Return-to-zero shift register FIRDAC

Jump to Latest
#761
RTZ DAC no2 with Ray's version of the filter board with the last stage removed:

IMG_20230714_171547.jpg


Also, LVDS receiver board mounted on my own daughter-board to hook up to the PCM2DSD below. This DAC is being fed from my Logitech Transporter via LDVS.
 
  • Like
Reactions: pgour, nautibuoy and MarcelvdG
#762
MarcelvdG said:
During discussions with Hans Polak, it was found that it can sometimes be handy to have a DSD file with repetitive 101010... patterns. For a FIRDAC with a notch at half the sample rate, it eliminates ultrasonic quantization noise, folding effects due to ultrasonic quantization noise and non-linearities or crosstalk issues, and it largely suppresses the effect of far-off clock jitter, leaving only the plain old analogue circuit noise. Attached is a zipped dsf file with just over one minute of 101010... in DSD64. Mind you, the file becomes more than 1000 times larger when you unzip it.
Click to expand...

There was a bug in the file: I miscalculated the number of samples by a factor of two. A DSD player can either ignore the stated number of samples and just play it properly or complain about a corrupt .dsf file. Hans's software just played it.

Anyway, attached is a corrected file and similar files for different DSD rates, all zipped. They all become about 1000 times as large when unzipped.

There are also two DSD128 files with periodic patterns representing large DC offsets, they are in plusmineenDSD128.zip. Whatever you do, don't play these at high volumes over completely DC coupled equipment! One of them has a periodic 11101110... pattern, that is, a positive offset of 0 dB DSD and the other a 00010001... pattern, corresponding to a 0 dB DSD negative offset.
 

Attachments

#763
Just in case it is of use to anyone: attached is a Pascal program that generates DSD512 .dsf test files with silence, an offset, or a sum of an offset and a sine wave, depending on what values you give the constants offsetL, offsetR, freqL, freqR, amplL and ampR at the beginning of the program. Mind you, when the sigma-delta modulator input signals are too large, you get a runtime error. The program's file extension is supposed to be .pp, but I've changed it into .txt to please the forum software.

Unlike the attachments of the previous post, the files this program generates have irregular patterns with noise-shaped quantization noise. It uses the very same algorithm I used for the PWM8 mode of my valve DAC, except that the code to prevent misbehaviour when the sigma-delta is overdriven is missing. It is a quasi-multibit sigma-delta modulator with an embedded pulse width modulator with randomly rotated output patterns, as explained here: https://linearaudio.net/sites/linearaudio.net/files/03 Didden LA V13 mvdg.pdf sections 3.1 and 9.

The program runs under Free Pascal Compiler. Whether it actually produces a valid .dsf file may depend on the endianness of your computer.
 

Attachments

Last edited:
  • Like
Reactions: rockies914 and Raj1
#767
Hans,

May I ask if you tried running Amanero from clean +5v power?

Also, how did it sound? In your opinion to what extent did the measurements correlate with good quality musical sound as perceived on your system?
 
Last edited:
#768
Hi Mark,
Good question, but since crosstalk between data and clock already seemed to play a role in this test, I was afraid of adding just another antenna and/or groundloop to the unshielded setup.
And since my goal was just testing and not listening, I didn’t use this option for which I’m sure that it may improve the listening experience.

But as you may have read in the review, I tried a double shielded very short USB interlink versus a very simple 1.5 meter USB cable and could not detect any difference.

Hans
 
Last edited:
#770
Instead of critisizing acronyms, it would be more positive when you could suggest a better name for dsd split into 3 channels.
I haven’t come across so far.
Marcel calls it raw DSD, and I called I2S DSD to indicate that the incoming DSD signal was split in 3 signals.

Hans
 
#773
Thx Bohrok,

Very interesting information on the archimago site.
In one of his reviews he compares several PCM to DSD options.
I was happy to see that JRiver came out as almost the best.

Also tried again to find out whether an official name exists for DSD after being split into Left, Right and Clock, but no succes.
In this world with so many acronyms this seems to be the one that's missing.
I'm sorry when using the term I2S DSD caused any confusion.

Hans
 
You must log in or register to reply here.
Top Bottom