Well, I just wanted to congratulate with a truly excellent project worthy of praise on the excellent site DIYaudio.com.
Cheers
Seconded. These are really nice guys, doing excellent work.
Converting to a higher sampling rate do not add any information about the original signal. Could you please explain what is so special about these converted files?
Hi, hirezfiles,
I think your question is very reasonable.
First, I want to explain my background.
I'm a computer system engineer. Therefore, I have some basic knowledge on information theory.
I have listened such PCM high resolution sources as DXD 352.8 kHz/24 bit or 192 kHz/24 bit.
ESS ES9018 is the only DAC chip I use for playing DSD256 sources so far.
(A Japanese DIY USB-DDC board designer says TI DSD1794A is also capable for playing DSD256.)
My observation is "Sources converted to DSD256 sound far better than original 44.1kHz/16bit PCM in general". Many audiophiles around me those who actually listened to the sounds agree with this opinion.
I have never experienced such any positive effect of upsampling before in the case of conversion from 44.1kHz/16bit PCM to 352kHz/24bit PCM.
However, in the case of DSD256 on ES9018, the effect is remarkable.
I'm not sure whether this is common to other DAC chips or not because I have never tried any DAC chips other than ES9018 as of now. One possible explanation might be that this is an unique finding to the ES9018 chip.
The quality of resulting sounds apparently depends on methods or programs involved in a conversion process from PCM 44.1kHz/16bit to DSD256.
(Within my experiences, "DSD Direct" program released by SONY gives the best DSD256 sources.)
My current interpretations are;
1. Usual 44.1kHz/16bit PCM sources inherently contain far richer information than that we expected before. However, many of conventional DAC have failed to exploit it.
2. DSD256 is accidentally the most suitable input data format for ES9018 DAC chip.
3. The conversion program might add some additional information.
(According to a blog article found in a Japanese web site, SONY DSD Direct uses FIR of 30,000 taps in its initial upsampling process.)
I hope my explanation can be a suitable answer for you.
Bunpei
Bunpei,
I fitted the capacitors you referred to in post 299 as instructed. They do make a major difference to the sound quality of the already very good sounding SDtrans.
The background is much quieter, micro dynamics are better reproduced and the bass definition improves. It is just more real and transparent.
Those who have heard my system are amazed how much closer to life my system sounds - even with just 16bit 44.1khz material.
Hi rez piano recordings are even more realistic and get many people who walk along the street looking into my home to try and "see" who is playing the piano. I tell some of them "no piano, only hifi" and they are surprised.
Your SDtrans is now my main reference transport. You guys should make a larger production run. I am sure if I show off the player to others that share this hobby, a fair few would pay for the privilege of owning it.
Cheers
David
I fitted the capacitors you referred to in post 299 as instructed. They do make a major difference to the sound quality of the already very good sounding SDtrans.
The background is much quieter, micro dynamics are better reproduced and the bass definition improves. It is just more real and transparent.
Those who have heard my system are amazed how much closer to life my system sounds - even with just 16bit 44.1khz material.
Hi rez piano recordings are even more realistic and get many people who walk along the street looking into my home to try and "see" who is playing the piano. I tell some of them "no piano, only hifi" and they are surprised.
Your SDtrans is now my main reference transport. You guys should make a larger production run. I am sure if I show off the player to others that share this hobby, a fair few would pay for the privilege of owning it.
Cheers
David
I still find this very interesting. This is adding bypass caps to an already well-supplied and regulated digital transport's digital processing chips!
BUT, when I did similar bypassing to my computer transport's Juli@ sound card's digital section, which supplies an I2S feed to a separate DAC, I got similar results. I did this partly after being inspired by the comments in post 299.
Thanks to Bunpei & Chiaki for not only making what may be one of the best digital sources available today, but also for advancing the digital transport state-of-the-art with their tweaks to their transport.
Greg in Mississippi
P.S. And I can heartily endorse a larger production run!
It's basic electronics that building an effective cut off filter for 20K signals from a 44.1K DAC is the devil's own task.
Dear Matthiasw,
I'm very sorry to such a delayed reply.
You might have read my post on J River Media Center BBS.
To our regret, the project has been suspended actually. Chiaki became very busy for preparing a new ES9018-based Dual Mono DAC board with 90/98 MHz oscillators. (The board includes no I/V stage.)
As I was not satisfied with any existing USB-I2S/DSD boards, I had asked him to make a preliminary design of his own USB interface board. However, he found it is not so easy.
Best regards,
Bunpei
Dear ihear21khz and Greg,
Thank you very much for your reports.
Chiaki had designed his SDTrans384 revision 3 board (originally it was referred as "SDTrans192") very carefully assigning a low noise voltage regulator for each power line. In spite of his effort, it has been proved that his original design was not the best because not a few users including me added their own favorite bypass caps or connected their own power supplies to the board and obtained their own better sounds.
I'd like to recommend SDTrans384 users add your best bypass caps or connect your best power supplies to your board. You efforts will be surely rewarded.
However, please be careful enough not to mis-connect a high voltage. Some users have broken their board because of high voltage power supplies.
Bunpei
Thank you very much for your reports.
Chiaki had designed his SDTrans384 revision 3 board (originally it was referred as "SDTrans192") very carefully assigning a low noise voltage regulator for each power line. In spite of his effort, it has been proved that his original design was not the best because not a few users including me added their own favorite bypass caps or connected their own power supplies to the board and obtained their own better sounds.
I'd like to recommend SDTrans384 users add your best bypass caps or connect your best power supplies to your board. You efforts will be surely rewarded.
However, please be careful enough not to mis-connect a high voltage. Some users have broken their board because of high voltage power supplies.
Bunpei
Dear thoglette,
Yes, you are correct when we think of frequency components stored in CD. An anti-aliasing filter applied before CD mastering cuts completely signals above 20 KHz.
However, I used the term "richer information" that covers wider meaning than "frequency components". For example, how about "a phase reproduction" or "a time-domain wave form reproduction", etc.?
I hope you can listen to DSD256 sources converted from CD sources actually and then build your theory.
Best regards,
Bunpei
Possible next release and pricing
To those who are interested in SDTrans384,
Neither Chiaki nor Bunpei has any stock of SDTrans384 now.
Chiaki made his preliminary agreement with a Japanese small audio device vendor, "Tachyon",
Tachyon - SDTrans384 (In Japanese only)
on a possible distribution of SDTrans384.
Tachyon is preparing the next release of SDTrans384. Chiaki will not release it by himself.
However, the company Tchyon has not been starting their actual distribution yet. So their announce is preliminary. No finalized release date is available.
Further more, they have not made any decisions on over-sea sales. (This does not necessarily means they will not handle over-sea distributions.)
I think they may consider over-sea shipping if they get many requests.
According to the web page quoted above, their preliminary pricing is;
1. a bare board (which is equivalent to Chiaki's distribution of the kit)
498,000 JPY (Tax included)
2. a board in Tchyon's original casing
698,000 JPY (Tax included)
Bunpei
To those who are interested in SDTrans384,
Neither Chiaki nor Bunpei has any stock of SDTrans384 now.
Chiaki made his preliminary agreement with a Japanese small audio device vendor, "Tachyon",
Tachyon - SDTrans384 (In Japanese only)
on a possible distribution of SDTrans384.
Tachyon is preparing the next release of SDTrans384. Chiaki will not release it by himself.
However, the company Tchyon has not been starting their actual distribution yet. So their announce is preliminary. No finalized release date is available.
Further more, they have not made any decisions on over-sea sales. (This does not necessarily means they will not handle over-sea distributions.)
I think they may consider over-sea shipping if they get many requests.
According to the web page quoted above, their preliminary pricing is;
1. a bare board (which is equivalent to Chiaki's distribution of the kit)
498,000 JPY (Tax included)
2. a board in Tchyon's original casing
698,000 JPY (Tax included)
Bunpei
DSD256 play without DAC chip
Recently in Japan, several digital audiophiles have been probing DSD playing without DAC chips by applying a simple LPF to digital output raw signals of DSD.
I also tried an "easy" LPF (LCR) directly connected to output pins of SDTrans384.
Wow! Clear analog sound of an enough volume level comes through such an easy filtering!
I think it's worth trying for SDTrans384 users those who are interested in DSD play.
Recently in Japan, several digital audiophiles have been probing DSD playing without DAC chips by applying a simple LPF to digital output raw signals of DSD.
I also tried an "easy" LPF (LCR) directly connected to output pins of SDTrans384.
Wow! Clear analog sound of an enough volume level comes through such an easy filtering!
I think it's worth trying for SDTrans384 users those who are interested in DSD play.
Hi, guglielmope & compressit,
Thank you very much for your replies. I will explain the "easy" LPF.
Please remember it is very primitive and not matured. It's just the first trial.
1. You need to add your coupling capacitors to the LPF output
when you have no coupling capacitors in your system.
2. In my case, I adopted a simple LCR type filter.
Some Japanese reported the use of only R, only L, CR, LC, etc. There can be many configurations as you like.
3. I referenced the LCR type filters in the following schematics;
http://www.ti.com/lit/ug/slou220/slou220.pdf (page 36, TI TAS5706B Class D amplifier chip evaluation board)
https://a4956983-a-62cb3a1a-s-sites.googlegroups.com/site/koonaudioprojects/dsd-playback-system/DSD%20Amp.png?attachauth=ANoY7cp2VaDZGwqMK9b9jri_GVR7fo2kRD911cmm8c3GWvRI9apmv1H1Q7L-3prXuyGAqYqzfbBm_T4ZfQ80zUSQFzchhQceJhb3JlBIGzDtVJiOXI_o5LkFbuiQzTobAdROhmX-wHI4zFPQn8nw1p_vhIp0lCO5jLJQ8BHJPMDTS190vUc54x-ayXyRjPeSDsHiTNFyg0-Fv3Pojlw3OKez5h5MCMPpRtQdMOgVosiup-_1ftZ3kM-vtdsFKxbuOxtVd1nVtuYL&attredirects=0 (Koon's DSD Class D amplifier)
4. If you pursue a certain level of quality, you may need to think of such further points;
A. The LPF segment should be isolated from SDTrans because the LPF segment is not of "Digital" but of "Analog".
B. The logic output of DSD signal should be bipolar or differential so as to eliminate the use of coupling caps.
C. The logic output should be drived by a certain analog amplifier.
D. The LPF should be of high order. for example, cascaded multiple LCRs or analog FIR.
E. LPF frequency should be adjusted depending on DSD64, DSD128, DSD256.
When I listened to sounds from the "easy" LPF, I felt that they were of higher resolution than I expected and very vivid or something wild. On the other hand, bass lacks energies and soft.
I think you need your certain efforts for possible improvements.
Bunpei
Thank you very much for your replies. I will explain the "easy" LPF.
Please remember it is very primitive and not matured. It's just the first trial.
1. You need to add your coupling capacitors to the LPF output
when you have no coupling capacitors in your system.
2. In my case, I adopted a simple LCR type filter.
Some Japanese reported the use of only R, only L, CR, LC, etc. There can be many configurations as you like.
3. I referenced the LCR type filters in the following schematics;
http://www.ti.com/lit/ug/slou220/slou220.pdf (page 36, TI TAS5706B Class D amplifier chip evaluation board)
https://a4956983-a-62cb3a1a-s-sites.googlegroups.com/site/koonaudioprojects/dsd-playback-system/DSD%20Amp.png?attachauth=ANoY7cp2VaDZGwqMK9b9jri_GVR7fo2kRD911cmm8c3GWvRI9apmv1H1Q7L-3prXuyGAqYqzfbBm_T4ZfQ80zUSQFzchhQceJhb3JlBIGzDtVJiOXI_o5LkFbuiQzTobAdROhmX-wHI4zFPQn8nw1p_vhIp0lCO5jLJQ8BHJPMDTS190vUc54x-ayXyRjPeSDsHiTNFyg0-Fv3Pojlw3OKez5h5MCMPpRtQdMOgVosiup-_1ftZ3kM-vtdsFKxbuOxtVd1nVtuYL&attredirects=0 (Koon's DSD Class D amplifier)
4. If you pursue a certain level of quality, you may need to think of such further points;
A. The LPF segment should be isolated from SDTrans because the LPF segment is not of "Digital" but of "Analog".
B. The logic output of DSD signal should be bipolar or differential so as to eliminate the use of coupling caps.
C. The logic output should be drived by a certain analog amplifier.
D. The LPF should be of high order. for example, cascaded multiple LCRs or analog FIR.
E. LPF frequency should be adjusted depending on DSD64, DSD128, DSD256.
When I listened to sounds from the "easy" LPF, I felt that they were of higher resolution than I expected and very vivid or something wild. On the other hand, bass lacks energies and soft.
I think you need your certain efforts for possible improvements.
Bunpei
The idea of using a LPF directly from DSD data, with or without post-filter buffer, is an old one, first seen in Vacuumstate SACD Player mods.
Some DIY examples can also be found here :
http://www.diyaudio.com/forums/digital-source/42441-out-stage-sony-cd-players.html
Regards,
Patrick
Some DIY examples can also be found here :
http://www.diyaudio.com/forums/digital-source/42441-out-stage-sony-cd-players.html
Regards,
Patrick