Banned
Joined 2002
Banned
Joined 2002
More information needed
I hadn't noticed this thread previously, but it looks like the perfect opportunity to make an absolutely cracking CD player. I looked at the website for the bits, but their serious electronic information was sparse.
Peter,
if you already have the transport, what are the part numbers of the major chips?
All,
Most CD players use I2S (can't do the fonts) internally, and this is the ideal way to connect to the DACs (forget SPDIF or AES3). Moreover, the clock can be put in the right place (by the DAC). Etc, etc.
It wouldn't be a problem to rival most $10k players. Think about it. At $10k, there aren't many buyers, so the dealer demands a larger mark-up because it takes longer to sell, and is risky. Taxes are higher. Manufacturer's sales are low, so development costs have to be defrayed over less units. In short, a $10k player can't be much better than a good Philips/Marantz.
You and I are prepared to take time, use nice components/layout, and make design choices that are commercially risky, so we can equal/surpass the commercial competition. Of course, if you add up your time, and cost it properly DIY audio is silly. But that's not why we do it...
I hadn't noticed this thread previously, but it looks like the perfect opportunity to make an absolutely cracking CD player. I looked at the website for the bits, but their serious electronic information was sparse.
Peter,
if you already have the transport, what are the part numbers of the major chips?
All,
Most CD players use I2S (can't do the fonts) internally, and this is the ideal way to connect to the DACs (forget SPDIF or AES3). Moreover, the clock can be put in the right place (by the DAC). Etc, etc.
It wouldn't be a problem to rival most $10k players. Think about it. At $10k, there aren't many buyers, so the dealer demands a larger mark-up because it takes longer to sell, and is risky. Taxes are higher. Manufacturer's sales are low, so development costs have to be defrayed over less units. In short, a $10k player can't be much better than a good Philips/Marantz.
You and I are prepared to take time, use nice components/layout, and make design choices that are commercially risky, so we can equal/surpass the commercial competition. Of course, if you add up your time, and cost it properly DIY audio is silly. But that's not why we do it...
Banned
Joined 2002
I don't know about kits...
At the risk of boring everybody, a little history is in order.
Once upon a time, 16 bits of resolution was very hard. You needed to be able to source the MSB current with LSB accuracy. This required very precise resistors. ICs aren't good at exact resistor values, but they are good at matched values, and the R2R ladder was born. However, (in 1983) this could only achieve 14 bit resolution, even with some rather cunning tweaks.
Oversampling works by looking at the area of a graph. If you only have 14 bit resolution, increasing to 16 bit resolution means reducing the area on a graph of current against time by a factor of 4. You can't do it by adjusting your current sources, but you can switch your current source on for a quarter of the time. This dumps the same charge into a capacitor as having 16 bit resolution for the entire sample period. Instantly, your 14 bit DAC achieves repeatable 16 bit resolution. A Philips masterstroke.
Later, oversampling was taken to its extreme, and 1 bit DACs are now common. Unfortunately, if you do the sums, a 1 bit DAC for CD needs to clock at 2.89GHz (44.1kHz x 2^16). That's extremely expensive, so fudges (noise shaping) are used. This is why very few DACs truly achieve even 20 bit resolution.
One of the major distortion problems occurs when the MSB switches (because of the LSB accuracy requirement). Push-pull operation, or in silicon parlance, differential operation, reduces this problem. Thus, most modern DACs naturally produce a balanced output.
Enough of history.
Modern DACs achieve linearity and naturally produce a differential output that doesn't require much filtering. Manufacturer's application notes give sufficient information to choose a DAC chip and interface it to an I2S bus. Changing one of these to balanced output actually simplifies the circuit.
Older DACs achieved linearity by individually laser trimming the R2R resistor networks. They are inherently unbalanced and can only be balanced by doubling the number of (expensive) DACs. They do not use noise shaping fudges. They are frighteningly expensive. Because the technology is older, the analogue bits of the manufacturer's application notes do not take into account advances in op-amp design or digital CMOS switching speeds (which reduce jitter).
Finally, converting from analogue to digital retains information, so the concept of distortion in the voltage (or current) domain is replaced by the concept of distortion in the time domain (jitter). A DAC converts distortion in the time domain into distortion in the voltage domain. This is why there are so many clock tweaks.
There is nothing to stop us pinching the best clock ideas, building them next to the DAC of our choice, and adding a little analogue circuitry to beat the bejasus out of anything else that ideally suits our own specific needs. This is why I don't know about kits...
At the risk of boring everybody, a little history is in order.
Once upon a time, 16 bits of resolution was very hard. You needed to be able to source the MSB current with LSB accuracy. This required very precise resistors. ICs aren't good at exact resistor values, but they are good at matched values, and the R2R ladder was born. However, (in 1983) this could only achieve 14 bit resolution, even with some rather cunning tweaks.
Oversampling works by looking at the area of a graph. If you only have 14 bit resolution, increasing to 16 bit resolution means reducing the area on a graph of current against time by a factor of 4. You can't do it by adjusting your current sources, but you can switch your current source on for a quarter of the time. This dumps the same charge into a capacitor as having 16 bit resolution for the entire sample period. Instantly, your 14 bit DAC achieves repeatable 16 bit resolution. A Philips masterstroke.
Later, oversampling was taken to its extreme, and 1 bit DACs are now common. Unfortunately, if you do the sums, a 1 bit DAC for CD needs to clock at 2.89GHz (44.1kHz x 2^16). That's extremely expensive, so fudges (noise shaping) are used. This is why very few DACs truly achieve even 20 bit resolution.
One of the major distortion problems occurs when the MSB switches (because of the LSB accuracy requirement). Push-pull operation, or in silicon parlance, differential operation, reduces this problem. Thus, most modern DACs naturally produce a balanced output.
Enough of history.
Modern DACs achieve linearity and naturally produce a differential output that doesn't require much filtering. Manufacturer's application notes give sufficient information to choose a DAC chip and interface it to an I2S bus. Changing one of these to balanced output actually simplifies the circuit.
Older DACs achieved linearity by individually laser trimming the R2R resistor networks. They are inherently unbalanced and can only be balanced by doubling the number of (expensive) DACs. They do not use noise shaping fudges. They are frighteningly expensive. Because the technology is older, the analogue bits of the manufacturer's application notes do not take into account advances in op-amp design or digital CMOS switching speeds (which reduce jitter).
Finally, converting from analogue to digital retains information, so the concept of distortion in the voltage (or current) domain is replaced by the concept of distortion in the time domain (jitter). A DAC converts distortion in the time domain into distortion in the voltage domain. This is why there are so many clock tweaks.
There is nothing to stop us pinching the best clock ideas, building them next to the DAC of our choice, and adding a little analogue circuitry to beat the bejasus out of anything else that ideally suits our own specific needs. This is why I don't know about kits...
Re: More information needed
I see some fresh enthusiasm.
I am also positive that with proper arrangement that transport can match most other high end players.
Check page 7 of the following link for chips: http://www.daisy-laser.nl/homeoptics/VAU1254bis.pdf
EC8010 said:
I see some fresh enthusiasm.
I am also positive that with proper arrangement that transport can match most other high end players.Check page 7 of the following link for chips: http://www.daisy-laser.nl/homeoptics/VAU1254bis.pdf
The DACs are %"£
Don't even think of using them. Noise-shaping chips specify their performance using the "A" weighting filter. This use is distinctly questionable. "A" wieghting vaguely corresponds to the sensitivity of an average human ear at a particular sound pressure level (SPL). The measurements were made some time ago, and average human hearing has deteriorated due to clubbing and personal stereos. (I kid you not, an AES study in the last decade was unable to find "typical" hearing amongst undergraduates.) 100dB(A) S/N isn't all that good.
Who cares? Interfaced to a decent DAC, far better performance can be achieved.
I understand DAC specifications, so I can look at the SAA7324, but I'm not a micro expert, so someone else needs to look at the 89C51RC+ controller. It probably only controls functionality, but it might control data recovery, which could turn out to be important.
Don't even think of using them. Noise-shaping chips specify their performance using the "A" weighting filter. This use is distinctly questionable. "A" wieghting vaguely corresponds to the sensitivity of an average human ear at a particular sound pressure level (SPL). The measurements were made some time ago, and average human hearing has deteriorated due to clubbing and personal stereos. (I kid you not, an AES study in the last decade was unable to find "typical" hearing amongst undergraduates.) 100dB(A) S/N isn't all that good.
Who cares? Interfaced to a decent DAC, far better performance can be achieved.
I understand DAC specifications, so I can look at the SAA7324, but I'm not a micro expert, so someone else needs to look at the 89C51RC+ controller. It probably only controls functionality, but it might control data recovery, which could turn out to be important.
cases building etc.
Hi,
Have any of you guys (other than peter obviously) thought about how you're going to implement the transport (cases, connection etc)?
Also using things like mdf for the case instead of a prebuilt case?
Are there any benefits from using alminium instead of mdf in places where sound could be affected?
Since some of us are trying to make a really good player but aren't capable of producing cases like peter's, I though it would be a good idea if we shared opinions and even plans of how to do the work:
1) So that someone with a reasonable ability for practical work can find an easy method to build the player.
2) So that people who are still unsure about the building are given the confidence and proper method and info to build a really good sounding player.
I'm talking about an easy and effective way of doing things, although peter's methods and designs are sound, not all of us could copy them, so for someone like myself who wants a player that looks and sounds good can build it. Anything could help here, stuff like possible cases etc, or even contacting a case supplier over the internet, and one of the more skilful amongst us to use panel design software to design a case on which we could do a further group buy. For instance we could have the hole cut in the top and also drilling/cutting for rca,12s, spdif, power inlet and the display already done. It shouldn't be overly expensive to do this, perhaps someone like
http://www.par-metal.com/
Thanks
Raj
Hi,
Have any of you guys (other than peter obviously) thought about how you're going to implement the transport (cases, connection etc)?
Also using things like mdf for the case instead of a prebuilt case?
Are there any benefits from using alminium instead of mdf in places where sound could be affected?
Since some of us are trying to make a really good player but aren't capable of producing cases like peter's, I though it would be a good idea if we shared opinions and even plans of how to do the work:
1) So that someone with a reasonable ability for practical work can find an easy method to build the player.
2) So that people who are still unsure about the building are given the confidence and proper method and info to build a really good sounding player.
I'm talking about an easy and effective way of doing things, although peter's methods and designs are sound, not all of us could copy them, so for someone like myself who wants a player that looks and sounds good can build it. Anything could help here, stuff like possible cases etc, or even contacting a case supplier over the internet, and one of the more skilful amongst us to use panel design software to design a case on which we could do a further group buy. For instance we could have the hole cut in the top and also drilling/cutting for rca,12s, spdif, power inlet and the display already done. It shouldn't be overly expensive to do this, perhaps someone like
http://www.par-metal.com/
Thanks
Raj
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.