I am working around the output stage of a Sony CDP player XB940 SACD, the XA50/30/20 and XB940/920/720 are similar (VC24+ CXD2562Q or CXD8735N - DAC chip’s). If someone is interested I can go further in this thread. I tested various configurations (OP-AMP’s / discrete) the last one uses a trans former to unbalance the output, but it can be used in balance mode. The sound is really very "natural" and the design is quite simple.
At once I have some problems with my ftp pages and with my provider, so I can’t post the schematics now.
For the moment I can tell you the evolution of the improvements :
The first and easiest way to improve this player is to bypass the “current pulse chips”, so you get the DSD directly from the dac in a differential voltage mode, both at 2.5V dc, I think you have already seen this in a few sites (VSE, LC-Audio etc...) For the output stage, there are various possibilities : Op-Amp, discrete differential to single ended, transformer…
The goal was also to unbalance the dac outputs, to make it compatible with my preamp/crossover or at least to get the choice between balanced or unbalanced outputs.
In order to “try” the signal directly at the output of the dac, and to achieve a smooth filtering, the first simple way, was to use a “good” op amp. The AD826 or AD828 are among the best ones for this application in my knowledge and the less tricky to implement. I prefer the AD828 (Gv>2) but the AD826 with a current sink of 5mA to V-. will give you about the same results. I didn’t tested the most recent ones from Analog Devices…
This simple mod, bypassing the current-chips and a few Burr Brown op amps improved greatly the sound in all the ways !
Well I hope tonight or tomorrow my provider will solve the problems !
Keep in touch.
Questions or suggestions are welcome… while waiting debugging !
For the moment I can tell you the evolution of the improvements :
The first and easiest way to improve this player is to bypass the “current pulse chips”, so you get the DSD directly from the dac in a differential voltage mode, both at 2.5V dc, I think you have already seen this in a few sites (VSE, LC-Audio etc...) For the output stage, there are various possibilities : Op-Amp, discrete differential to single ended, transformer…
The goal was also to unbalance the dac outputs, to make it compatible with my preamp/crossover or at least to get the choice between balanced or unbalanced outputs.
In order to “try” the signal directly at the output of the dac, and to achieve a smooth filtering, the first simple way, was to use a “good” op amp. The AD826 or AD828 are among the best ones for this application in my knowledge and the less tricky to implement. I prefer the AD828 (Gv>2) but the AD826 with a current sink of 5mA to V-. will give you about the same results. I didn’t tested the most recent ones from Analog Devices…
This simple mod, bypassing the current-chips and a few Burr Brown op amps improved greatly the sound in all the ways !
Well I hope tonight or tomorrow my provider will solve the problems !
Keep in touch.
Questions or suggestions are welcome… while waiting debugging !
HOOOUUUUUFFF !!!
Well after working hard, (with a big help from my wife !, she is my blind listener too...), I find another way to post my diagrams.
Below the first ones.
They may appear not really original, but the improvement taking into account simplicity is worth.
XB-output1
The DAC output offset between +out and –out, is not 0V but a few mV, and moreover, change according to the playing mode CD (pcm) or SACD (bitstream).
DC-servo add here.
XB-output2
Well after working hard, (with a big help from my wife !, she is my blind listener too...), I find another way to post my diagrams.
Below the first ones.
They may appear not really original, but the improvement taking into account simplicity is worth.
XB-output1
The DAC output offset between +out and –out, is not 0V but a few mV, and moreover, change according to the playing mode CD (pcm) or SACD (bitstream).
DC-servo add here.
XB-output2
DAC Out
Richard,
If I am correct, VSE is not using the signal from the DAC output, but rather uses the differential bit-stream output directly from the VC24+ before the DAC and then low-pass filter / buffer.
Have you tried this, or are there reasons why you prefer to tap after the DAC ?
Also would you have datasheets for the VC24+ / DAC chips ?
Or some pin assignments would already be a GREAT help.
Many thanks in advance,
Patrick
Richard,
If I am correct, VSE is not using the signal from the DAC output, but rather uses the differential bit-stream output directly from the VC24+ before the DAC and then low-pass filter / buffer.
Have you tried this, or are there reasons why you prefer to tap after the DAC ?
Also would you have datasheets for the VC24+ / DAC chips ?
Or some pin assignments would already be a GREAT help.
Many thanks in advance,
Patrick
Re: DAC Out
No, I do the same way as VSE or LC audio. If you refer to the XB940, the VC24+ is under the PCB, and this is the “real” DAC. What you see on both sides in skew and lengthened, with a “copper tape” are the current pulse converters. These chip’s are very sensitive to their supply (+7,-7V) and their reference. I tested in this way, after the current pulse converters, with battery power supplies and batt for the reference, (battery for the reference was the biggest improvement), this modification is not as effective in the reproduce sound quality as the one I describe here, and is more complicated and difficult to implement. Well… I must say that I didn’t test it with a jocko homo I/V converter... By the way, if someone here remember the thread where he speaks about it, I am curious to see his design, well maybe just a simplified one, I saw so much people in this forum speaking about it !
I get from Sony the service manuals of the SCD XB 940, CDP XA-50 and CDP XB-920.
I don’t have a scanner or a digital camera for the moment, maybe next week…but you can take a look at the VSE or LC-audio (Zap filter) sites, for their photos. I hope that Allen Wright or Lars Clausen will not be annoyed.
Davet, EUVL,
I can try to draw a simpified diagram for the output assignements pins of the VC24+, or the input pins of the current converters, I think maybe it’s easier to get the signal at the input of those chips.
(Maybe someone in this forum has them in computer file ?)
EUVL said:
No, I do the same way as VSE or LC audio. If you refer to the XB940, the VC24+ is under the PCB, and this is the “real” DAC. What you see on both sides in skew and lengthened, with a “copper tape” are the current pulse converters. These chip’s are very sensitive to their supply (+7,-7V) and their reference. I tested in this way, after the current pulse converters, with battery power supplies and batt for the reference, (battery for the reference was the biggest improvement), this modification is not as effective in the reproduce sound quality as the one I describe here, and is more complicated and difficult to implement. Well… I must say that I didn’t test it with a jocko homo I/V converter... By the way, if someone here remember the thread where he speaks about it, I am curious to see his design, well maybe just a simplified one, I saw so much people in this forum speaking about it !
I get from Sony the service manuals of the SCD XB 940, CDP XA-50 and CDP XB-920.
I don’t have a scanner or a digital camera for the moment, maybe next week…but you can take a look at the VSE or LC-audio (Zap filter) sites, for their photos. I hope that Allen Wright or Lars Clausen will not be annoyed.
Davet, EUVL,
I can try to draw a simpified diagram for the output assignements pins of the VC24+, or the input pins of the current converters, I think maybe it’s easier to get the signal at the input of those chips.
(Maybe someone in this forum has them in computer file ?)
Very useful information
Richard,
Your information is and will be most useful for someone like me who is thinking of doing the same modification (on an XB940). So any picture or sketches of your work is very welcome indeed. If you would be kind enough to take the trouble, please send me a copy of the manual and datasheet by post. I'll reimburse your costs. Send me a email so that I can let you have my postal address.
Many thanks for sharing with us.
Patrick
Richard,
Your information is and will be most useful for someone like me who is thinking of doing the same modification (on an XB940). So any picture or sketches of your work is very welcome indeed. If you would be kind enough to take the trouble, please send me a copy of the manual and datasheet by post. I'll reimburse your costs. Send me a email so that I can let you have my postal address.
Many thanks for sharing with us.
Patrick
Patrick,
Sorry but I can’t join you, by your forum mail.
Peter,
Did you detected a real improvement by changing only the clock ?
I tried too, a discrete op amp, but the sounding results where not as good as the simples fet’s designs below, I am working on…
The firsts fet’s designs were simply intended to “see” how a really basic design would sound, I didn’t expect such good results, they exceeded my hopes.
The sound was astonishing real, natural, when you reach this degree of transparency. Sometimes in some cd’s you wonder whether certain noises come from ! Your room ? Are they recorded ? Until you listen the track once more…
Following these very good results I refined the design to make it reliable in the player. It is the design with dc-servo and transformer.
Evaluation schematic.
Fet with dc-servo.
Final schematic.
Sorry but I can’t join you, by your forum mail.
Peter,
Did you detected a real improvement by changing only the clock ?
I tried too, a discrete op amp, but the sounding results where not as good as the simples fet’s designs below, I am working on…
The firsts fet’s designs were simply intended to “see” how a really basic design would sound, I didn’t expect such good results, they exceeded my hopes.
The sound was astonishing real, natural, when you reach this degree of transparency. Sometimes in some cd’s you wonder whether certain noises come from ! Your room ? Are they recorded ? Until you listen the track once more…
Following these very good results I refined the design to make it reliable in the player. It is the design with dc-servo and transformer.
Evaluation schematic.
Fet with dc-servo.
Final schematic.
I did the clock mod first and it made an "incredible" difference. Everything about the sound was improved.... a real jawdropper.
Actually me and a friend felt that both the 940 and a Marantz CD63KI-SE got the most improvement from the clock mod. Sure the Zapfilter made a lot as well, but the clock was the bigger difference overall.
/Peter
Actually me and a friend felt that both the 940 and a Marantz CD63KI-SE got the most improvement from the clock mod. Sure the Zapfilter made a lot as well, but the clock was the bigger difference overall.
/Peter
Re: Great Tip
Hi Patrick,
I presume the Zapfilter can also be connected to the DAC chip as Allen Wright is doing. Only the current pulse chip is bypassed, NOT the DAC. I don't have experience with the Zapfilter only with my own clock and the Vacuumstate modification in the Sony SCD-1. My friend the owner of the SCD-1 prefers the sound of the original implementation with opamps. To me it seems that the clock is even more critical than for CD.
Many things could be done to improve the Sony: powersupply, bypassing, other opamps etc. etc. The many SMD parts discourage me from doing it.........
EUVL said:
Hi Patrick,
I presume the Zapfilter can also be connected to the DAC chip as Allen Wright is doing. Only the current pulse chip is bypassed, NOT the DAC. I don't have experience with the Zapfilter only with my own clock and the Vacuumstate modification in the Sony SCD-1. My friend the owner of the SCD-1 prefers the sound of the original implementation with opamps. To me it seems that the clock is even more critical than for CD.
Many things could be done to improve the Sony: powersupply, bypassing, other opamps etc. etc. The many SMD parts discourage me from doing it.........
Hi jean paul,
The version I prefer, is the fet one with servo and transformer, the op amp version is not very far in terms of "very very good reproduction", but the last fet version with transformer is simply real, well... much more real.
Actually I use the transformer, to unbalance the output (my crossover input is unbalanced). I tested the schematic below, with just one "leg" the +output, and I increase R101/110 to about 390 Ohms to get 0V dc in the output. It's just a very little better than the transfo version, but unfortunately a little bit noisy, well i can hear it with my ears near the horn of my TAD's (of course, noise is cancelled by the balanced mode).
Noise in my 20 kHz FFT is at -114 dB in CD mode and balanced, about -130 dB in SACD mode.
Sorry, in my last post, the second link doesn't show the good schematic.
XB_out_fet_dc
Fet's are matched by pairs and about 8 mA of Idss.
Transfos are LUNDHAL LL5402, because I have 2 of them from an audio distribution amplifier, and their caracteristics seems to match my requirements. I will test next days an LL7902.
I am not very qualified in transformers, so some advises are welcome.
The output impedance of the fet followers is about 30 ohms, but it's better not to load them at less than 3 kOhms.
The version I prefer, is the fet one with servo and transformer, the op amp version is not very far in terms of "very very good reproduction", but the last fet version with transformer is simply real, well... much more real.
Actually I use the transformer, to unbalance the output (my crossover input is unbalanced). I tested the schematic below, with just one "leg" the +output, and I increase R101/110 to about 390 Ohms to get 0V dc in the output. It's just a very little better than the transfo version, but unfortunately a little bit noisy, well i can hear it with my ears near the horn of my TAD's (of course, noise is cancelled by the balanced mode).
Noise in my 20 kHz FFT is at -114 dB in CD mode and balanced, about -130 dB in SACD mode.
Sorry, in my last post, the second link doesn't show the good schematic.
XB_out_fet_dc
Fet's are matched by pairs and about 8 mA of Idss.
Transfos are LUNDHAL LL5402, because I have 2 of them from an audio distribution amplifier, and their caracteristics seems to match my requirements. I will test next days an LL7902.
I am not very qualified in transformers, so some advises are welcome.
The output impedance of the fet followers is about 30 ohms, but it's better not to load them at less than 3 kOhms.
Elso,
as i wrote above, I use Zapfilter direct at VC24+. I believe I was the first one that did this mod after Allen, however I did the clock before Allen so I was the first one to hear what the Sony SACDs really can do with both Jitter reduction and discrete analog circuitry
.
The clock mod did as much for CD as for SACD.
If I had to chose between the LClock or the Zapfilter in the 940 I´d go with the clock without doubt. My friend with the CD63 felt the same.
/Peter
as i wrote above, I use Zapfilter direct at VC24+. I believe I was the first one that did this mod after Allen, however I did the clock before Allen so I was the first one to hear what the Sony SACDs really can do with both Jitter reduction and discrete analog circuitry
.The clock mod did as much for CD as for SACD.
If I had to chose between the LClock or the Zapfilter in the 940 I´d go with the clock without doubt. My friend with the CD63 felt the same.
/Peter
Clocks
Has anyone made a comparison between different clocks -- e.g. Tent, LCAudio, and that from Elso.
Elso, I know you are probably fed up by now with people asking you for details of your clock, but I have a hard time finding a comprehensive description in the forum. If it exists, could you kindly post a link ?
Many thanks to all. Very useful information indeed.
Patrick
Has anyone made a comparison between different clocks -- e.g. Tent, LCAudio, and that from Elso.
Elso, I know you are probably fed up by now with people asking you for details of your clock, but I have a hard time finding a comprehensive description in the forum. If it exists, could you kindly post a link ?
Many thanks to all. Very useful information indeed.
Patrick
Re: Clocks
Can't find back the post, lousy searchengine.......
I did, but of course I am completely biased!EUVL said:
Can't find back the post, lousy searchengine.......
Ok, I will enable sending emails to me, temporarily. Just hit the email button below this post and I will send you the schematic. Even the wildest avatar does not seem to scare you guys off.
Sorry ! When I redrew the simulation diagrams to clean and comment them, I made a drawing error on the fet’s dc servo ! (dc servo inputs not crossed).
I will change the drawings in the link, in order to update them.
Below the updated schematics.
XB-output1.
XB-output2.
Fet evaluation schematic.
Fet with dc-servo.
Fet final schematic.
I will change the drawings in the link, in order to update them.
Below the updated schematics.
XB-output1.
XB-output2.
Fet evaluation schematic.
Fet with dc-servo.
Fet final schematic.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.