I can switch in NOS mode whenever I like. I'm trying it again with the "accurate" firmware and it's much nicer than I remember it being before. It seems to hit a nice spot between precision and the "I don't want the music to stop" listenability I first experienced.

You are incorrect and exaggerating. Without oversampling, you don’t get extreme levels of aliasing inside the audio band in well-made NOS DAC. Have you ever heard a NOS DAC like TotalDac?

Upsampling is a method to help filter out high frequency digital artifacts without using sharp analogue filters. When a DAC converts from digital to analog minor inaccuracies in this process introduce artifacts or so called aliasing over the replay frequency. At 44 kHz sample rate aliasing are introduced at just above 22 kHz, which is out of the audio band. A NOS DACs require an analog filter to be placed to cut everything above 22 kHz to eliminate artifacts. By upsampling to a higher frequency a DAC essentially moves those artifacts to a higher frequency, farther from the audible spectrum and making them easier to filter out without impacting the audible frequency spectrum. Both OS and NOS has its pros and cons and both need quality filter to sound good. Digital filters are not easy to get right and bad digital filters can ruin the sound of an otherwise good OS DAC. So with a good anti-aliasing filter its other things that comes in to play. Namely noise, jitter and better phase response. I repeat it’s not aliasing that is the problem in a well-made NOS DAC.

“One of the key measurement parameters for analog-to-digital converters (ADCs) is signal-to-noise ratio (SNR). SNR measures the relative level between the desired signal power and the entirety of the noise power within the first Nyquist zone. The Nyquist zone bandwidth is the sampling rate divided by two (Fs/2). Recall that all signals and noise will fold back into the first Nyquist zone. This zone effectively represents the entire bandwidth of the device. One benefit of over-sampling is that the image components are separated farther in frequency space. This allows easier analog filtering to eliminate interfering signals that can alias down into the captured bandwidth and desensitize the receiver. The over-sampled case provides a more realizable analog anti-aliasing filter. Over-sampling can improve the SNR performance of the device beyond the theoretical quantization noise limitations. The quantization noise is equally distributed across the Nyquist bandwidth. By increasing the sampling rate, the same quantization noise is spread over a larger Nyquist bandwidth. The desired signal remains fixed.”

Read about noise in link below: https://e2e.ti.com/blogs_/b/analogw...ampling-how-over-sampling-is-cheating-physics

“Since the jitter bandwidth in a sample clock can extend to half the sampling frequency of the converter, the jitter in an oversampled converter will be spread over a wider spectrum than the jitter in a non-oversampled converter. The error caused by jitter modulation is related to the jitter spectrum, so the error signal from an oversampled converter is also spread across a wider spectrum. To illustrate this: consider a 1 kHz signal being sampled with 1 ns of spectrally flat, noise-like jitter. By calculation, this will produce a total error 104 dB below the signal. This total error figure remains the same regardless of the sample rate of the converter. As you can see in Figure 16, in a 4X oversampled DAC this error signal will be spread over four times the frequency range compared with a 1X converter. For audio purposes, of course, we limit our interest to the 20 Hz to 20 kHz bandwidth, and a measurement made over that range contains only one-quarter of the power of the full spectrum of error noise. One-quarter the power implies one-half the voltage, resulting in an error 6 dB lower than that for the non-oversampled converter.”

Read about jitter on page 11-12 in link below: http://www.audiophilleo.com/zh_hk/docs/Dunn-AP-tn23.pdf and here: https://www.stereophile.com/asweseeit/344/index.html

I repeat it’s not aliasing that is the problem in a well-made NOS DAC. I don’t have the time and desire to quote all the articles that are written about this, but much more can be fund if searching the Internet.

 

[Currawong]

This is a 20 kHz (iirc, the labels disappeared with the forum software transition) sine wave output of a NOS DAC. It's very visibly heavily aliased. On my other computer I have similar ones showing 1kHz, 5kHz and 10kHz with varying levels of aliasing for the NOS measurements, and perfect sine waves for the oversampled output. Since the worst of it occurs high in the audio band, above 10 kHz and mostly into the harmonics of instruments, it seems to mostly affect imaging.

 

[FredA]

One key point missing above is that due to the non linear responses beyond audible range of the output stage and equipements downstream including the speakers and cables, the entire audible spectrum will be polluted from distortion of these high-frequency contents the overall filter (digital plus analog) could not eliminate.

So basically, if you can implement a filter with perfect response (infinite slope, zero phase difference), you will recover the exact signal and without anything beyond the audio range. And it will remain more intact once it has become a sound wave. In practice, it is possible to very well approximate such a filter if introducing a delay in the signal and using a large time window. You need a filter with an impulse response that is (sin x)/x, which in it's ideal form corresponds to a response that ranges from -infinity to +infinity in the time domain.

Another very important consideration is that for the ideal filter to work perfectly, it has to be fed samples with infinite precision and it's response should be such that these original samples be left intact at the output. In practice there is a quantization error. In my opinion, the best you can do is to implement a filter that will only allow a deviation corresponding to the quantization error, the original samples (assumed accurate in the first place) thus remaining as intact as can be.

Yet another consideration however is rise time and dynamics. To have sharp transitions, you possibly need a banwidth that goes beyond the audible range. It would mean that for perfect dynamics, you need to extend the filter's output beyond, trying to figure out the best interpretation of the samples. If this is the case, were are not just talking low-pass filtering but also of trying to figure out what was beyond audible range in real life when the actual sound was recorded. I think that amplifiers and output stages try to acheive this goal as well so matching equipment can be tricky with this regard.

So as you can see, sound reproduction requires some sort of interpretation given approximate signal representions.

 

[FredA]

Another key point in digital audio is jitter. Jitter can alter frequency response significantly although it will not show in graphs because of the way it is measured. And it has more severe effects as well. Il will cause a loss of accuracy.

 

[Articnoise]

This is a 20 kHz (iirc, the labels disappeared with the forum software transition) sine wave output of a NOS DAC. It's very visibly heavily aliased. On my other computer I have similar ones showing 1kHz, 5kHz and 10kHz with varying levels of aliasing for the NOS measurements, and perfect sine waves for the oversampled output. Since the worst of it occurs high in the audio band, above 10 kHz and mostly into the harmonics of instruments, it seems to mostly affect imaging.

Seriously is that a sine wave? Look corrupt and definitely a lot of aliasing. Which DAC is it?

 

[Articnoise]

One key point missing above is that due to the non linear responses beyond audible range of the output stage and equipements downstream including the speakers and cables, the entire audible spectrum will be polluted from distortion of these high-frequency contents the overall filter (digital plus analog) could not eliminate.

So basically, if you can implement a filter with perfect response (infinite slope, zero phase difference), you will recover the exact signal and without anything beyond the audio range. And it will remain more intact once it has become a sound wave. In practice, it is possible to very well approximate such a filter if introducing a delay in the signal and using a large time window. You need a filter with an impulse response that is (sin x)/x, which in it's ideal form corresponds to a response that ranges from -infinity to +infinity in the time domain.

Another very important consideration is that for the ideal filter to work perfectly, it has to be fed samples with infinite precision and it's response should be such that these original samples be left intact at the output. In practice there is a quantization error. In my opinion, the best you can do is to implement a filter that will only allow a deviation corresponding to the quantization error, the original samples (assumed accurate in the first place) thus remaining as intact as can be.

Yet another consideration however is rise time and dynamics. To have sharp transitions, you possibly need a banwidth that goes beyond the audible range. It would mean that for perfect dynamics, you need to extend the filter's output beyond, trying to figure out the best interpretation of the samples. If this is the case, were are not just talking low-pass filtering but also of trying to figure out what was beyond audible range in real life when the actual sound was recorded. I think that amplifiers and output stages try to acheive this goal as well so matching equipment can be tricky with this regard.

So as you can see, sound reproduction requires some sort of interpretation given approximate signal representions.

Yes NOS DACs are inherently sensitive to the upstream equipment, all DAC are, but NOS DACs are extra sensitive.

 

[FredA]

I have an iUSB 3.0 feeding an F-1, which has been a good combination.



Incorrect. Without oversampling, you get quite extreme levels of aliasing inside the audio band.

Here are the NOS modes from an email I got from Kingwa when I asked about them:

Mode 1 is simply process .
Mode 2 is FIFO mode process.
Mode 3 is ultra high speed process.
​

I gather that Mode 3 is technically the best, but the most transport-dependent.

Edit: I've pulled out my bag of spare connectors and bits and found a switch that would fit in the air vents in the top plate, so now I can switch in NOS mode whenever I like. I'm trying it again with the "accurate" firmware and it's much nicer than I remember it being before. It seems to hit a nice spot between precision and the "I don't want the music to stop" listenability I first experienced.

Maybe the accurate firmware is the way to go with the configuration i use since last night. I could possibly obtain a slightly more dynamic sound, which would be nice. But i like this laid back, relaxed and amazingly accurate sound i get now. I strongly recommend you give it a try if not done already.

To pass the switch wire through the venting slot is a good idea. How did you manage to put the cover back into place? Using a long wire? A bit tricky. I will try to locate a place where i could install a permanent switch. Again, that would by nice if Kingwa offered a kit for us to buy with replacement faceplate, wire and toggle button.