Hi, Sumotan,
Thanks, for that tip. At this stage, however, we are only endeavoring to identify the objective technical differences between OS and NOS DAC playback. With the eventual goal of discovering exactly why, technically, they exhibit differing subjective sound characteristics when, on the surface, it seems they should not. Please feel free to add your thoughts on that. ��
If the same digital samples are repeated n times within the original sampling period T without a digital filter, that is NOS, isn't it? On the other hand, the shifted signal idea was used by Cambridge Audio, where four identical DAC (TDA1541A) worked in simultaneus mode, and received the same samples shifted by T/4, T/2, 3T/4, T. Each DAC worked in NOS mode by itself, and their analog outputs were summed. I don't know what would be the equivalent digital filter for this "moving averaging".
I'm always asked myself if the sim mode was better because of the speed reduction of the data and spliting the channel in two so not working in time overlaping (or was it in frequency domain for the non sim mode ?) or if it was the change in the grounding and using the 4 pins and the grounding substract of the TDA1541 chip that made it better because the different pins and also internal applied polarity????
That's also true when you leave or not the resistors in front of the I2S input of that chip, it can be heard, whatever the purpose was for input voltage reduction or for anti-bouncing purposes of the I2S line. When this last problem is not there there is like something more details... crazy, but experienced it a lot on the TDA1545A for instance. Though ECdesigns liked to reduce at minima the voltage input of the I2S TDA1541 inputs and the devices were said to sound very fine !
.
As said in the list it's hard to make iso experience with hardware between the OS and non OS because all the things you add or remove for each. Of course the main differences are certainly about the upsampling and or added filter such hardware play in the OS devices.
Would like to hear about John from ECdesign who certainly tried that around all hisprotos.
The fact is it sounds totally different.
Maybe the first wisch of Cambridge about 4 TDA1541 was more about reducing the I/V purposes ???
That's also true when you leave or not the resistors in front of the I2S input of that chip, it can be heard, whatever the purpose was for input voltage reduction or for anti-bouncing purposes of the I2S line. When this last problem is not there there is like something more details... crazy, but experienced it a lot on the TDA1545A for instance. Though ECdesigns liked to reduce at minima the voltage input of the I2S TDA1541 inputs and the devices were said to sound very fine !
.As said in the list it's hard to make iso experience with hardware between the OS and non OS because all the things you add or remove for each. Of course the main differences are certainly about the upsampling and or added filter such hardware play in the OS devices.
Would like to hear about John from ECdesign who certainly tried that around all hisprotos.
The fact is it sounds totally different.
Maybe the first wisch of Cambridge about 4 TDA1541 was more about reducing the I/V purposes ???
Yes, this is a moving-average low-pass filter as seen from the frequency-domain, and linear-interpolation as seen from the time-domain. In your example, they are being performed entirely in analog (no digital FIR filter) by directly summing the current outputs of multiple, time-interleaved DACs.
As I had mentioned to Lampie519 in post #89, this is a rather clever idea which has been around for some time. Unfortunately, it doesn't really accomplish much that is useful, IMHO. The point of oversampling is to move the image-bands to higher frequency bands. However, in the process of accomplishing this, the existing image-bands still must be effectively removed.
Moving-average filters are very poor at seperating adjacent frequency bands. Therefore, the original image-bands will largely remain, even through samples are now being output from the DAC at a higher rate. Effectively moving the image-bands requires not only a higher output sample rate - which opens additional spectrum for the relocation - but also requires sharp filtering/removal of the existing image-bands, which you will recall begin at 22KHz for CD format.
Last edited:
ELIMINATION PHASE IS OPEN
SUGGESTED ELIMINATION PROCESS:
As we now begin the phase of suspected reason elimination, it seems obvious to me that our arguments for why a given suspect reason should, or should not, be removed from the list often will necessarily involve our own individual anecdotal listening experiences with NOS and OS. There certainly will be purely technical arguments as well.
For example, the -3dB droop @ 20KHz produced with NOS is easily corrected with simple analog equalization. My experience is that when equalized, the NOS characteristics I hear without EQ, clearly persists with EQ. Therefore, I would argue that the -3dB @ 20KHz droop is NOT responsible for the difference in character between OS and NOS, which I hear. However, in order for that suspected reason to be removed from our list should require at least a majority of concurring reports.
We must keep in mind that we are tasked only with evaluating the subjective character of the sound. To assess changes, or none, in that character. So, we will be dependent on the non-scientific reports of each other's subjective listening experiences. Is this method open to error? Yes, it is. Is there a better alternative, given the practical realities of our being a group of dispersed hobbyists? If there is, I can't think of it at the moment. If any of you feels that you have a superior method, please feel free to share it with the thread.
Alright, with all of that said, I'd like to launch the first round of eliminations by suggesting the we remove the 20KHz frequency response droop item from the list, for the reasons I've already stated above. This would effectively eliminate category 'A' entirely from the list. Anyone with a concurring subjective experience of analog EQ of the 20KHz droop, please share it. Anyone who has a counter subjective experience with analog EQ, please share that. This is just my vote, as it were. This is OUR joint investigation, not my personal investigation.
SUGGESTED ELIMINATION PROCESS:
As we now begin the phase of suspected reason elimination, it seems obvious to me that our arguments for why a given suspect reason should, or should not, be removed from the list often will necessarily involve our own individual anecdotal listening experiences with NOS and OS. There certainly will be purely technical arguments as well.
For example, the -3dB droop @ 20KHz produced with NOS is easily corrected with simple analog equalization. My experience is that when equalized, the NOS characteristics I hear without EQ, clearly persists with EQ. Therefore, I would argue that the -3dB @ 20KHz droop is NOT responsible for the difference in character between OS and NOS, which I hear. However, in order for that suspected reason to be removed from our list should require at least a majority of concurring reports.
We must keep in mind that we are tasked only with evaluating the subjective character of the sound. To assess changes, or none, in that character. So, we will be dependent on the non-scientific reports of each other's subjective listening experiences. Is this method open to error? Yes, it is. Is there a better alternative, given the practical realities of our being a group of dispersed hobbyists? If there is, I can't think of it at the moment. If any of you feels that you have a superior method, please feel free to share it with the thread.
Alright, with all of that said, I'd like to launch the first round of eliminations by suggesting the we remove the 20KHz frequency response droop item from the list, for the reasons I've already stated above. This would effectively eliminate category 'A' entirely from the list. Anyone with a concurring subjective experience of analog EQ of the 20KHz droop, please share it. Anyone who has a counter subjective experience with analog EQ, please share that. This is just my vote, as it were. This is OUR joint investigation, not my personal investigation.
Last edited:
If it's due to the response droop, then an oversampling DAC with the same response droop should sound the same. Hence the proposal of post #147, https://www.diyaudio.com/forums/digital-line-level/371931-makes-nos-sound-15.html#post6654297 , where I assume that repeating the same sample a couple of times is equivalent to playing it for a longer time (which may be wrong if the DAC has settling issues), and that the oversampling for the oversampling case can be done outside of the DAC without changing anything (which is contrary to Lampie519's impressions).
Of course it would be nicer to equalize out the response droop to within +/- 0.1 dB with an analogue filter and then use an ABX switchbox, because then these assumptions aren't needed. On the other hand, if that would make the difference disappear, the blame would probably be put on imperfections of the equalizing filter or the switchbox.
Of course it would be nicer to equalize out the response droop to within +/- 0.1 dB with an analogue filter and then use an ABX switchbox, because then these assumptions aren't needed. On the other hand, if that would make the difference disappear, the blame would probably be put on imperfections of the equalizing filter or the switchbox.
No, but I can very easily generate the impulse responses.
I'm still looking for a simple way to do the zero insertion and convolution with audio data - homemade C++ program with a wav file reading and writing library like GitHub - adamstark/AudioFile: A simple C++ library for reading and writing audio files. or using the audio features of Gnu Octave? I'm not good at C++ and I never used Octave.
A well made DAC function, most probably using OS to ease the filtering task, will recreate a more truthful copy of the recorded original. Now, compared to the rest of the really high distorting parts of a music system, this DAC is superior. But when we put this DAC into a system some gets disappointed of the result of the system chain output. They think it soned dull and flat and harsh - the problem is that one now hear the rest of the system in its "full glory". The flaws in the stereo system and the high distorsion in speakers seem to need at least one tweaked/crazy component to make the result listenable. Why not a wild DAC? Or a crazy tube amp.
It is all a compensation game at this stage of the path to correct recreation of a live performance in a hall. It will be along journey still. The whole recording/recreation architecture need to be re-defined and the electrical/mechanical transducers need at least one proper great step of improvement before we will reach the goal. Maybe only the youngest here will experience it.
Until then, the (system component) compensation game will continue and we will all find our own favourite personal way of making the best of the half mediocre situation. Some get a technically perfect DAC and seek ways to make the system sound good with it (e.g. me). Others are in love with their amps and/or speakers and compensate on the DAC side - pick your poison
My view is: why start uphill with a distorting DAC (i.e. NOS)? It just means that you have just predestined yourself to *never* reach the goal.
//
Or a crazy tube amp.It is all a compensation game at this stage of the path to correct recreation of a live performance in a hall. It will be along journey still. The whole recording/recreation architecture need to be re-defined and the electrical/mechanical transducers need at least one proper great step of improvement before we will reach the goal. Maybe only the youngest here will experience it.
Until then, the (system component) compensation game will continue and we will all find our own favourite personal way of making the best of the half mediocre situation. Some get a technically perfect DAC and seek ways to make the system sound good with it (e.g. me). Others are in love with their amps and/or speakers and compensate on the DAC side - pick your poison
My view is: why start uphill with a distorting DAC (i.e. NOS)? It just means that you have just predestined yourself to *never* reach the goal.
//
WADIA as with the 27xx model (as I have only the 27) did this shifting too using PCM1704's as 2.8Mhz SR using a DSP SPLINE oversampling
As I could hear using tube & magnepan, was clear winner against using AK4490 & ESS9038Pro DAC gear.
The WADIA 27ix have a clock link (master clock using a 3. oscillator) back to CD player to solve SPDIF jitter related degradation's.
The ordinary WADIA 27 SPDIF receiver needs to tuned while the RC filter did not get the recommended loop filter. Burning the CD's to carbon CDR's helped also in this regard and important recordings may 180° of phase.
In other words NO more spline oversampling seen so far, as with today's R2R DAC may possible as 64x factor or more...
Hp
We largely agree about digital reconstruction filters. Many DACs do, or would, sound better without their on-chip filter. I suspect those of producing artifacts stemming from an implementation which prioritizes chip cost over other objectives. One approach is address that problem is to just accept that many digital filters are problematic and dispense with them entirely. I find doing that (NOS) brings certain subjective benefits, but also leaves other subjective benefits behind. So, I'm hoping to marry the subjective benefits of both NOS and OS, while divorcing each from their respective faults. Which requires an understanding of exactly why each produces the sound character it does. If I found eliminating the digital filter to be entirely satisfying, I would simply do that and be done.
Unfortunately, I hear some characteristics from OS sound which are desireable to my ears, and that NOS doesn't seem to provide. Which doesn't preclude me from enjoying my present DAC, which is switchable between NOS or OS mode.
By the way, no need for apologies regarding your english skills. I'm quite certain they are far superior to my skill with your native language.
A lot of the problems stems from that the companies need to put out a spec that says: 0-22,05kHz +/- 0,001 dB. I dont see any notion of a "guard band" (20-22k) being utilzed as a way to ease the requirement on the brick-wall-ness. If one relaxes this and let the FR drop say a dB at 20k, the situation could be better I think. Sound quality is offered on the alter of spec numbers in the sales material.
//
I agree Ken. Just imagine all those loudspeakers we are using everywhere in the world to do this "subjective tests". deviations of +/- 3dB round 20kHz are pretty common. Still the general consent is that the typical NOS sound is perceived. So Frequency curve alone is just not it. Just dump it from your list - my 2 cents worth
Jumping ahead (or it was said already - sorry in that case): for me the difference is only in the FIR filtering. I say that because I always recognized the type of SQ perception with and without it regardless of the style of the DAC chip. PCM63, TDA1543, PCM1794. All sounded similarly with digital filter as typical CD sound and without digital filter sounding more analog NOS style - again my 2 cents worth (so already you have 4...)
A/D probably is a reason for the digital sound is not reaching its full potential. Here, also spec war is requiring super flat FR tp Fs/2 which leads to energy sneaking into the pass band below Fs/2 - this is a catastrophe and this error can never be washed out once it has happened ruining the recording forever.
//
Neither the CD2 or the CD3 were NOS. The linear interpolation was preceded by the SAA7220P.
Until recenctly i also thought that digital filtering is NOT they way to go (played many years without any with great results). But when adding a filter (seperate box in my case) that is up to the job then everything falls into place and it does exactly what it was meant for in the first place. This gives the extra "space and detail" that makes the sound so "analog".
Naturally, if this has not yet been experienced it is no use to discuss it, like trying to explain the taste of chocolate (or the milk of water buffalo) if you never had any .
Naturally, if this has not yet been experienced it is no use to discuss it, like trying to explain the taste of chocolate (or the milk of water buffalo) if you never had any .
Last edited:
You are not giving much away, and I am not sure why exactly... But I'll ask a few questions and see if you are willing to answer.
Digital filters can do anything and everything. Are you adding the bits (headroom and noise bits) and/or shifting certain spectrum (like equalizer in analog domain)? While at it (at noise...), are you talking IIR or FIR? What is the actual DAC that proceeds the filter? And, what is the preceding "block" (that provides input to this DF of yours)?
EDIT: is it M Scaler?
Digital filters can do anything and everything. Are you adding the bits (headroom and noise bits) and/or shifting certain spectrum (like equalizer in analog domain)? While at it (at noise...), are you talking IIR or FIR? What is the actual DAC that proceeds the filter? And, what is the preceding "block" (that provides input to this DF of yours)?
EDIT: is it M Scaler?
Last edited:
you are asking the right questions Boky. This is DIY forum, so if Lampie has a great box turning milk into chocolate milk, he should kindly publish it....
I tried filtering BEFORE my "NOS" DDDAC1794 several times. Including upsampling etc by use of the Roon DSP. It never was nicer to listen to. May be some one would say, it falls in place. My System is very revealing (After passive I/V only 3 SE Tube stages till speaker (with Diamond Thiel)) so may this helps hearing the difference.
But anyway lampie, if you found the holy grail, let us know !
I tried filtering BEFORE my "NOS" DDDAC1794 several times. Including upsampling etc by use of the Roon DSP. It never was nicer to listen to. May be some one would say, it falls in place. My System is very revealing (After passive I/V only 3 SE Tube stages till speaker (with Diamond Thiel)) so may this helps hearing the difference.
But anyway lampie, if you found the holy grail, let us know !