@TNT
Its not a rotation. It is a reversal.
https://www.browserling.com/tools/reverse-binary
https://onlinetoolz.net/bitshift#base=2&value=0&bits=8&steps=1&dir=r&type=log&allsteps=1
Its not a rotation. It is a reversal.
https://www.browserling.com/tools/reverse-binary
https://onlinetoolz.net/bitshift#base=2&value=0&bits=8&steps=1&dir=r&type=log&allsteps=1
I’ll come back with more accurate sims for DSD256 silence, now played with 88.6 nsec steps, with all further details.
Hans
Impressive analysis.
I’ll try to see whether any of this can be detected in a simulation.
If so, changing the filter topology could give more insight.
Hans
Yes true, because that was only a very short time window, so I will need at least a 1 second window to get a 1Hz resolution.
Will be quite a challenge
The other thing is that I’m using .dsf files converted by JRiver from PCM.
Will they produce the same anomalies as Bohrok’s PCMtoDSD converter ???
Hans
@bohrok
Did you also have these low level artefacts at DSD64 ?
I’m asking because creating a large time window in DSD256 will exceed by far the borders of what’s possible in my simulation.
I can maximally achieve a 0.35 second window in DSD64 in one go, that’s the equivalent of 1Mio samples.
Eventually I could possibly concatenate three of those windows if needed.
Hans
Did you also have these low level artefacts at DSD64 ?
I’m asking because creating a large time window in DSD256 will exceed by far the borders of what’s possible in my simulation.
I can maximally achieve a 0.35 second window in DSD64 in one go, that’s the equivalent of 1Mio samples.
Eventually I could possibly concatenate three of those windows if needed.
Hans
Yes true, because that was only a very short time window, so I will need at least a 1 second window to get a 1Hz resolution.
Will be quite a challenge
The other thing is that I’m using .dsf files converted by JRiver from PCM.
Will they produce the same anomalies as Bohrok’s PCMtoDSD converter ???
Hans
That 300 Hz or so resolution should be quite sufficient to see the noise shaping curve in the audio band. If I were you, I would first simulate an ideal DAC with ideal filter and tweak the reltol, maximum time step and DFT settings until I saw a noise shaping curve, and then think about longer time windows for bohrok2610's low-level artefacts.
Who knows? It must be better in this respect: two out of three modulator algorithms of the solid-state logic gate DAC are quasi-multibit, it has one extra order of passive filtering before the active part of the filter (constructed such that it spoils the virtual ground at low frequencies) and it uses FET op-amps.
Last edited:
At a given signal frequency, to get the same modulation index at lower sample rates, you have to increase the signal level inversely proportionally to the sample rate. You could also keep the signal level constant and reduce its frequency (250 Hz, -60 dB).
Here are measurements of 1kHz -60dBFS at DSD64, DSD128 and DSD256. The output stage is the dual OPA1632 "hat" I used in my output stage shootout. All signals were generated from sox 1kHz -60dBFS 44k1 PCM using sox-dsd with sdm8 filter. Player was foobar2000.
The low level distortion artefacts are clearly depending on the used modulator but sox-dsd with sdm8 seems to work well. These results look quite satisfactory to me.
The splitted harmonic peaks of PCM2DSD is my main concern as PCM2DSD is by far the easiest and most economic way to use this RTZ dac. Performance is very good apart from the splitted peaks.
The low level distortion artefacts are clearly depending on the used modulator but sox-dsd with sdm8 seems to work well. These results look quite satisfactory to me.
The splitted harmonic peaks of PCM2DSD is my main concern as PCM2DSD is by far the easiest and most economic way to use this RTZ dac. Performance is very good apart from the splitted peaks.
Attachments
Duh... may bad. I was wrong. See that now.
//
... odd operation... byte swap.. what does it arithmetically mean?
Last edited:
