Hi 454Casull,
Yes, it even affects sound quality through RCA and speaker interlinks, as the whole listening room and all equipment installed in it are affected, not only the digital audio source. This would require screening the whole audio set, interlinks included, this isn't very practical.
Yes, it even affects sound quality through RCA and speaker interlinks, as the whole listening room and all equipment installed in it are affected, not only the digital audio source. This would require screening the whole audio set, interlinks included, this isn't very practical.
Hi phofman,
Yes I an referring to ac ground loops. The best isolation transformers still have a certain capacitance between both primary and secondary windings (up to a few pF), this provides a low impedance for RF / HF signals.
That's why I decided to use Toslink, as this optical interface doesn't have this problem.
Yes I an referring to ac ground loops. The best isolation transformers still have a certain capacitance between both primary and secondary windings (up to a few pF), this provides a low impedance for RF / HF signals.
That's why I decided to use Toslink, as this optical interface doesn't have this problem.
Hi soundcheck,
Toslink (SPDIF) always contains jitter, jitter amplitude isn't the biggest problem, it's jitter frequency spectrum. Using Toslink you have at least 2 jitter sources, source jitter spectrum and PLL jitter spectrum. When (synchronous) reclocking is used, the master clock jitter spectrum is added to the mix. This is basically a no-win situation that always includes making compromises.
Sure improvements can be made by using better Toslink cables (real glass fibre interlinks), and faster optical transmitters / receivers. But performance will always be less than optimal.
Here is a photograph of the recovered master clock from a dual SPDIF receiver (each receiver has different loop filter), both run in slave-clock mode. They are followed by a micro controller-based FFL VCXO. The scope is put on highest time resolution, and the oscillogram shows one of the pulses that follows the trigger pulse. The signal is measured directly at the DAC chip bit clock input. The electron-beam trace is blurred (photo camera). The actual trace thickness is less than a tenth of a millimeter.
This is the (worst case) starting point, this master clock still has way too high jitter, it's barely visible on the oscilloscope, set at maximum time resolution, yet it already has big audible effect on sound quality. I already use a NOS DAC that's least sensitive to timing jitter.
In order to maintain low (bit) clock jitter levels at the DAC chip internal circuits, I attenuate the master clock amplitude to approx. 400mV, and create a low load impedance (approx. 33 Ohm), this is possible with both TDA1543 / TDA1541A as these have TTL-compatible current steering inputs.
Still this doesn't solve the jitter spectrum "issue". Using a single master clock for both source and DAC, that are placed close together, and are connected through I2S is a good starting point.
Toslink (SPDIF) always contains jitter, jitter amplitude isn't the biggest problem, it's jitter frequency spectrum. Using Toslink you have at least 2 jitter sources, source jitter spectrum and PLL jitter spectrum. When (synchronous) reclocking is used, the master clock jitter spectrum is added to the mix. This is basically a no-win situation that always includes making compromises.
Sure improvements can be made by using better Toslink cables (real glass fibre interlinks), and faster optical transmitters / receivers. But performance will always be less than optimal.
Here is a photograph of the recovered master clock from a dual SPDIF receiver (each receiver has different loop filter), both run in slave-clock mode. They are followed by a micro controller-based FFL VCXO. The scope is put on highest time resolution, and the oscillogram shows one of the pulses that follows the trigger pulse. The signal is measured directly at the DAC chip bit clock input. The electron-beam trace is blurred (photo camera). The actual trace thickness is less than a tenth of a millimeter.
This is the (worst case) starting point, this master clock still has way too high jitter, it's barely visible on the oscilloscope, set at maximum time resolution, yet it already has big audible effect on sound quality. I already use a NOS DAC that's least sensitive to timing jitter.
In order to maintain low (bit) clock jitter levels at the DAC chip internal circuits, I attenuate the master clock amplitude to approx. 400mV, and create a low load impedance (approx. 33 Ohm), this is possible with both TDA1543 / TDA1541A as these have TTL-compatible current steering inputs.
Still this doesn't solve the jitter spectrum "issue". Using a single master clock for both source and DAC, that are placed close together, and are connected through I2S is a good starting point.
Attachments
John, whilst I respect your designs I think this project is a bit outdated before it even is in production.
No Flac support "because SD cards can be rewritten many times" will end up in useless time consuming erasing and rewriting the SD cards. A repetative and frustrating experience ! The more data fits on the card the less erasing and copying will be necessary. In this case SD cards will be a kind of replacement for cd's but with better sound quality...
As a sidenote: I just returned to my old friend Hifidelio and despite the fact that it is not a high end player the convenience is terrific. Operation is a breeze and it plays Wav, FLAC and MP3. With a 160 gb disk it is a one time copying action and you're set.
So please rethink this project, it must be possible with a "larger than SD" notebook harddisk or even a SSD.
No Flac support "because SD cards can be rewritten many times" will end up in useless time consuming erasing and rewriting the SD cards. A repetative and frustrating experience ! The more data fits on the card the less erasing and copying will be necessary. In this case SD cards will be a kind of replacement for cd's but with better sound quality...
As a sidenote: I just returned to my old friend Hifidelio and despite the fact that it is not a high end player the convenience is terrific. Operation is a breeze and it plays Wav, FLAC and MP3. With a 160 gb disk it is a one time copying action and you're set.
So please rethink this project, it must be possible with a "larger than SD" notebook harddisk or even a SSD.
jean-paul said:
Hard disks and SSDs have controllers in them, which will add latency and jitter. I think the point of this project is to eliminate jitter and all the other craps that modern electronic introduces. I think this project is analogous to playing vinyl. Absolutely pain in the butt, but it sounds great.
On the other hand, and this mostly refers to the DAC connected to the SD player, I like a switch for I2S between the best quality SD mode, and a lower quality auxiliary input from slimserver duet.
Those who want to play their music through a SSD or whatever else with 1TB capacity can buy a Squeezebox Duet or a TEAC WAP. The WAP2200 starts at 120 Euros.
I myself don't mind using SD that contains 99 CDs. I don't think I need more. I don't even mind using 2GB SD as Koon's Ultimate Player is restricted to. Large storage capacity is not why I want this player. The only reason for going to this trouble is that it plays original WAV codes without error, and without all the reclocking troubles associated with SPDIF. Simply put, I want the perfect digital source; all solid state, no moving parts, and no lossy codes.
If you want convenience, buy an iPod.
Patrick
I myself don't mind using SD that contains 99 CDs. I don't think I need more. I don't even mind using 2GB SD as Koon's Ultimate Player is restricted to. Large storage capacity is not why I want this player. The only reason for going to this trouble is that it plays original WAV codes without error, and without all the reclocking troubles associated with SPDIF. Simply put, I want the perfect digital source; all solid state, no moving parts, and no lossy codes.
If you want convenience, buy an iPod.
Patrick
99 cd's in wav would be around 70 gb. That is a pretty expensive SD card. An SSD will be cheaper and it is also solid state....
My point is that convenience and good sound quality can go together which would make this a truly exceptional product that stands out in the crowd. Why go for something that sounds very good but is a pain to use and that has limited storage capacity ? And which will take months of development for a very small market so it has a big chance to be a time consuming loss creating project. If a project is limited from the start it will be dead when it is finally born. Especially storage capacity is an important factor for user satisfaction.
Good sound quality and convenience also do not exclude eachother. I will stick with my Hifidelio I guess. An Ipod is very convenient and it would be very nice if Apple would make one in 19" format for home use with a very good DAC ( inside ! ). Or if Hermstedt would only have made a HF with a better DAC...
BTW Patrick, FLAC is not a lossy format. It is WAV but compressed without loss.
My point is that convenience and good sound quality can go together which would make this a truly exceptional product that stands out in the crowd. Why go for something that sounds very good but is a pain to use and that has limited storage capacity ? And which will take months of development for a very small market so it has a big chance to be a time consuming loss creating project. If a project is limited from the start it will be dead when it is finally born. Especially storage capacity is an important factor for user satisfaction.
Good sound quality and convenience also do not exclude eachother. I will stick with my Hifidelio I guess. An Ipod is very convenient and it would be very nice if Apple would make one in 19" format for home use with a very good DAC ( inside ! ). Or if Hermstedt would only have made a HF with a better DAC...
BTW Patrick, FLAC is not a lossy format. It is WAV but compressed without loss.
jean-paul said:
Jean-Paul
Since the bginnings of audio reproduction we have been using some kind of mechanical support: rollers, 78's, 45's, LP, magnetic tapes, CD, DVDs, hard disk, etc.
I always dreamed of getting rid of all this stuff with their associated accessories: needles, phono cartriges, tape heads, laser pickups, name it.
This project may not be perfect, but it shows the way, where we should be heading, not returning in the old footsteps.
I welcome you, John or any other interested EE to push reasearch in the direction of new concepts for audio reproduction where any mechanical support is left out for ever.
Hi Jean-Charles. An SSD is a Solid State Disk. A replacement for a harddisk with no moving parts and high reliability. Like a big memory stick with a PATA/SATA connector instead of USB.
If a SSD would be used the less fortunate can use its antique equivalent: a harddisk. Both share the same connector. So two birds with one stone. An expensive version with an SSD with no moving parts and let's say 64 gb. A cheaper version which has the same hardware except for the fact that it has a 2,5 " harddisk which has the added bonus that they are relatively cheap and much larger in size ( up to 500 gb at this moment ). Two versions with relatively small differences but somewhat larger differences in price will attract two different groups of audio afficionados.
Since the hardware and software would be equal more energy can be put in user friendliness and connectivity like LAN ( which should possible to switch off ) and a fancy GUI preferably with a OLED display. Most mediaplayers that can be bought today need a computerscreen ( how sad ) to see the contents of the harddisk.They also sound pretty mediocre but that is another story. A good display for standalone operation is necessary.
Upgrading from harddisk to SSD would be possible. Downgrading from SSD to harddisk too for those that like to run in circles ( or if they need more space for a smaller price )
If a SSD would be used the less fortunate can use its antique equivalent: a harddisk. Both share the same connector. So two birds with one stone. An expensive version with an SSD with no moving parts and let's say 64 gb. A cheaper version which has the same hardware except for the fact that it has a 2,5 " harddisk which has the added bonus that they are relatively cheap and much larger in size ( up to 500 gb at this moment ). Two versions with relatively small differences but somewhat larger differences in price will attract two different groups of audio afficionados.
Since the hardware and software would be equal more energy can be put in user friendliness and connectivity like LAN ( which should possible to switch off ) and a fancy GUI preferably with a OLED display. Most mediaplayers that can be bought today need a computerscreen ( how sad ) to see the contents of the harddisk.They also sound pretty mediocre but that is another story. A good display for standalone operation is necessary.
Upgrading from harddisk to SSD would be possible. Downgrading from SSD to harddisk too for those that like to run in circles ( or if they need more space for a smaller price )
Before SD we had Compact Flash cards and some mediaplayers used them as well. Compact Flash is 100 % compatible with the ATA standard so I guess a full-blown computer is exxagerated if you see the pretty simple mediaplayers that used the CF standard. In fact Compact Flash is superior to SD but they are physically bigger than SD.
I see, the word "full-blown" was not the right choice. There are countless SoC solutions available, supporting ATA, USB, etc. while being in fact systems capable of running an operating system, e.g. linux or BSD. It does not seem ecdesigns intends to use such complicated solution.
It was just a remark, I am not connected to the project in any way.
It was just a remark, I am not connected to the project in any way.
No problem, discussion within certain boundaries can give good ideas that can make a project more worthwhile. The way of the least resistance is to see practical objections before they even exist
Since the know-how on SoC etc. lies in the Far East I think definitions on quality should best be made by the designer and practical manufacturing and implementation by an OEM that already produces mass market stuff. it would require some very well defined quality demands but it is possible.
I do not think serious audio components with solid state storage will enter the market soon as demand is too low and a lot of people do not care a thing about quality. The large traditional companies already have problems keeping up with the market which makes chances even smaller.
Since the know-how on SoC etc. lies in the Far East I think definitions on quality should best be made by the designer and practical manufacturing and implementation by an OEM that already produces mass market stuff. it would require some very well defined quality demands but it is possible.
I do not think serious audio components with solid state storage will enter the market soon as demand is too low and a lot of people do not care a thing about quality. The large traditional companies already have problems keeping up with the market which makes chances even smaller.
I'd be surprised if reading from IDE is much more complex than reading from SD or CF. 10 years ago, people experimented with playing MP3s using the MAS3507D decoder chip. There was a guy in Alaska working with it... IIRC it only required a low end microcontroller to play from IDE.
How about those Zoom digital recorders? If they record, I assume they also play back.
For cheap lossless wav playback, I got a 2 GB USB MP3/WMA/WAV player from Staples for $20, and it will play 16bit/44k WAV files, not just mono speech-quality wavs. I've seen a similar player at Walmart that had 4 GB for another $8... that's nearly as much music as I could fit in my car CD changer 15 years ago, for 1/20 the cost. (Assuming many CDs didn't occupy the full 74 to 80 minutes). (And assuming the Walmart player also handles full 16/44 wav).
I've heard that Western Digital is selling some kind of versatile media player thing that plays from hard drives. It, or some similar gadget, can play DVD images off the hard drive, so presumably if it can handle Dolby Digital and DTS bitstreams, it can also put out faithful S/PDIF of PCM content.
How about those Zoom digital recorders? If they record, I assume they also play back.
For cheap lossless wav playback, I got a 2 GB USB MP3/WMA/WAV player from Staples for $20, and it will play 16bit/44k WAV files, not just mono speech-quality wavs. I've seen a similar player at Walmart that had 4 GB for another $8... that's nearly as much music as I could fit in my car CD changer 15 years ago, for 1/20 the cost. (Assuming many CDs didn't occupy the full 74 to 80 minutes). (And assuming the Walmart player also handles full 16/44 wav).
I've heard that Western Digital is selling some kind of versatile media player thing that plays from hard drives. It, or some similar gadget, can play DVD images off the hard drive, so presumably if it can handle Dolby Digital and DTS bitstreams, it can also put out faithful S/PDIF of PCM content.
this western digital player is called WD TV,it costs 90$in the US and 100EUR in europe,it can handle a lot of formats and it has optical digital output,this week new firmware will be out,but we are limited to known USB transfer problems with lots of jitter
Music - MP3, WMA, OGG, WAV/PCM/LPCM, AAC, FLAC, Dolby Digital, AIF/AIFF, MKA
Photo - JPEG, GIF, TIF/TIFF, BMP, PNG
Video -MPEG1/2/4, WMV9, AVI (MPEG4, Xvid, AVC), H.264, MKV, MOV (MPEG4, H.264),
MTS, TP, TS
Playlist - PLS, M3U, WPL
Subtitle -SRT (UTF-, SMI, SUB, ***, SSA
official page
http://www.westerndigital.com/en/products/products.asp?driveid=572
but,i would stick with EC's SD player just for the audio part
Music - MP3, WMA, OGG, WAV/PCM/LPCM, AAC, FLAC, Dolby Digital, AIF/AIFF, MKA
Photo - JPEG, GIF, TIF/TIFF, BMP, PNG
Video -MPEG1/2/4, WMV9, AVI (MPEG4, Xvid, AVC), H.264, MKV, MOV (MPEG4, H.264),
MTS, TP, TS
Playlist - PLS, M3U, WPL
Subtitle -SRT (UTF-, SMI, SUB, ***, SSA
official page
http://www.westerndigital.com/en/products/products.asp?driveid=572
but,i would stick with EC's SD player just for the audio part
Unfortunately it does need a tv to display the contents of the harddisk which is a showstopper IMO.
An audio device should have the possibility to operate it without a computer screen or a tv. The same counts for a keyboard and a mouse. If those are necessary one could use a pc for the same purpose which is quite the opposite direction an audiophile would go AFAIK. An audio device should not necessarily play video content as well.
But what does ECdesigns ( John ) think about this ? I feel a bit sorry for bending this thread in another direction. It was for a cause though.
An audio device should have the possibility to operate it without a computer screen or a tv. The same counts for a keyboard and a mouse. If those are necessary one could use a pc for the same purpose which is quite the opposite direction an audiophile would go AFAIK. An audio device should not necessarily play video content as well.
But what does ECdesigns ( John ) think about this ? I feel a bit sorry for bending this thread in another direction. It was for a cause though.
Anyway, after skimming the whole thread, I realize that the WD box is a different animal. If it runs Linux, any software faults are likely to be remedied by users before long, but it does need a video display, and sucks a lot of power, and doesn't support multichannel lossless audio over HDMI. I still want one for my truck, though.
One advantage of FLAC is that there's less data to transfer. Even if flash memory is infinite and free, reducing the time spent copying by 40% is useful. And possibly if you spend less time reading stuff off the card, that'll make up for the overhead of decoding the FLAC to PCM. I think FLAC is poised to replace MP3 in the long run for digital audio storage, so supporting it would drastically increase the number of potential buyers for your player.
One advantage of FLAC is that there's less data to transfer. Even if flash memory is infinite and free, reducing the time spent copying by 40% is useful. And possibly if you spend less time reading stuff off the card, that'll make up for the overhead of decoding the FLAC to PCM. I think FLAC is poised to replace MP3 in the long run for digital audio storage, so supporting it would drastically increase the number of potential buyers for your player.
Hi jean-paul,
We carefully evaluated all possible options, and the end result of this was the SD-card player.
Both SD-card capacity (up to 32Gb) and price are now getting interesting for digital audio playback.
Certain options like FLAC were deliberately left-out in order to maximize data transfer speed from SD-card to RAM buffer, and lowering processor interference levels by keeping processor load as low as possible. In order to guarantee reliable 44.1/16 playback from (slower) SD-cards, data transfer needs to be as fast and efficient as possible.
Data flow is mainly DMA-(hardware) controlled, and software intervention is reduced to absolute minimum during playback. This way, lowest possible interference levels could be achieved (we attempt to minimize interference levels since the player has to be integrated in the DAC).
The concept is limiting the amount of CDs stored on one CD-card, so the user interface can be kept as simple and basic as possible.
The SD-card is intended as medium to transport selected audio data (play-list), stored on a computer, to the stand-alone SD-card player. So SD-card capacity is of minor importance since the computer still holds all CD images.
Multiple (cheap) SD-cards could be used for quickly accessing favorite music.
I already explained the main reason for choosing this approach. It's about acieving best possible playback quality using 44.1/16.
I don't see any need for higher resolution formats as these result in higher ground-bounce (higher sample frequency) and resulting increased sample timing jitter. Most of the high quality audio sets barely manage to effectively resolve 16 bits, so using 24 bit performance still results in effective 16 bit resolution or less.
In order to maximize 44.1/16 performance, we used 32 bits / frame (only sending the 32 bits that contain data for both L and R channels). As a result of this the bit clock could be lowered to 1.4112 MHz, providing even lower ground-bounce.
Both SD-card player and DAC chip share the same low-jitter master clock, introducing only intrinsic master clock -no source clock- jitter. The latest dynamic jitter attenuation circuits, placed in all I2S signals, can further reduce output sample timing jitter into the femto-second range.
In order to maximize performance (increase resolution), I also decided to skip (pre) amplifier and conventional volume control.
The balanced output DAC runs on the SD-card player, it drives HVIV converters, these generate up to 40Vpp output signal without using amplifiers. The voltage is generated by the DAC output current, running through a variable passive I/V resistor (volume control). The voltage across these I/V resistors then directly drives the unity gain bridge power buffers. These buffers then drive the speaker. This concept eliminates the amplifiers and related distortion. The circuit can be fully DC-coupled (balanced operation) so best possible transparency can be had.
The SD-card player, DAC, and unity gain bridge power buffers can all be placed in the same housing, also eliminating analogue interlink issues. This provides a one-box digital audio playback solution.
We carefully evaluated all possible options, and the end result of this was the SD-card player.
Both SD-card capacity (up to 32Gb) and price are now getting interesting for digital audio playback.
Certain options like FLAC were deliberately left-out in order to maximize data transfer speed from SD-card to RAM buffer, and lowering processor interference levels by keeping processor load as low as possible. In order to guarantee reliable 44.1/16 playback from (slower) SD-cards, data transfer needs to be as fast and efficient as possible.
Data flow is mainly DMA-(hardware) controlled, and software intervention is reduced to absolute minimum during playback. This way, lowest possible interference levels could be achieved (we attempt to minimize interference levels since the player has to be integrated in the DAC).
The concept is limiting the amount of CDs stored on one CD-card, so the user interface can be kept as simple and basic as possible.
The SD-card is intended as medium to transport selected audio data (play-list), stored on a computer, to the stand-alone SD-card player. So SD-card capacity is of minor importance since the computer still holds all CD images.
Multiple (cheap) SD-cards could be used for quickly accessing favorite music.
I already explained the main reason for choosing this approach. It's about acieving best possible playback quality using 44.1/16.
I don't see any need for higher resolution formats as these result in higher ground-bounce (higher sample frequency) and resulting increased sample timing jitter. Most of the high quality audio sets barely manage to effectively resolve 16 bits, so using 24 bit performance still results in effective 16 bit resolution or less.
In order to maximize 44.1/16 performance, we used 32 bits / frame (only sending the 32 bits that contain data for both L and R channels). As a result of this the bit clock could be lowered to 1.4112 MHz, providing even lower ground-bounce.
Both SD-card player and DAC chip share the same low-jitter master clock, introducing only intrinsic master clock -no source clock- jitter. The latest dynamic jitter attenuation circuits, placed in all I2S signals, can further reduce output sample timing jitter into the femto-second range.
In order to maximize performance (increase resolution), I also decided to skip (pre) amplifier and conventional volume control.
The balanced output DAC runs on the SD-card player, it drives HVIV converters, these generate up to 40Vpp output signal without using amplifiers. The voltage is generated by the DAC output current, running through a variable passive I/V resistor (volume control). The voltage across these I/V resistors then directly drives the unity gain bridge power buffers. These buffers then drive the speaker. This concept eliminates the amplifiers and related distortion. The circuit can be fully DC-coupled (balanced operation) so best possible transparency can be had.
The SD-card player, DAC, and unity gain bridge power buffers can all be placed in the same housing, also eliminating analogue interlink issues. This provides a one-box digital audio playback solution.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.