Several Sony CDP's from the early 1990s received high praise from the audiophile community ... and continue to do so. The resale value of certain Sony CDPs from 1991-1994 seem to hold up ...
What they seem to have in common is Sony's own CXD2565M 1-bit (PLM) dac chip. This was an DF+DAC device that was apparently an evolution from separate DF and DAC devices [CXD1244S, CXD2552]
In any case, the vintage 1-bit Sony dacs (or complementary DF chips) are hard to find on ebay, etc. Also, not sure what the input format is. EIAJ or I2S. I can't tell from the datasheet timing diagrams.
Sony seems to have kept a tight lid on the details and distribution of these devices -- compared to Philips, AD, Burr-Brown, that is. Not sure why Sony was overly protective of these devices?
Would be interesting to make a diy dac based on CXD2565M, maybe being fed by a USB adapter.
I have a feeling that such a project -- even if everything was very optimized and cleanly powered -- would only yield so-so sound.
Something magical and dreamlike happens when the whole OEM box -- say, classic Sony or Philips CDPs , with all those manhours of R&D -- is used. Something about the dream of totality.
Also see:
https://www.datasheetbank.com/datasheet-download/340323/1/Sony/CXD2565M?v=V2
https://pdf1.alldatasheet.com/datasheet-pdf/view/46935/SONY/CXD1244S.html
https://pdf1.alldatasheet.com/datasheet-pdf/view/47027/SONY/CXD2552.html
https://www.hifiengine.com/manual_library/sony.shtml?category=cd-player
What they seem to have in common is Sony's own CXD2565M 1-bit (PLM) dac chip. This was an DF+DAC device that was apparently an evolution from separate DF and DAC devices [CXD1244S, CXD2552]
In any case, the vintage 1-bit Sony dacs (or complementary DF chips) are hard to find on ebay, etc. Also, not sure what the input format is. EIAJ or I2S. I can't tell from the datasheet timing diagrams.
Sony seems to have kept a tight lid on the details and distribution of these devices -- compared to Philips, AD, Burr-Brown, that is. Not sure why Sony was overly protective of these devices?
Would be interesting to make a diy dac based on CXD2565M, maybe being fed by a USB adapter.
I have a feeling that such a project -- even if everything was very optimized and cleanly powered -- would only yield so-so sound.
Something magical and dreamlike happens when the whole OEM box -- say, classic Sony or Philips CDPs , with all those manhours of R&D -- is used. Something about the dream of totality.
Also see:
https://www.datasheetbank.com/datasheet-download/340323/1/Sony/CXD2565M?v=V2
https://pdf1.alldatasheet.com/datasheet-pdf/view/46935/SONY/CXD1244S.html
https://pdf1.alldatasheet.com/datasheet-pdf/view/47027/SONY/CXD2552.html
https://www.hifiengine.com/manual_library/sony.shtml?category=cd-player
Looking at page 11 of the CXD2565M datasheet, it looks like the LSB is transmitted first, so then it can't be I2S.
Edit after seeing the schematic at page 12:
It's from the early 1990's and has a third-order noise shaper. Looking at page 12 of the datasheet, the two outputs per channel appear to form a differential output.
If it is a single-loop third-order single-bit design, it is a very early one. Before 1990, only very few people knew that it was possible to make single-loop, single-bit modulators of order greater than two.
Edit after seeing the schematic at page 12:
It's from the early 1990's and has a third-order noise shaper. Looking at page 12 of the datasheet, the two outputs per channel appear to form a differential output.
If it is a single-loop third-order single-bit design, it is a very early one. Before 1990, only very few people knew that it was possible to make single-loop, single-bit modulators of order greater than two.
Last edited:
Early literature about high-order single-bit single-loop sigma-delta modulators:
Robert W. Harris, US patent US4509037, application filed 1 December 1982, date of patent 2 April 1985
Kirk C.-H. Chao, Shujaat Nadeem, Wai L. Lee and Charles G. Sodini, "A high order topology for interpolative modulators for oversampling A/D converters", IEEE Transactions on Circuits and Systems, vol. 37, no. 3, March 1990, pages 309...318
As far as I know, almost no one was aware of the 1980's patent, and the 1990 article led to the noise transfer function synthesis method to design high-order single-bit sigma-delta modulators.
Robert W. Harris, US patent US4509037, application filed 1 December 1982, date of patent 2 April 1985
Kirk C.-H. Chao, Shujaat Nadeem, Wai L. Lee and Charles G. Sodini, "A high order topology for interpolative modulators for oversampling A/D converters", IEEE Transactions on Circuits and Systems, vol. 37, no. 3, March 1990, pages 309...318
As far as I know, almost no one was aware of the 1980's patent, and the 1990 article led to the noise transfer function synthesis method to design high-order single-bit sigma-delta modulators.
Many of those ES series Sony players use ONE CXD2565M per channel (two total).
I'm not sure what differentiated the early noise shapers from each other? Matsushita (MASH), Philips (Bitstream), Sony (???).
The following are my only direct experiences with early noise shapers -- holding all devices nearly on equal playing fields: combi LD/CD players.
Some of the Panasonic combi LD/CD players -- even the high end ones like my LX-900 -- used MASH. And that sounded a bit harsh. The Philips CDV-600 was a bit better -- but not as good as the CDV487/488 (7220/1541A) .
The early/mid-1990s Pioneer combi players had best best soinics .
When you don't know how to directly design a sigma-delta modulator/noise shaping loop of order greater than two (or if patents prevent you from doing so), you can make a cascade of low-order loops. For example, you make a second-order loop, subtract its input and output signals and put the difference into another first-order loop. When you then add the outputs of the two loops in exactly the right proportions, the quantization error of the second-order loop is supposed to cancel, and the overall result is third-order noise shaping. You do get more idle tone issues and an accuracy problem, as that addition with the right proportions is an addition in the analogue domain. This trick is called MASH.
What I remember from old articles about early Philips noise shapers, is that they used a cascade of a multibit sigma-delta modulator and a pulse width modulator. In the end, that resulted in a higher bit rate than they could have made with just a sigma-delta modulator with the technologies they had back then. I don't remember what the order of the modulator was (I would guess 2).
Apparently, if I understand the datasheet correctly, Sony used a third-order single-loop single-bit sigma-delta in the early 1990's. That's remarkably early. The so-called "DC dithering" is probably a trick to move the main idle tone out of the audio band when playing silence by adding an offset.
It seems as if Philips was really into the process of up- and oversampling, even with 1-bit.
Here ...
https://www.dutchaudioclassics.nl/the_evolution_of_dac_the_digital_filter/
... it is noted:
https://www.dutchaudioclassics.nl/philips_tda1547/
And even so far as to suggest another DF manuf [NPC, SM5803APT]
It's unclear how much this "meat tenderizing" process has actually "improved", even with flawless measurements and worshipful reviews of AKM or ESS or even the FPGA processes. The recipe has changed -- but just for a DIFFERENT flavor.
Adding more noise-shaping stages adds more parts and signal traces and complexity. Not all that beneficial to real-word signal integrity ... unless you then beef up the power supply, or add a nice tubed output stage. So ... back to where you started!
Here ...
https://www.dutchaudioclassics.nl/the_evolution_of_dac_the_digital_filter/
... it is noted:
On the next page (Fig. 4) shows the SAA7350 between the digital filter IC [TDA1307 I assume ] and he Bitstream TDA1547.
https://www.dutchaudioclassics.nl/philips_tda1547/
And even so far as to suggest another DF manuf [NPC, SM5803APT]
It's unclear how much this "meat tenderizing" process has actually "improved", even with flawless measurements and worshipful reviews of AKM or ESS or even the FPGA processes. The recipe has changed -- but just for a DIFFERENT flavor.
Adding more noise-shaping stages adds more parts and signal traces and complexity. Not all that beneficial to real-word signal integrity ... unless you then beef up the power supply, or add a nice tubed output stage. So ... back to where you started!
Sony is famous for over complicating equipment, their Trinitron TV circuits were weird, and that is putting it mildly.
Anyway, it is a piece of history now, you cannot buy CD players as new production at this point, and many computer CD-ROM can give raw digital output, which can then be converted with many programs, choose the one that sounds best to your ears.
As for the techniques used, my P-III computer was new when CD players started going out of fashion, newer computers and processors are much more capable.
Even my dish TV box has dedicated processors, and a flash ROM, which I suppose could be hacked to make a DAC, as per your preference of method of processing harmonics.
Signal processing has changed from hard wired to software now, though the basic filter processing is similar.
Anyway, it is a piece of history now, you cannot buy CD players as new production at this point, and many computer CD-ROM can give raw digital output, which can then be converted with many programs, choose the one that sounds best to your ears.
As for the techniques used, my P-III computer was new when CD players started going out of fashion, newer computers and processors are much more capable.
Even my dish TV box has dedicated processors, and a flash ROM, which I suppose could be hacked to make a DAC, as per your preference of method of processing harmonics.
Signal processing has changed from hard wired to software now, though the basic filter processing is similar.
Preliminary SAA7350 datasheet from November 1991:
https://pdf1.alldatasheet.com/datasheet-pdf/view/1161463/NXP/SAA7350.html
Third-order single-bit noise shaper with a DC offset to push the main idle tone with 0 input out of the audio band, some degree of dithering, built-in switched-capacitor DAC and outputs to drive another DAC (TDA1547).
This is also an early order > 2 sigma-delta DAC. An advantage of a switched-capacitor DAC is less sensitivity to the phase noise floor (period-to-period jitter) and a disadvantage that you have to handle nasty current spikes linearly. They use an internal op-amp with an external capacitor across its inputs for that.
https://pdf1.alldatasheet.com/datasheet-pdf/view/1161463/NXP/SAA7350.html
Third-order single-bit noise shaper with a DC offset to push the main idle tone with 0 input out of the audio band, some degree of dithering, built-in switched-capacitor DAC and outputs to drive another DAC (TDA1547).
This is also an early order > 2 sigma-delta DAC. An advantage of a switched-capacitor DAC is less sensitivity to the phase noise floor (period-to-period jitter) and a disadvantage that you have to handle nasty current spikes linearly. They use an internal op-amp with an external capacitor across its inputs for that.
No, it couldn't be the TDA1307 as that only has bitstream output. SAA7350 needs PCM input. TDA1307 takes over the function of both the digital filter and the modulator.
TDA1547 datasheet:
https://pdf1.alldatasheet.com/datasheet-pdf/view/19217/PHILIPS/TDA1547.html
Single-bit switched-capacitor DAC made in a process with bipolar transistors and MOSFETs. Better performance than the DAC inside the SAA7350 due to less crosstalk from the digital part and the more suitable process.
https://pdf1.alldatasheet.com/datasheet-pdf/view/19217/PHILIPS/TDA1547.html
Single-bit switched-capacitor DAC made in a process with bipolar transistors and MOSFETs. Better performance than the DAC inside the SAA7350 due to less crosstalk from the digital part and the more suitable process.
Hypothetically you mean, or have you actually listened to a really good modern dac (and not a $200 one that measures well)?
Your remark is too generic. It's neither here not there.
=========
Shifting gears ...
About TDA1307 and TDA1547. Both of these are very difficult to find in eBay or Ali market place. [Even the Saa7350 isn't cheap]
And if you do - - especially the TDA1547 -- expect to PAY for it. Not sure why? Maybe the were not manuf. in as large qtys as TDA1541. The combo was used in some of the high-end gear from Philips and Marantz.
TDA1307 and TDA1547 used in 2000's high-end Marantz SA-1 SACD player [$7500.00 USD, in 2000]:
https://www.stereophile.com/hirezplayers/271/index.html
Last edited:
Seemed like your comments were that. Or maybe the comments were meant to be more sarcastic as versus accurate?
First ... are you being serious (seriously)?
Second ... (if you are serious), then the "Incredible high quality" is out of the box or after modding?
Or are you only referring to "high quality" measurements?
I have one of high-rated SMSL headphone amps. It measures "perfect" but sonics don't impress compared to a worse-measuring amp. Like a 2-decade old Meier Audio.
Do you mean that older digital stuff is better? That is common thinking among the grey audio brigade. When digital was introduced it were the record players then that were superior. Now those have disintegrated it is old digital stuff that takes their place. It is never the "now" that is OK as the "now" is available to most.
Both SAA7350 and TDA1547 were mediocre DAC chips already in their time. As a listener one patiently waited for better, the DIYer installed better DAC boards and better PSUs/clocks. I have modded many Teac CD players with new Burr Brown based DAC boards. BB also made mistakes, anyone will recall PCM1710.
Sanskrit 10th mkII is being modded indeed but it is an example of those devices that give a lot while costing less than a second hand expensive DAC with a "name". One can condemn newer technology without being in modern technology but it would be OK to just listen and then judge. Time goes on, today we get more than we ever expected. 20 years from now the then grey brigade will search for NIB/NOS AK4493 "from before the fire as those have the best highs"
Both SAA7350 and TDA1547 were mediocre DAC chips already in their time. As a listener one patiently waited for better, the DIYer installed better DAC boards and better PSUs/clocks. I have modded many Teac CD players with new Burr Brown based DAC boards. BB also made mistakes, anyone will recall PCM1710.
Sanskrit 10th mkII is being modded indeed but it is an example of those devices that give a lot while costing less than a second hand expensive DAC with a "name". One can condemn newer technology without being in modern technology but it would be OK to just listen and then judge. Time goes on, today we get more than we ever expected. 20 years from now the then grey brigade will search for NIB/NOS AK4493 "from before the fire as those have the best highs"
Last edited: