I'm sorry if this has been answered previously in the thread but with the version 3 of the firmware, was that with ODDR2 or without please? I seem to recall there was only one bit file in that version?
I can try both (RTZ versions) and listen but I just wondered if you have any idea what the actual audible differences might be?
Thanks for all your work!
I can try both (RTZ versions) and listen but I just wondered if you have any idea what the actual audible differences might be?
Thanks for all your work!
Version 3 was with ODDR2, but some people wanted it without, so now there are two. What's better - your choice
Just listened to the new firmware. My initial impression is that its a change for the worse. However, could be I'm having a bad ear day or something. I will get some other people to listen to it and see what they think. Way it sounds to me right now is that instruments are mushed together, and some individual instrument dynamics and some of individual sonority (timbre) has been lost. To put it another way, it sounds over-compressed in the dynamics sense, and lossy compressed in terms of loss of individual instrument sonority and distinct spatial positioning along the L/R soundstage. Weird. Maybe a little DC offset affects more than would be predicted by PSS FFT? Don't know.
Last edited:
Thinking about this a little more, I wonder if we don't need a new test for DSD modulators. Standard FFTs traditionally used for nominally linear amplifiers may not have the same interpretation/meaning when applied to something so highly nonlinear as a modulator. Anyone else have any thoughts on this subject?
Also, in terms of a possible new test I am wondering if a prospective inter-channel phase coherency test using a multi-tone waveform might be worth discussing. Could be talked about here, or in the context of Marcel's dac and modulator stuff, or maybe in its own thread? Question is if there is enough interest in the subject to make it into its own thread? Also, seems like such a test could be used for other devices besides modulators. Basic thought is that a multitone signal in the frequency domain can correspond to a complex time-domain signal with a long or short PSS repetition period (depending). Question would be what if two multitone signals at different volume levels were sent to a stereo modulator? Would phase coherency between channels of the modulator hold despite different PCM input volume levels for the two channels? What about if the multitone signals were different to begin with, would relative phase coherency/time-domain-synchronization still stay in the same time alignment between channels post modulation?
Seems to me something like that is necessary for proper stereo imaging, and its something that could potentially be measured using FFT methods, rather than requiring listening tests.
Also, in terms of a possible new test I am wondering if a prospective inter-channel phase coherency test using a multi-tone waveform might be worth discussing. Could be talked about here, or in the context of Marcel's dac and modulator stuff, or maybe in its own thread? Question is if there is enough interest in the subject to make it into its own thread? Also, seems like such a test could be used for other devices besides modulators. Basic thought is that a multitone signal in the frequency domain can correspond to a complex time-domain signal with a long or short PSS repetition period (depending). Question would be what if two multitone signals at different volume levels were sent to a stereo modulator? Would phase coherency between channels of the modulator hold despite different PCM input volume levels for the two channels? What about if the multitone signals were different to begin with, would relative phase coherency/time-domain-synchronization still stay in the same time alignment between channels post modulation?
Seems to me something like that is necessary for proper stereo imaging, and its something that could potentially be measured using FFT methods, rather than requiring listening tests.
Did you do your listening with a RTZ dac? AFAIK the latest firmware includes modifications which were specifically aimed at improving issues found with a specific RTZ dac. Not sure if they are applicable to other dacs. Based on what @PJotr25 explained to me the changes included extending the data length, modifying dither and adding a small DC offset. Extending data length sounds like a general refinement. OTOH adding DC offset sounds like something needed specifically for a specific RTZ dac. Modifications to dither in this case were probably also used as fine tuning for a specific RTZ DAC.
Looks like the discussion is gravitating towards Marcel's thread: https://www.diyaudio.com/community/threads/return-to-zero-shift-register-firdac.379406/post-7657111
If the added offset is meant to correct for a very small systematic offset elsewhere, like the offset of about -870 that the previous firmware seemed to have and that shifted the idle tones around half the DSD sample rate by just over 1 Hz, it is not at all obvious that that correction would degrade anything with any DAC.
Did anyone measure any DSC* DACs or other raw DSD DACs with the previous and latest configuration files, with a -60 dB, 1 kHz signal, with a DFT noise floor below -140 dB DSD and with a frequency resolution well below 1 Hz?
Did anyone measure any DSC* DACs or other raw DSD DACs with the previous and latest configuration files, with a -60 dB, 1 kHz signal, with a DFT noise floor below -140 dB DSD and with a frequency resolution well below 1 Hz?
Last edited:
Some measurements with previous firmware and DSC 2.5.2 are on this thread
https://www.diyaudio.com/community/threads/simple-dsd-modulator-for-dsc2.370177/post-7407497
The -60dB graph looks very familiar.
https://www.diyaudio.com/community/threads/simple-dsd-modulator-for-dsc2.370177/post-7407497
The -60dB graph looks very familiar.
The shape is similar, but the peaks around the fourth harmonic are about -110 dB DSD. They are closer to -130 dB DSD in https://www.diyaudio.com/community/threads/return-to-zero-shift-register-firdac.379406/post-7635705 Of course it is logical that they are higher for a DSC* DAC, as it has no return to zero.
There is a clear difference between the old and new FW seen in measurements. Why wouldn't that account for the difference in sound?
Old and new firmware may sound different for any number of reasons. Just reverted to the previous release and its super obvious it sounds far better than the new version, at least on this system. Dynamics are back, imaging in back, small details of the human voice are back.
Of course, it could be that some of the changes in the new FW would sound even better on this system. Maybe there is just one change that is dominating the bad sound.
Regarding currently used measurements, I am with @ThorstenL on that. Streetlight effect. Also WYSIATI. Thor knows because he went to see the elephant.
Of course, it could be that some of the changes in the new FW would sound even better on this system. Maybe there is just one change that is dominating the bad sound.
Regarding currently used measurements, I am with @ThorstenL on that. Streetlight effect. Also WYSIATI. Thor knows because he went to see the elephant.
Parallax Propeler P2 migh be an very interesting ic for pcm2dsd, dsd2pcm, dsp and many more things in relation to audio things, anybody tried something like this https://www.parallax.com/propeller-multicore-concept/ ?