I think that including SDM would open our investigation to too many variables for now. I suggest that, since the NOS versus OS subjective difference is audible with non-SDM DACs, we keep our investigation focused only on the technology differences there and see where we get. Our conclusions here could later be tested with regards to SDM DACs.
I do note, however, that the NOS charater difference must be apparent even via SDM operation, judging from the popularity of the (FIR filter bypassed) DDD PCM1794A SDM based, NOS DAC thread.
I do note, however, that the NOS charater difference must be apparent even via SDM operation, judging from the popularity of the (FIR filter bypassed) DDD PCM1794A SDM based, NOS DAC thread.
I would be careful making the logical leap that just because something is popular that means there is anything behind it.
I agree that using PCM179x is not a good test case unless you are going to define NOS as simply not having a classic digital reconstruction filter.
True.
To minimize the number of variables, I suggest that we define NOS as a DAC which does not feature any filter, digital or analog, specifically tasked with the role of reconstructing the signal. Meaning, mostly, or very significantly suppressing the signal's image-bands. Although I've already expressed my early suspicions of Equiripple FIR interpolators, I will resist prejudging any possibility. So, we now have collected a list of suspects, and have entered a phase of logical elimination, where analog reconstruction filters may quickly get removed from suspicion.
Last edited:
Indeed, I had a TDA1543 (multi parallel) DAC system in the earlier days in true NOS (also very popular, > 1000 kits sold). Very similar sound feeling with the "NOS" DDDAC 1794 system setup.
Does this mean, that the sound differences are mostly caused by the FIR filtering? I do believe so, as I had the FIR filter in front of the DAC also in the early days and it sounds, well like CD player typically sounds....
It would be very interesting to check by ABX or similar blind testing whether anyone can hear a difference between an interpolation filter that only repeats a sample n times and a 0.00001 dB equiripple low-pass filter at 20 kHz convolved with n repeating samples - and if so, if the difference disappears when the signal is attenuated by a few dB.
Clarification: some pre-filtered test files.
Of course, there will additional filtering in any playback system. But, if any obvious differences in files it may still come through on a transparent enough system. Music samples might include very well recorded CD quality clips of simple and of complex music. A well recorded simple music CD might be, Dire Straits, "Love Over Gold." For complex music, perhaps something orchestral, with or without vocals (e.g. symphony/opera).
Last edited:
It seems to me, that the list can be grouped in to 4 categories as follows:
A) FREQUENCY RESPONSE ERRORS
============================
1) SINC aperture based -3dB droop @ 20KHz.
B) RECONSTRUCTION/IMAGE-BAND HANDLING
===============================================
2) Lack of an FIR interpolation-filter, freeing the DAC from certain processing 'artifacts' , such as:
a) time-domain signal echoes produced within Equiripple on-chip FIR filters.
b) impulse response ringing (pre or post)
c) half-band filters plainly violating Nyquist
d) are prone to clip on peak sample normalized recordings - the intersample overshoot issue.
e) purely analog image-band suppression inherently sounds different than digital suppression?
3) Phase-modulation of the baseband signal due to insufficiently suppressed image-bands. In other words, because the signal waveform is not fully reconstructed according to the sampling theorem requirements.
4) The unsuppressed image-bands are, somehow, producing audible IM products directly within the ear.
C) ALTERED JITTER IMPACT
=============================================
5) Different jitter impact due to fewer D/A conversion cycles per second.
6) Reduced supply and ground noise due to slower clock rates.
D) SAMPLE-PERIOD RELATED QUANTIZATION ERRORS
==============================================
7) Converter settling-time becomes a smaller percentage of each conversion period as the conversion rate is made slower.
8) Harmonic-distortion may be sample-rate dependent. (per, Markw4)
A) FREQUENCY RESPONSE ERRORS
============================
1) SINC aperture based -3dB droop @ 20KHz.
B) RECONSTRUCTION/IMAGE-BAND HANDLING
===============================================
2) Lack of an FIR interpolation-filter, freeing the DAC from certain processing 'artifacts' , such as:
a) time-domain signal echoes produced within Equiripple on-chip FIR filters.
b) impulse response ringing (pre or post)
c) half-band filters plainly violating Nyquist
d) are prone to clip on peak sample normalized recordings - the intersample overshoot issue.
e) purely analog image-band suppression inherently sounds different than digital suppression?
3) Phase-modulation of the baseband signal due to insufficiently suppressed image-bands. In other words, because the signal waveform is not fully reconstructed according to the sampling theorem requirements.
4) The unsuppressed image-bands are, somehow, producing audible IM products directly within the ear.
C) ALTERED JITTER IMPACT
=============================================
5) Different jitter impact due to fewer D/A conversion cycles per second.
6) Reduced supply and ground noise due to slower clock rates.
D) SAMPLE-PERIOD RELATED QUANTIZATION ERRORS
==============================================
7) Converter settling-time becomes a smaller percentage of each conversion period as the conversion rate is made slower.
8) Harmonic-distortion may be sample-rate dependent. (per, Markw4)
Last edited:
I am using the Ian Canada FiFoPi to re-clock theI2S before it goes into the DDDAC1794
To elaborate a bit on my previous comments, it wouldn't surprise me at all if it were found that the reason NOS vs OS dacs sound different are at least in part complex interactions of factors that may be hard to untangle. For example, OS makes analog filtering easier but a dac chip may perform worse at higher sample rates because, oh, say, there is some internal capacitance at a node that has negligible effect at low frequencies, but starts to affect clocck signal timing at higher frequencies. Or, maybe transistors switching resistors or capacitors may cause output pulses or steps to be unsettled (in a transient state) for greater percentage of a step lasting a shorter time period.
The list of possible interactions could get so complex that people give up trying to make sense of it. We'll see.
However, this may actually be a more difficult problem than trying to fully understand subjective reports of cable sound. IME it isn't all the listeners head, but it may be in the case of some listeners in some situations. IOW, it gets complicated and some people can only accept simple explanations.
The list of possible interactions could get so complex that people give up trying to make sense of it. We'll see.
However, this may actually be a more difficult problem than trying to fully understand subjective reports of cable sound. IME it isn't all the listeners head, but it may be in the case of some listeners in some situations. IOW, it gets complicated and some people can only accept simple explanations.
Last edited:
