Lossless SD-card player

Status
Not open for further replies.
To build on Gobble's idea, on v 2.0 why not just build the display into a remote like squeezebox duet and remove the display and circuitry from the chassis. The chassis--being Class II double insulated already--would be less expensive to produce. Sitting 9 feet away from my audio rig--like most people I suspect--it's hard to read a display anyway. It would have to be different though as the duet controls the PC which sends the wav file via ethernet to the squeezebox.
 
Hi JBdV,

I am interested to connect your SD transport to my NAD M2 (fully digital amp) , when do you think your new version of the SD Transport will be ready?

I was first planning to use SPDIF interface and option to slave the source. In practice, lots of problems occur when attempting to slave the source. One of these is high SPDIF jitter at the DAC, despite slaving the source, another problem is high jitter at the source.

What I am looking for is low jitter at both, DAC and source, while maintaining compatibility with SPDIF. Main problem with SPDIF is combining data and timing signals. This in turn makes it very difficult to extract low jitter timing signals at the receiver.

This lead to development of let's call it ESPDIF (Enhanced SPDIF) interface that I am planning to use in the new SD-transport.

The idea is to split both data and timing signals prior to feeding them into an interlink. ESPDIF uses dual shielded twisted pairs (RS422 interface), one for the existing SPDIF signal, the other for the master clock (timing). Third shielded twisted pair (power supply) can be added to power the synchronous reclocker. I choose RS422 (large bandwidth) in order to maintain low jitter (master clock) and preventing timing errors that could lead to data corruption.

At the end of the ESPDIF interlink, a synchronous SPDIF reclocker is used to attenuate SPDIF jitter prior to feeding it into the connected device. Both coax or Toslink (real glass fibre) interlinks can be used for the short! interlink between SPDIF reclocker and connected device.

The SPDIF reclocker can be kept very simple, cheap and compact. It could have the size of a small connector.

If one can modify the connected device, the masterclock signal can be used for local I2S signal reclocking.


Ultimate performance can be had by integrating both SD-transport and DAC, using 32 bits / frame I2S interface. I2S outputs are available inside the SD-transport for this purpose.

The SD-transport will run on a balanced master clock, the same as used in the final TDA1541A-MK4 module.
 
I am not sure if I can compeltely follow what you want to do.
A simple block diagram might be helpful.

I think my amp (more a sort of Power Dac) is a common example as most of the DACs around regarding connectivity.
It has coax as well as Toslink input and for now I have no intentions to mod it internally regarding I2S (warranty)

Unfortenately it has no separate clock (trigger) input
 
Hi JBdV,

I am not sure if I can compeltely follow what you want to do.
A simple block diagram might be helpful.

The source (SD-transport) outputs both, SPDIF and clock signal, each on a separate twisted pair interlink. This interlink can be up to say 10 meters.

On the other side we have jitter on the SPDIF signal because SPDIF contains both data and timing information.

But we also have a relatively clean clock signal because there is only timing information on the clock interlink.

At the end of both interlinks we place a small box (synchronous reclocker) that attenuates the jitter that built up along the SPDIF interlink. So the reclocker simply reclocks the complete SPDIF signal.

The output of this box is a clean low jitter SPDIF signal, either Toslink or Coax. It contains considerable lower jitter compared to a conventional Toslink or coax SPDIF interlink of this length.

Now we can use shortest possible Toslink or Coax interlink to feed this cleaned-up SPDIF signal into DAC or power amp. This short interlink ensures lowest possible SPDIF interlink jitter.
 
The NAD M2 has Digital Coax, Optical as well as AES/EBU inputs

DIGITALE INGANG (COAXIAL, OPTICAL, AES/EBU)
Ingangsimpedantie 75Ω (coaxiale)
110Ω (AES/EBU)
Monsterfrequentie 32kHz tot 192kHz
Frequentierespons ±0,5dB (ref. 20Hz - 20kHz)-3dB (ref. 96kHz)
Kanaalscheiding >90dB (ref. 10kHz, 4Ω, 1/3 nominaal vermogen)
 
New SD-transport firmware was developed providing 64 bits / frame I2S output. This enables connecting most popular DAC chips including the Sabre DAC through I2S interface.

The product-brief from ESS says "The SABRE32 Reference audio DAC’s 32-bit Hyperstream architecture can handle full 32-bit PCM data via I2S input, as well as DSD or SPDIF data." So to me it seems the Sabre dac can only handle 32 bits. Could you explain how it will handle 64 bits?

John, i've already asked you by email if one clock (tentlabs XO) is able to serve as the masterclock for both the sd-transport and my TP Opus dac (wm8741). Can i do this, or do i need a reclocking circuit like in your sd-card player?

Bartel
 
Hi Telrab,

The SD-transport can run on a 11.2896 MHz Tent clock that also feeds the DAC (slaved transport).

The existing SD-transport only supported 32 bits / frame (16 bit L, followed by 16 bit R).

This was no problem for the TDA1541A DAC module or TDA1543, however, many other DACs or digital amplifiers require 64 bits / frame (16 ... 32 bits L, followed by 16 ... 32 bits R). Unused bits (16 bit source for example) are usually set to zero.

This means that the first SD-player version was not able to drive these devices through I2S, nor could it drive a SPDIF converter to generate 64 bits / frame output.
 
Hi folks.

Just to let you know.

We compared the SD-Player against an SB Touch - Anagram-384khz converter (sold by Silvercore) on a DIY-fair in Germany (Hifi-Music-World 2010 Stuttgart) a couple of weeks ago.
I'had been invited to present my modified version of the SB-Touch as a Transport on a Silvercore/Bastanis system on that fair.

This system IMO sounded pretty good. The sound and quality ofBastanis speakers I'm used to ( I still own a pair of Apollos) .
I was surprised how well the Silvercore products were performing.
The overall system sounded IMO pretty good - revealing slightest details, textures, timbre etc. - and this on a show.

We did the comparison of both digital solutions right after dinner after the first exhibition day. 10 hours of music were not enough. 😉

As far as I understood, we've been testing a slightly modified SD-player. It was hooked up on a passive transformer pot of Silvercore.

Just to make it short:

The SD-player is extremely sensitive to incoming distortions or it it is outputing those.
There were clearly audible distortions in the air. The famous "towel in front of the
speakers".
This I considered a pretty awful performance. At this level the Touch/Anagram combo was performing ways better.

Luckily we had Robert Bastani with his great Bastanis Cables in the team.
His cable has some special HF filtering in place. Robert came up with the idea of applying his cable.

Things changed after applying that one.

Suddenly a nice stage were present. And everything else sounded just right.

The most obvious characteristic of the SD-player was the lack of "inner shiver".
The overall tonal character was also pretty good. At this point I wasn't sure
anymore which DAC was leading the competition. To be honest I do think
the SD-player passed the Touch-Anagram combo on certain aspects.

Another issue seemed to be the passive, low voltage, high impedance output stage of the SD-player. It really needs a perfectly matching amp/pre-amp to perform best otherwise you'll immediately realize the lack of energy. It tends to get a bit boring. That's nothing new though.

Last but not least. Operating the SD-Player ( as register a printed sheet of paper with album titles on it - not even tracks - were available ) was a pretty "cute" attempt to control your collection. In comparison to my iPad controlled iPeng/SB database
handling, the SD-card handling must be considered even worse compared to good old CDs.

Bottom line:
The SD-player must be called a "Diva". It needs a pretty pampered environment to sound best. It very easily gets off track. However, in a perfectly set environment it'll be able to outperform quite some other DACs out there. That has been John's goal all the time.

Handling your collection on that SD-player is a pretty awful experience once you're used to full text search functions and album selection by coverart over a couple hundred of albums on your iPad. Meanwhile this fact got a higher priority then squeezing out the last bit of sound of your DAC in my case.

I do think that John got a story to sell for a little niche market of 44.1/16-sound-only-enthusiasts out there.


Cheers
 
Last edited:
Hi Soundcheck,

As far as I understood, we've been testing a slightly modified SD-player. It was hooked up on a passive transformer pot of Silvercore.

The SD-player was a MK4 with 2 current buffer JFETs removed and it didn't perform optimally.

The latest DAC is the TDA1541A-MK5 with following mods:

- 2.8224 MHz DEM clock (smoother sound, more detail).
- Discrete voltage regulators replacing the 78XX and 79XX regulators (much better refinement and flow).
- Modified hybrid coupling cap (MKT / tin foil combination), V-cap TFTF and DC-coupling was used as reference (greatly reduced grain, much better flow).
- Filter caps at the TDA1541A analogue output pins (reduces bandwidth before signal is fed to connected circuits like I/V converter).
- Power MOSFET current buffer.
- Much higher value bias resistors and extra filter for -15V rail.


Another issue seemed to be the passive, low voltage, high impedance output stage of the SD-player. It really needs a perfectly matching amp/pre-amp to perform best otherwise you'll immediately realize the lack of energy. It tends to get a bit boring. That's nothing new though.

TDA1541A-MK4 / TDA1541A-MK5 output impedance equals 500 Ohms, this is low enough to drive a 10K load like the silvercore TVC, absolutely no problem here. Problem is that the output is tapped directly across the passive I/V resistor for highest possible resolution (no extra semiconductors in the signal path). This however puts the connected load (highly inductive TVC) in parallel with the passive I/V resistor. This results in significant performance degradation and possible RF oscillation.

The RF oscillation (distorted awful sound) can be stopped by paralleling a capacitor (filter cable). By now the once almost pure resistive I/V resistor is converted into a RLC parallel resonance circuit and performance goes down the drain.

When the MK4 / MK5 is connected to a (wire wound) 10K pot (resistive), I/V conversion works optimal and resolution is maximal. This is the configuration I currently use.

With the MK5, the power amplifier needs to be highly transparent as the slightest flaws will now become audible. So the audible distortion is not caused by the DAC but by the power amplifier. This forced me to design yet another power amplifier that matches TDA1541A-MK5 resolution and transparency. It is a balanced 4-MOSFET Circlotron power amp (attached picture) that offers 25W rms in 8 Ohms with approx. 470mV rms input signal (DAC). With 700mV rms input it soft clips like a tube amplifier (no irritating distortion during clipping). The slight distortion that was still audible with various tube and transistor power amplifiers is now completely gone.

Last but not least. Operating the SD-Player ( as register a printed sheet of paper with album titles on it - not even tracks - were available ) was a pretty "cute" attempt to control your collection. In comparison to my iPad controlled iPeng/SB database
handling, the SD-card handling must be considered even worse compared to good old CDs.

The latest ECDSD utility prints album artwork, CD number, CD title, track number, track duration, and track name of all CDs stored on a SD-card. Like I mentioned before, the SD-transport was designed for ultimate performance, and a fancy user interface conflicts with this goal.

Bottom line:
The SD-player must be called a "Diva". It needs a pretty pampered environment to sound best. It very easily gets off track. However, in a perfectly set environment it'll be able to outperform quite some other DACs out there. That has been John's goal all the time.

At these performance levels one has to observe correct matching of equipment.

the SD-card handling must be considered even worse compared to good old CDs.

I am glad I don't have to use CDs anymore, in practice the SD-card is so much easier. I can tell because I already used it for almost one year on a daily basis, and have been using CDs since introduction of the first CD players.
 

Attachments

  • pcbcir1.jpg
    pcbcir1.jpg
    135.7 KB · Views: 832
Hi John,.

Good to see that you're getting rid off the obvious flaws, such as 78xx regulators or poor coupling caps, slowly but surely. 😀

However -- "500R = good enough" !?!?

78xx were good enough for a long time; your 6$ coupling cap was good enough; an Alps pot ( which is heavily impacting the impedance - if used ) was good enough asf. asf. Now we're at SDP-MK5.

John what's really bothering me is your "I know it all - basta" teacher-like attitude.
Fact is - if you don't have a better solution to offer - "it's good enough".
Two weeks later you come up with a new MKx version to claim that this is good enough now. Now it needs your new amplifier which is exactly matched to the DAC output level,
which was and is supposed to be good enough.

Honestly I consider the entire discussion over here a commercial and marketing campaign for your products. You should consider ( or the moderators should consider that) to get this thread moved over to the commercial section. To me this thread is nothing else then a marketing thread for your products. Which would be OK if located at another place.


Fact is - as I said before - your products perform pretty well under certain conditions.
Though it is almost impossible to nail that performance down. As soon as some feedback hits the outer world, you're coming back with -- "Now I'm at MKx. Your comments are not valid anymore resp. I fixed them all already".
And that's going on since MK1. Honestly I would have a hard time to buy
such a product. (Of course you're not any better or worse then any other company out there - the difference is -- this is a DIY-forum and it is "your" thread.)

And 2nd when I talked about CDs prefered over your SDC handling, I was talking about the lack of booklets&song informations on your SD player.

Hint:
You might know that dbpoweramp seem to be able to apply tags to .wav files.
With a little LCD you could present song info on a display, without causing extra load.
Shouldn't be too complicated to accomplish for your brother.



Cheers
 
Soundcheck,

you say it in clear words. Unfortunately... the TVCs are oscillating 😱 😛 😉

78xx were good enough for a long time; your 6$ coupling cap was good enough; an Alps pot ( which is heavily impacting the impedance - if used ) was good enough asf. asf. Now we're at SDP-MK5.

And the 1541/1543 are still good enough, although they are a dead end for various reasons.

John what's really bothering me is your "I know it all - basta" teacher-like attitude.
Fact is - if you don't have a better solution to offer - "it's good enough".
Two weeks later you come up with a new MKx version to claim that this is good enough now. Now it needs your new amplifier which is exactly matched to the DAC output level, which was and is supposed to be good enough.

It is more a god-like attitude, John switches to "ignore mode" as soon as somebody criticises him on his designs.
And then he will or will not change his design, telling us that he found out, something else performs better.

One of many examples: I was't asonished that the wire wound pot is now presented as a cure for the selfmade problem with the ALPS pot.
However, that ALPS pot was praised earlier as the ultimate solution in that "highly critical application". People who were keen to doubt the perfomance of the ALPS pot were told why they are wrong, nothing else can be used here...

Not to forget, at these performance levels, not only the amplifier, but also the speakers are higly citical. They must be able to match the high resolution and ultimate transparency of the whole chain.
And so on and on...

Honestly I consider the entire discussion over here a commercial and marketing campaign for your products. You should consider ( or the moderators should consider that) to get this thread moved over to the commercial section. To me this thread is nothing else then a marketing thread for your products. Which would be OK if located at another place.

Agreed, despite the fact that he publishes most of his schematics.


Though it is almost impossible to nail that performance down. As soon as some feedback hits the outer world, you're coming back with -- "Now I'm at MKx. Your comments are not valid anymore resp. I fixed them all already".
And that's going on since MK1. Honestly I would have a hard time to buy
such a product.

I remember claims like "live sound" and "now crystal clear", xx DAC incarnations before the mkIV, and still ... ?
 
Last edited:
Anyone has to supose that there are little changes in the sound. And after some mods, when you arrive to a better configuration in the design, you can claim that now it is still better. So I understand John's reasons to upgrade to MK...infinite. Like you say, after all he shows the schematics. We are all smart enough to know when dealing with our own design expertise, while further informed by the forum. And last, I'm enjoying a lot building this kind of electronics, and hearing it against a supposed high end references, and comparing all.
 
Hi soundcheck,

John what's really bothering me is your "I know it all - basta" teacher-like attitude.

I don't know it all, and don't want to give this impression. My remarks are based on at least 4 years of extensive (daily) research on these projects, and I am still learning. I don't have, nor would I want to have a "I know it all - basta" teacher-like attitude". You visited me several times, so I am a bit surprised about this remark.

Good to see that you're getting rid off the obvious flaws, such as 78xx regulators or poor coupling caps, slowly but surely.

You tell me I have a "I know it all - basta" teacher-like attitude" and then you make a remark like this.

an Alps pot ( which is heavily impacting the impedance - if used ) was good enough asf. asf

Please have a look at this ("This is our solution Optimisation of the Impedance"):

http://www.silvercore.de/index.php?grundlagen-1

Now we're at SDP-MK5.

Yes we are at TDA1541A-MK5, it was your suggestion to use version numbering. I followed your advice and now that's no good either. The different versions were a direct result of feedback from diyaudio members, the goal is improving performance.

Fact is - if you don't have a better solution to offer - "it's good enough".

I plan to offer (much) better solutions. Time will tell if I succeed.

Two weeks later you come up with a new MKx version to claim that this is good enough now. Now it needs your new amplifier which is exactly matched to the DAC output level,
which was and is supposed to be good enough.

Things are indeed "good enough" for a while, then feedback from users can lead to further improvements. Since I don't care much about selling this stuff, I can shorten the intervals between different versions and can achieve improved performance within a shorter time period.

The Circlotron power amplifier is not specifically matched to the DAC output level and works very well with other digital and analogue sources. The key words are resolution, symmetry and control.

Honestly I consider the entire discussion over here a commercial and marketing campaign for your products. You should consider ( or the moderators should consider that) to get this thread moved over to the commercial section. To me this thread is nothing else then a marketing thread for your products. Which would be OK if located at another place.

I share schematics, detailed description, photographs and measurements (given the restrictions of my measuring equipment) of my projects. If my objective was selling this stuff I would not share this information at all.

Though it is almost impossible to nail that performance down.

Direct comparison with reference equipment / double blind listening tests.

Honestly I would have a hard time to buy such a product.

That's fine with me as selling these projects is not my objective.

Of course you're not any better or worse then any other company out there - the difference is -- this is a DIY-forum and it is "your" thread.

That's your opinion, based on "snapshots" of project performance under less than optimal conditions. I got quite some different feedback from other diy audio members that took the time to better match connected equipment. They also listened over longer time periods (weeks / months), and were able to compare the differences between subsequent versions.

And 2nd when I talked about CDs prefered over your SDC handling, I was talking about the lack of booklets&song informations on your SD player.

And I mentioned how the SD-user interface was implemented and that it included printed booklets of SD-card contents like attached picture shows. printed booklets have way higher resolution than LC displays and can offer a much better overview. The pdf or booklet shows CD artwork, CD number is right on top and track number, name and duration are on the left. browse the pdf or booklet, select desired CD, enter CD and track number and press play. This booklet is an essential part of the SD-player user interface.

Hint:
You might know that dbpoweramp seem to be able to apply tags to .wav files.
With a little LCD you could present song info on a display, without causing extra load.
Shouldn't be too complicated to accomplish for your brother.;

We used a separate text file placed in the CD directory. It is a piece of cake to provide text on a display, however interference levels will rise as the processor now has to perform more tasks. This was verified by noise measurements. The other problem is the size of the display, if you are sitting few meters away from the player and you don't have hawk eye vision, you won't be able to read the information on flimsy LC display. I assume most people are able to read a booklet with printed text or a pdf file on an hand held computer.

When using an ipad or cellphone to control the player, WiFi needs to be used. Since the distance between EMI source and equipment is rather small (few meters) power levels are substantial and interference is accordingly. This fact alone (EMI) will prevent best performance unless every last piece of the audio equipment, interlinks and mains distribution are fully screened against EMI. In practice this is almost not possible.
 

Attachments

  • Screenshot-SD-card 1.jpg
    Screenshot-SD-card 1.jpg
    186.5 KB · Views: 699
Status
Not open for further replies.