New Japanese IV transformers
Some of you might be interested in reading the mini review of a new IV transformer for the Buffalo2 I wrote today.
http://www.diyaudio.com/forums/twisted-pear/164294-buffalo-ii-transformers-18.html#post2696689
Some of you might be interested in reading the mini review of a new IV transformer for the Buffalo2 I wrote today.
http://www.diyaudio.com/forums/twisted-pear/164294-buffalo-ii-transformers-18.html#post2696689
DPLL Bandwidth parameter "no bandwidth"
On his BBS, Chiaki reported his successful result of playing DXD 352.8 kHz /24 bit PCM files with ES9018 DAC chip by setting DPLL Bandwidth parameter "no bandwidth" under "OSF=ON" and "Jitter reduction=ON".
The key point is feeding a "synchronous master clock" generated in SDTrans384 to ES9018 DAC chip. The master clock frequency is 90.3168 MHz.
On his BBS, Chiaki reported his successful result of playing DXD 352.8 kHz /24 bit PCM files with ES9018 DAC chip by setting DPLL Bandwidth parameter "no bandwidth" under "OSF=ON" and "Jitter reduction=ON".
The key point is feeding a "synchronous master clock" generated in SDTrans384 to ES9018 DAC chip. The master clock frequency is 90.3168 MHz.
Connecting to TAS5706 Evaluation Board
I could connect TI TAS 5706 (full digital input Class-D amplifier) to SDTrans384 via I2S by setting some registers on the chip using Arduino.
The chip on its evaluation board could play PCM WAV files ranging from 44.1 kHz/16 bit to 192 kHz/24 bit without any problem.
DAC chips and full digital input Class-D Amplifier chips I have confirmed the compatibility of SDTrans are;
<DAC>
Wolfson Microelectronics: WM8741 ( - 192 kHz/24 bit)
ESS Technology: ES9018 ( - 352.8 kHz/24 bit)
Asahi Kasei Microdevices: AK 4399 ( - 352.8 kHz/24 bit)
Texas Instruments: PCM1794 ( - 192 kHz/24 bit)
<Class-D amplifier>
STMicroelectronics: STA328 ( - 192 kHz/24 bit)
Texas Instrumenents: TAS5706 ( - 192 kHz/24 bit)
I could connect TI TAS 5706 (full digital input Class-D amplifier) to SDTrans384 via I2S by setting some registers on the chip using Arduino.
The chip on its evaluation board could play PCM WAV files ranging from 44.1 kHz/16 bit to 192 kHz/24 bit without any problem.
DAC chips and full digital input Class-D Amplifier chips I have confirmed the compatibility of SDTrans are;
<DAC>
Wolfson Microelectronics: WM8741 ( - 192 kHz/24 bit)
ESS Technology: ES9018 ( - 352.8 kHz/24 bit)
Asahi Kasei Microdevices: AK 4399 ( - 352.8 kHz/24 bit)
Texas Instruments: PCM1794 ( - 192 kHz/24 bit)
<Class-D amplifier>
STMicroelectronics: STA328 ( - 192 kHz/24 bit)
Texas Instrumenents: TAS5706 ( - 192 kHz/24 bit)
Attachments
I am very spoilt :-D
This is because Bunpei personally works on my entire Buffalo 3 and SDTrans 384 playback system and visits my home in Setagaya, Tokyo frequently. It is of course a give-and-take, because both Bunpei and myself get to learn from the various experiments done.
The latest iteration is 2 x Buffalo 3 converter boards connected in parallel stereo with the exotic Urushi Painted "Feastrex Branded" line transformers. Bunpei has also programmed the ARDUINO controller for 9-bit quantizer operation of the ESS 9018s. Finally a pair of hefty OS Con electrolytic caps has been connected to the to the regulators. I understand that this is a risky proposition that is not recommended by the supplier, but I think that as long as I take the responsibility for any eventual damage myself, everything should be allowed in DIY audio bar causing fire or explosions :-D
The system does of course include the "synchronous master clocking" scheme and the Cornell Dubilier Electronics (CDE) film capacitors in the power supply. BTW: the SDTrans 384 itself is powered by three or four Panasonic Evolta D-Cells.
The previous iteration was based on a single Buffalo 2 which also had been set to 9-bit quantizer, but without any separate regulators or caps in the regulation. I can report that the sound improvement today was one of the biggest jumps upward that I have experienced in the Buffalo/SDTrans development. The sound is richer, creamier, higher resolution, more elegant, more vivid. It simply makes you feel like listening to more great music!
I normally use the excellent Fidelix Caprice, and I have not yet done a direct comparison. This isn't really about any "shoot-out" anyway. I look at the Buffalo development as one direction and the Caprice as another. I am quite happy that both can deliver satisfactory sound at their individual "vectors".
This is because Bunpei personally works on my entire Buffalo 3 and SDTrans 384 playback system and visits my home in Setagaya, Tokyo frequently. It is of course a give-and-take, because both Bunpei and myself get to learn from the various experiments done.
The latest iteration is 2 x Buffalo 3 converter boards connected in parallel stereo with the exotic Urushi Painted "Feastrex Branded" line transformers. Bunpei has also programmed the ARDUINO controller for 9-bit quantizer operation of the ESS 9018s. Finally a pair of hefty OS Con electrolytic caps has been connected to the to the regulators. I understand that this is a risky proposition that is not recommended by the supplier, but I think that as long as I take the responsibility for any eventual damage myself, everything should be allowed in DIY audio bar causing fire or explosions :-D
The system does of course include the "synchronous master clocking" scheme and the Cornell Dubilier Electronics (CDE) film capacitors in the power supply. BTW: the SDTrans 384 itself is powered by three or four Panasonic Evolta D-Cells.
The previous iteration was based on a single Buffalo 2 which also had been set to 9-bit quantizer, but without any separate regulators or caps in the regulation. I can report that the sound improvement today was one of the biggest jumps upward that I have experienced in the Buffalo/SDTrans development. The sound is richer, creamier, higher resolution, more elegant, more vivid. It simply makes you feel like listening to more great music!
I normally use the excellent Fidelix Caprice, and I have not yet done a direct comparison. This isn't really about any "shoot-out" anyway. I look at the Buffalo development as one direction and the Caprice as another. I am quite happy that both can deliver satisfactory sound at their individual "vectors".
Last edited:
ES9018 "No Bandwidth" experience
A few month ago I tried No bandwidth setting with syncronous ES9018 clocking, just after seeing Chiaki's report on his BBS.
It seemed that ideal mclk frequency varies depends on samplelates.
MCLK frequencies I had succeeded to play music files - not perfectly, BII failed to lock every 30-40 seconds - under No Bandwidth setting are as follows.
Tested with Buffalo II dac (OSF on) - i2s - SDtrans384 with number17 EEPROM.
352.8khz - 90.3168mhz from mclk output
96khz - 24.576mhz, tapped from CN11
44.1khz - haven't succeeded with STtrans-BuffaloII combo, only succeeded when using QA-550 sd card transport and tapping 11.2896mhz XO from it.
I feel ES9018's master clock should be close to correct mclk requirements when aiming "No Bandwidth" DPLL setting under syncronous clocking scheme.
But honestly to say, I couldn't find obvious sonic improvement with changing DPLL setting from "Lowest" to "No Bandwidth".
Lowest or even higher DPLL is fine to me under syncronous clocking.
About "correct mclk requirements", forum member glt made great effort to gathering informations and well explains in his "Hifiduino" blog.
A few month ago I tried No bandwidth setting with syncronous ES9018 clocking, just after seeing Chiaki's report on his BBS.
It seemed that ideal mclk frequency varies depends on samplelates.
MCLK frequencies I had succeeded to play music files - not perfectly, BII failed to lock every 30-40 seconds - under No Bandwidth setting are as follows.
Tested with Buffalo II dac (OSF on) - i2s - SDtrans384 with number17 EEPROM.
352.8khz - 90.3168mhz from mclk output
96khz - 24.576mhz, tapped from CN11
44.1khz - haven't succeeded with STtrans-BuffaloII combo, only succeeded when using QA-550 sd card transport and tapping 11.2896mhz XO from it.
I feel ES9018's master clock should be close to correct mclk requirements when aiming "No Bandwidth" DPLL setting under syncronous clocking scheme.
But honestly to say, I couldn't find obvious sonic improvement with changing DPLL setting from "Lowest" to "No Bandwidth".
Lowest or even higher DPLL is fine to me under syncronous clocking.
About "correct mclk requirements", forum member glt made great effort to gathering informations and well explains in his "Hifiduino" blog.
Hi, mrbear,
It's a tough question.
In an objective way, there is no way to measure it properly for me.
On SDTrans384, I2S signals (LRCLK, SDAT, BCLK) are generated in FPGA based on an external clock. As for MCLK, we have two selections. One way is to use the original external clock as MCLK and another way is to derive MCLK from the FPGA. When I compared the subjective sonic impressions before, I felt the former way was better. However, at the experiment, my preparations for power lines were not adequate enough.
If you can explain your question more in detail, I may be able to give you more suitable answer.
It's quite an interesting finding! I think we need to have more experiments on this point.
perhaps a little off topic...
I have a SD card "music sampler" used in music stores. It uses an AVI amp with a SD card music "file" reader and speakers. The board is made by Somerset (I think). Digital readout for song number. I didn't get the face plate, so can't tell if there is a "next" button it. If any want I could take photos of it.
I have a SD card "music sampler" used in music stores. It uses an AVI amp with a SD card music "file" reader and speakers. The board is made by Somerset (I think). Digital readout for song number. I didn't get the face plate, so can't tell if there is a "next" button it. If any want I could take photos of it.
Just recently I am trying a combination of SDTrans with a full digital class-D amplifier, TI TAS5706. I think your music player can be positioned in the same product area. However, our SDTrans was originally designed for proof of concept of "Bit perfect play of PCM sources" and it discarded playing of MP3 and flac. Therefore, SDTrans may not be suitable for that kind of "music sampler" application.
Anyway, I appreciate your kind post of photos of the sampler very much.
Chiaki made a new experiment on synchronous MCLKing for an ES9018-based DAC.
The detailed explanation is posted here.
http://www.diyaudio.com/forums/digital-line-level/117238-ess-sabre-reference-dac-8-channel-189.html#post2785373
The result is very excellent.
The detailed explanation is posted here.
http://www.diyaudio.com/forums/digital-line-level/117238-ess-sabre-reference-dac-8-channel-189.html#post2785373
The result is very excellent.
The next kit release plan
I talked with Chiaki yesterday.
To our regret, he has no clear plan for the next release of SDTrans384 board so far.
As for the optional synchronous clocking sub boards for ES9018 DAC chip, he has no firm forecast on the demand because the optional board set is rather expensive.
I talked with Chiaki yesterday.
To our regret, he has no clear plan for the next release of SDTrans384 board so far.
As for the optional synchronous clocking sub boards for ES9018 DAC chip, he has no firm forecast on the demand because the optional board set is rather expensive.
DSD, please . . .
@Bunpei,
Excellent thread
Dear Mr. Bunpei, please, can You suggest to Mr. Chiaki to read these my heavens wishes with a lot of patience and if possible
hypothetically consider to include them in the next transport release: (1.st. wish) a native reading-decoding
(sabre dac can do) the 1bit DSD stream: 2,8MHz and 5,6MHz from Korg MR2000S master recorder.
The most life-like - (1bit DSD 5,6MHz) - realistic digital recording - playback to date, noticeably better than DXD,
from my own experiences in live performance recordings.
All the digital media playback to analogue output playback from DAC should be Auto-Upsampled to highest Fs avail.
from the DAC can send to its output. (2.nd. vish) Beside current HC SD Card reader, it is a really
another big heavens wish to include the 2.nd media storage option - an integrated SATA controller with
external Esata or USB2-3 connector for the ultimate HD Content-storage of more than 500GB--1TB on a single
2,5 inch HDD. This way it'll become an ultimate Stand Alone & the highest quality digital audio playback source.
- If this would happen in the proceeding time of my life, than I'd be one of the happiest music lovers on the planet.
- Please put me on the waiting list for the any kind future-release of the Mr. Chiaki work to be avail. to the public, transport . . . dac . . .
@Bunpei,
Excellent thread
Dear Mr. Bunpei, please, can You suggest to Mr. Chiaki to read these my heavens wishes with a lot of patience and if possible
hypothetically consider to include them in the next transport release: (1.st. wish) a native reading-decoding
(sabre dac can do) the 1bit DSD stream: 2,8MHz and 5,6MHz from Korg MR2000S master recorder.
The most life-like - (1bit DSD 5,6MHz) - realistic digital recording - playback to date, noticeably better than DXD,
from my own experiences in live performance recordings.
All the digital media playback to analogue output playback from DAC should be Auto-Upsampled to highest Fs avail.
from the DAC can send to its output. (2.nd. vish) Beside current HC SD Card reader, it is a really
another big heavens wish to include the 2.nd media storage option - an integrated SATA controller with
external Esata or USB2-3 connector for the ultimate HD Content-storage of more than 500GB--1TB on a single
2,5 inch HDD. This way it'll become an ultimate Stand Alone & the highest quality digital audio playback source.
- If this would happen in the proceeding time of my life, than I'd be one of the happiest music lovers on the planet.
- Please put me on the waiting list for the any kind future-release of the Mr. Chiaki work to be avail. to the public, transport . . . dac . . .
Hi, I posted how to read and play DSF file, here.
http://www.diyaudio.com/forums/digi...-system-dsf-player-usb-ddc-dsd-amplifier.html
http://www.diyaudio.com/forums/digi...-system-dsf-player-usb-ddc-dsd-amplifier.html
Bunpei brought Chiaki's latest work to our listening room yesterday. It consists of an SD transport with an add-on board, and a Caprice DAC with an external expansion board. The Caprice has been hacked so that its internal clock is bypassed.
Communication between the DAC and transport is via HDMI cable.
The expansion board inside the Caprice contains two master oscillators which clocks both the SD Transport and Caprice DAC at once.
These oscillators (which were recommended by NDK) are a custom order, and cost something like JPY 13,000 each. Since two oscillators are needed, it is obvious that this will not be cheap!
Still, I am reasonably sure that this is the best-sounding digital setup that I have experienced to date. Yes, I have heard BAD, dCs, Weiss, Accuphase and many other digital setups, and I have also listened to various mastering studio-grade digital installations, but IMHO Chiaki's latest work sets a high bar for other manufacturers to overcome.
Kudos also to Shin Nakagawa of Fidelix, who masterminded the Caprice DAC. I do not think that Chiaki would have obtained as good results if he had used a different DAC as a starting platform.
Given the content and engineering, this combination cannot be cheap, and the pricing may take it out of consideration for nearly all DIYers. Nonetheless, IMHO this is a setup that can challenge any digital audio setup available, regardless of price, and I would be quite surprised if it can be bested.
We will be sending our own SDTransport and Caprice to Chiaki for his modifications. Believe me, we felt that we had no alternative to doing so, after hearing what the SDTransport and Caprice could do.
FWIW, our Caprice is slightly different in key parameters from the standard Caprice (which is what Chiaki used), and the ES9018 registers of the Caprice were slightly differently to what we have found to sound the best, so I suspect we will get even better sonic results than with Chiaki's test unit (^o^).
Incidentally, the increase in sonic performance can be noted on 44.1kHz or 48kHz material as well as 96kHz, 192kHz or 352.8kHz, on 16-bit as well as 24-bit. You do not need to limit your listening to high-rez to appreciate what the SDTrans and Caprice combo can do.
Strongly recommended, not only to DIYers, but High-End digital audiophiles in general.
kind regards, jonathan carr
Communication between the DAC and transport is via HDMI cable.
The expansion board inside the Caprice contains two master oscillators which clocks both the SD Transport and Caprice DAC at once.
These oscillators (which were recommended by NDK) are a custom order, and cost something like JPY 13,000 each. Since two oscillators are needed, it is obvious that this will not be cheap!
Still, I am reasonably sure that this is the best-sounding digital setup that I have experienced to date. Yes, I have heard BAD, dCs, Weiss, Accuphase and many other digital setups, and I have also listened to various mastering studio-grade digital installations, but IMHO Chiaki's latest work sets a high bar for other manufacturers to overcome.
Kudos also to Shin Nakagawa of Fidelix, who masterminded the Caprice DAC. I do not think that Chiaki would have obtained as good results if he had used a different DAC as a starting platform.
Given the content and engineering, this combination cannot be cheap, and the pricing may take it out of consideration for nearly all DIYers. Nonetheless, IMHO this is a setup that can challenge any digital audio setup available, regardless of price, and I would be quite surprised if it can be bested.
We will be sending our own SDTransport and Caprice to Chiaki for his modifications. Believe me, we felt that we had no alternative to doing so, after hearing what the SDTransport and Caprice could do.
FWIW, our Caprice is slightly different in key parameters from the standard Caprice (which is what Chiaki used), and the ES9018 registers of the Caprice were slightly differently to what we have found to sound the best, so I suspect we will get even better sonic results than with Chiaki's test unit (^o^).
Incidentally, the increase in sonic performance can be noted on 44.1kHz or 48kHz material as well as 96kHz, 192kHz or 352.8kHz, on 16-bit as well as 24-bit. You do not need to limit your listening to high-rez to appreciate what the SDTrans and Caprice combo can do.
Strongly recommended, not only to DIYers, but High-End digital audiophiles in general.
kind regards, jonathan carr
+ DSD . . . . post #354
If ever p#354happened, than this would be the best & last product in Digital audio history with a pure lfe-like sound as a target, fare away form current fashion HiEnd agenda.
Let's all together summarize the pure positive energy and by anybody available any kind of individual help to Mr. Chiaki-san in any possible way that he can realize this-his masterpiece of Art. so all of pure sound minded-oriented individuals will have a chance to enjoy a really pure art of elementary music-sound.
Andrew
If ever p#354happened, than this would be the best & last product in Digital audio history with a pure lfe-like sound as a target, fare away form current fashion HiEnd agenda.
Let's all together summarize the pure positive energy and by anybody available any kind of individual help to Mr. Chiaki-san
in any possible way that he can realize this-his masterpiece of Art. so all of pure sound minded-oriented individuals will have a chance to enjoy a really pure art of elementary music-sound. Andrew
I'm sorry very sorry for my non-immediate replies.
1. DSD play functionality on SDTrans384
Since spring this year, some SDTrans384 users including me have requested the functionality to Chiaki. I believe this is already on his development agenda though the schedule is not planned yet. In this case, only a new version of FPGA configuration ROM and new version of MCU firmware will enable the play of DSD64 and DSD128.
To Koon,
Thank you very much for your kind link on useful information on DSD player development. I told it to Chiaki and I think he has read those posts on your thread.
I always respect your pioneering projects and generous and open attitude for publishing your successful work on your web site.
2. SDTrans384-Sync option and ES9018 based-DAC
To jcarr,
I'd like to express our gratitude to your comments.
As the purpose of the experiment is to demonstrating effectiveness of synchronous master clocking scheme for ES9018 without any PLL mechanism, I hope more people can try the scheme in rather easy and your own ways.
The reason why Chiaki adopted 90MHz clocks is that he sticks to successful 384 kHz/32 bit play without any noise on oversampling mode with "the lowest" DPLL bandwidth setting. However, if you need no oversampling mode, you can try synchronous master clocking at lower frequencies.
Though Chiaki has not planned his next release of the SDTrans384-Sync option, he seems to begin to consider it.
1. DSD play functionality on SDTrans384
Since spring this year, some SDTrans384 users including me have requested the functionality to Chiaki. I believe this is already on his development agenda though the schedule is not planned yet. In this case, only a new version of FPGA configuration ROM and new version of MCU firmware will enable the play of DSD64 and DSD128.
To Koon,
Thank you very much for your kind link on useful information on DSD player development. I told it to Chiaki and I think he has read those posts on your thread.
I always respect your pioneering projects and generous and open attitude for publishing your successful work on your web site.
2. SDTrans384-Sync option and ES9018 based-DAC
To jcarr,
I'd like to express our gratitude to your comments.
As the purpose of the experiment is to demonstrating effectiveness of synchronous master clocking scheme for ES9018 without any PLL mechanism, I hope more people can try the scheme in rather easy and your own ways.
The reason why Chiaki adopted 90MHz clocks is that he sticks to successful 384 kHz/32 bit play without any noise on oversampling mode with "the lowest" DPLL bandwidth setting. However, if you need no oversampling mode, you can try synchronous master clocking at lower frequencies.
Though Chiaki has not planned his next release of the SDTrans384-Sync option, he seems to begin to consider it.