Konnichiwa,
I refer to a PSU as "low leakage" is there is not much current flow between earth and the secondary. Hifi/audio gear is usually earthed somewhere, somehow, so such current always flows. The presence of these current loops in conjunction with single ended RCA connections is also the main reason for audible differences between mains cables.
You can achieve low leakage in many ways. With toroidal transformers it often works to use a pair of them, both mounted identical to the chassis with non-conductive screws and wiring one primary out of polarity with the other, secondaries wired normally. The leakage is usually identical but now out of phase, around 20db reduction in leakage current are possible.
The jitter matters most at the DAC, however it is the DAC's WORDCLOCK that determines the jitter output for multibit DAC's, while for Delta/Sigma DAC's the main clock tends to be the system clock. How all of this works out depends hence upon the DAC and other chips used.
Ideally we re-clock the Wordclock of the Multibit DAC, hence we ground our clock signal (hopefully the clock has a seperate supply so that we do not need to worry about PSU line currents) at the re-clocker. If we use the same supply for reclocker and clock again we ground at the reclocker. For a Delta-Sigma DAC we clock directly the DAC and pull our joint supply from the DAC too and we do not connect ANY grounds at the other clock fanout receivers, only the signal.
Retro-fitting such clock distribution systems to existing gear is often messy, as the grounding is not designed for it.
BTW, if you start to implement multiple re-clocking also track down all other clocked devices (Microcontrollers for drive functions, displays etc.) and slave them to an integer multiple of the main clock. Genertaing the main closk from a very high multiple (eg. 33....MHz instead of 16.93MHz) allows dividing down to the commonly found 12 & 8MHz clocks on such PIC's easily.
Once ALL clocks are fully synced (as I did on my "giant killer" CD-67) and all supplis maximally insulated against crosstalk you start hearing what CD is actually capable of in terms of resolution, on both good and bad recordings.... ;-)
Of course, given that I rarely use CD for "serious" music listening I could not care less, really and prefer hence pleasant sounding, visually stunning Digital gear over boring looking but excellent sounding black boxes.
Sayonara
Bricolo said:
I refer to a PSU as "low leakage" is there is not much current flow between earth and the secondary. Hifi/audio gear is usually earthed somewhere, somehow, so such current always flows. The presence of these current loops in conjunction with single ended RCA connections is also the main reason for audible differences between mains cables.
You can achieve low leakage in many ways. With toroidal transformers it often works to use a pair of them, both mounted identical to the chassis with non-conductive screws and wiring one primary out of polarity with the other, secondaries wired normally. The leakage is usually identical but now out of phase, around 20db reduction in leakage current are possible.
Bricolo said:And now, how would the grounding have to be with a complex clock distribution? (like a master clock feeding a decoder, a filter and a dac)
The jitter matters most at the DAC, however it is the DAC's WORDCLOCK that determines the jitter output for multibit DAC's, while for Delta/Sigma DAC's the main clock tends to be the system clock. How all of this works out depends hence upon the DAC and other chips used.
Ideally we re-clock the Wordclock of the Multibit DAC, hence we ground our clock signal (hopefully the clock has a seperate supply so that we do not need to worry about PSU line currents) at the re-clocker. If we use the same supply for reclocker and clock again we ground at the reclocker. For a Delta-Sigma DAC we clock directly the DAC and pull our joint supply from the DAC too and we do not connect ANY grounds at the other clock fanout receivers, only the signal.
Retro-fitting such clock distribution systems to existing gear is often messy, as the grounding is not designed for it.
BTW, if you start to implement multiple re-clocking also track down all other clocked devices (Microcontrollers for drive functions, displays etc.) and slave them to an integer multiple of the main clock. Genertaing the main closk from a very high multiple (eg. 33....MHz instead of 16.93MHz) allows dividing down to the commonly found 12 & 8MHz clocks on such PIC's easily.
Once ALL clocks are fully synced (as I did on my "giant killer" CD-67) and all supplis maximally insulated against crosstalk you start hearing what CD is actually capable of in terms of resolution, on both good and bad recordings.... ;-)
Of course, given that I rarely use CD for "serious" music listening I could not care less, really and prefer hence pleasant sounding, visually stunning Digital gear over boring looking but excellent sounding black boxes.
Sayonara
Dave,
I would suggest you consider using some film/foil capacitors in place of the output electrolytics. With speakers flat to 21Hz, 3.3uF was the lowest value to which I could go without noticeable impact on low bass, though the input impedance of the following preamp might influence your results.
Can you be more specific with regard to what you term as ‘mediocre’, it is not much to go on.
Tim.
I would suggest you consider using some film/foil capacitors in place of the output electrolytics. With speakers flat to 21Hz, 3.3uF was the lowest value to which I could go without noticeable impact on low bass, though the input impedance of the following preamp might influence your results.
Can you be more specific with regard to what you term as ‘mediocre’, it is not much to go on.
Tim.
Tim,
Firstly, your spellling is better than mine!!
Secondly, expanding on mediocre:-
In my experience cheap CD players sound either soft and lacking in detail or harsh and bright. I can make the CD723 sound either way but it still lacks any real resolution and depth. The bass is boomy and thin rather than full and tight, instruments lack tonal colouring, treble is thin - "common digitalis".
Maybe I'm being hard on the poor little CD723 because I thought a Naim CD5i I demo'ed had the same problems (erring on the soft side) and nearly all (new) sub £1500 players do. That was why I asked the original question - is this machine worth further effort?
Sreten thinks not. KYW etc thinks I can do better than my 1st round of mods. Time will tell (although it might be quite a while).
Firstly, your spellling is better than mine!!
Secondly, expanding on mediocre:-
In my experience cheap CD players sound either soft and lacking in detail or harsh and bright. I can make the CD723 sound either way but it still lacks any real resolution and depth. The bass is boomy and thin rather than full and tight, instruments lack tonal colouring, treble is thin - "common digitalis".
Maybe I'm being hard on the poor little CD723 because I thought a Naim CD5i I demo'ed had the same problems (erring on the soft side) and nearly all (new) sub £1500 players do. That was why I asked the original question - is this machine worth further effort?
Sreten thinks not. KYW etc thinks I can do better than my 1st round of mods. Time will tell (although it might be quite a while).
Konnichiwa,
I am not Dave, but to quote:
"I have a CD723 with all the "Chris Found" mods plus:
KC-7,
AD826 I/V"
KC-7 here stands for Kwak Clock 7.
This is exactly the kind of hype that get's me annoyed. I know quite a bit of Ditital gear that relies on bog standard clocks (single gate on a big LSI Chip) and merely feeds them with a clean supply and guess what, it sounds just fine AND measures in < 200pS region for Jitter.
I wish people would stop this "Add clock - great", "No new Clock = Bad sound" nonsense. This new clock business reminds me by far too much of the Emporers new Cloth & Groove.
Yes, in virtually ALL cases changing the clock will change the sound. To the better? Sometimes. Often add-in clocks actually worsen the jitter levels (not reducing them as claimed) and plenty of digital gear sounds just fine without them, plenty more can be modded to sound great with a "Clock". True, fitting correctly a proven low jitter clock module CORRECTLY can reduce jitter and improve the sound, but such is rather rare.
Sayonara
carlosfm said:
I am not Dave, but to quote:
"I have a CD723 with all the "Chris Found" mods plus:
KC-7,
AD826 I/V"
KC-7 here stands for Kwak Clock 7.
carlosfm said:
This is exactly the kind of hype that get's me annoyed. I know quite a bit of Ditital gear that relies on bog standard clocks (single gate on a big LSI Chip) and merely feeds them with a clean supply and guess what, it sounds just fine AND measures in < 200pS region for Jitter.
I wish people would stop this "Add clock - great", "No new Clock = Bad sound" nonsense. This new clock business reminds me by far too much of the Emporers new Cloth & Groove.
Yes, in virtually ALL cases changing the clock will change the sound. To the better? Sometimes. Often add-in clocks actually worsen the jitter levels (not reducing them as claimed) and plenty of digital gear sounds just fine without them, plenty more can be modded to sound great with a "Clock". True, fitting correctly a proven low jitter clock module CORRECTLY can reduce jitter and improve the sound, but such is rather rare.
Sayonara
Now we're getting somewhere!
In order to start on the new mods I have reversed some of the old mods. I removed the emitter follower type supplies and lashed in a 7815/7915 PSU (liberated from some pro-audio piece of kit only 7 years old but already obsolete!!) just to power the I/V, all the other supplies are back to the original. The change in sound was minimal, possibly indicating:
a) I'm deaf
b) I did something wrong with the EF supplies
c) Power supplies don't matter
d) 7815/7915 ain't that bad.
e) Bigger problems elsewhere.
Oh well here we go, at the risk of starting an argument.....
Next I removed the KC-7 and refitted the original clock components (even those horrible axial 22pF caps), the same crystal was used (the original). The sound improved significantly with the standard clock arrangement, although it is still has some of the red book upper mid glare, the resolution, dynamics and general musicallity is much better.
The KC-7 was built on Elso's pcb and fitted under the ECO-SL board with a short clock connection to the sig processor and the 3 supply wires were connected at the main smoothing caps on the DAC board. The only deviation from Elso's recommendations was that the 1uF caps were naked stacked film polyesters not polypropylene - in order to keep the unit low profile.
The next step is to try KYW's current f/b I/V stage (AD812 in lieu of LM6182) but I need a bit more help here. KYW suggests selecting the DAC reference resistors for 2mA full scale current. The data sheet shows 1mA and R(ref) is 11K, with 22K from VDD to Vref and 33K from vref to Pin 7. My CD723 has the corresponding values - 11K (internal to the DAC), 15K and 10K.
I've read the data sheet about 10 times and the explanation is not at all clear to me, so how do I calculate the resistor values neccessary for 2mA FS current?
Thanks,
Dave
In order to start on the new mods I have reversed some of the old mods. I removed the emitter follower type supplies and lashed in a 7815/7915 PSU (liberated from some pro-audio piece of kit only 7 years old but already obsolete!!) just to power the I/V, all the other supplies are back to the original. The change in sound was minimal, possibly indicating:
a) I'm deaf
b) I did something wrong with the EF supplies
c) Power supplies don't matter
d) 7815/7915 ain't that bad.
e) Bigger problems elsewhere.
Oh well here we go, at the risk of starting an argument.....
Next I removed the KC-7 and refitted the original clock components (even those horrible axial 22pF caps), the same crystal was used (the original). The sound improved significantly with the standard clock arrangement, although it is still has some of the red book upper mid glare, the resolution, dynamics and general musicallity is much better.
The KC-7 was built on Elso's pcb and fitted under the ECO-SL board with a short clock connection to the sig processor and the 3 supply wires were connected at the main smoothing caps on the DAC board. The only deviation from Elso's recommendations was that the 1uF caps were naked stacked film polyesters not polypropylene - in order to keep the unit low profile.
The next step is to try KYW's current f/b I/V stage (AD812 in lieu of LM6182) but I need a bit more help here. KYW suggests selecting the DAC reference resistors for 2mA full scale current. The data sheet shows 1mA and R(ref) is 11K, with 22K from VDD to Vref and 33K from vref to Pin 7. My CD723 has the corresponding values - 11K (internal to the DAC), 15K and 10K.
I've read the data sheet about 10 times and the explanation is not at all clear to me, so how do I calculate the resistor values neccessary for 2mA FS current?
Thanks,
Dave
Dave,
With your initial (I notice you have just made some progress so I hope this is not too late) CD723 criticisms in mind, here are some thoughts based on my own experience with this player and in comparison with a slightly modified (for a dc-coupled output) Marantz CD17KI.
Output capacitors...
10uF Oscons (bypassed with small polyprops) were reasonably good although there was a degree of congestion and harshness coupled with a ‘small’ sound. My own preference here is for paper-in-oil types but from what you have written and again from experience an Audyn tin-foil ’KP-SN’ could suit your requirements nicely, being not so glassy in the treble as a Hovland Musicap for example and sounding more spacious than Oscons and other electrolytics.
Op-amps…
I wonder if you might like the NJM2114, significantly more dynamic than the NE5534 and in many ways more refined, though to my ears a little harsh in the treble. My preference is for the NE5534 (of recent Philips manufacture) though I got fine results from a pair of LM6171s (with 500nF across the +/- pins). I have tried 1980s Signetics NE5534s in other circuits with disappointing results. The OPA627 was impressive but I couldn’t live with it. I find a 2mA constant current source on the op-amp output usually leads to a noticeable improvement in refinement, though sometimes a very slightly diminished amount of bass ‘drive’ accompanies this – not a problem for me, but for you…?
Regulation…
I echo those who suggest regulating the digital supplies. There is a rather light-weight, flimsy, diluted quality to the sound without regulation. This was particularly noticeable through an external DAC when compared with the digital output of CD17KI, though it was not unpleasant. I regulated the existing supply to around 10V (if I remember correctly) for the servo circuits with a feed off this to a shunt regulator for the 5V supplies. Further listening between the two digital outputs proved a waste of time, any remaining differences between the two circuits were now so small as to be insignificant.
Other changes…
Removed output muting transistors, some A.N. tantalum resistors here and some Elna Silmics there.
The end result…
The CD723 became (after C.F.’s recommended changes and after adding the regulation) subjectively indistinguishable from the CD17KI when their digital outputs were fed to an external DAC. The analogue sound of the CD17KI remained the more analytical while the modified CD723 (NE5534/paper-in-oil) became the more musically involving. Fitting an NJM2114 into the CD723 output narrows the gap between the two considerably, making the CD723 more ‘up-beat’, many people would prefer this.
Tim.
With your initial (I notice you have just made some progress so I hope this is not too late) CD723 criticisms in mind, here are some thoughts based on my own experience with this player and in comparison with a slightly modified (for a dc-coupled output) Marantz CD17KI.
Output capacitors...
10uF Oscons (bypassed with small polyprops) were reasonably good although there was a degree of congestion and harshness coupled with a ‘small’ sound. My own preference here is for paper-in-oil types but from what you have written and again from experience an Audyn tin-foil ’KP-SN’ could suit your requirements nicely, being not so glassy in the treble as a Hovland Musicap for example and sounding more spacious than Oscons and other electrolytics.
Op-amps…
I wonder if you might like the NJM2114, significantly more dynamic than the NE5534 and in many ways more refined, though to my ears a little harsh in the treble. My preference is for the NE5534 (of recent Philips manufacture) though I got fine results from a pair of LM6171s (with 500nF across the +/- pins). I have tried 1980s Signetics NE5534s in other circuits with disappointing results. The OPA627 was impressive but I couldn’t live with it. I find a 2mA constant current source on the op-amp output usually leads to a noticeable improvement in refinement, though sometimes a very slightly diminished amount of bass ‘drive’ accompanies this – not a problem for me, but for you…?
Regulation…
I echo those who suggest regulating the digital supplies. There is a rather light-weight, flimsy, diluted quality to the sound without regulation. This was particularly noticeable through an external DAC when compared with the digital output of CD17KI, though it was not unpleasant. I regulated the existing supply to around 10V (if I remember correctly) for the servo circuits with a feed off this to a shunt regulator for the 5V supplies. Further listening between the two digital outputs proved a waste of time, any remaining differences between the two circuits were now so small as to be insignificant.
Other changes…
Removed output muting transistors, some A.N. tantalum resistors here and some Elna Silmics there.
The end result…
The CD723 became (after C.F.’s recommended changes and after adding the regulation) subjectively indistinguishable from the CD17KI when their digital outputs were fed to an external DAC. The analogue sound of the CD17KI remained the more analytical while the modified CD723 (NE5534/paper-in-oil) became the more musically involving. Fitting an NJM2114 into the CD723 output narrows the gap between the two considerably, making the CD723 more ‘up-beat’, many people would prefer this.
Tim.
Kuei Yang Wang said:
Sorry Kuei, I forgot he was already using an external clock.
Anyway, I think you can always do better with a good clock, that's my experience.
On my CD650 I double-regulated the 7220 and, still with the original clock, bass was not tight at all, mucked the whole music.
Only when I put a Tent clock (with dedicated trafo/PSU) things got much better.
The same happened to several other CDPs I own.
But you may have your tricks, don't discuss that.
And each circuit may be a different thing and a different approach.
Re: Now we're getting somewhere!
Konnichiwa,
I think it's a mix. First, your current Op-Amp's are fairly tolerate to PSU problems. Second the really nasty drive and digital section supply really causes troubles.
Okay. Now, if you take my recommendation get a simple clock (like the cheap Tent Module) and give it it's own supply, low leakage transformer for the supply and a really nicely low noise supply, then the correct logic level to the Servo Chip.
And for goodness sake give the actual Player logic it's own well regulated and clean 5V supply plus a nice solid regulated (7812/7912 is fine) supply for the servos.
Using this hookup will maximise all possible problems with ground noise etc to maximise jitter. A typhical example why "Clock" is nowhere near the "cure all digital ills" many claim it to be.
The Datasheet (it is on page 9 Note 5) suggests two 11K resistors. Why you are at it, the capacitor on the Reference section is also quite critical for sonics. I'd use 47uF/6.3V BG NX-HiQ (I do occasionally use BG's, where they actually work). With the above the feedback resistors for the OPA (as discussed previously) become ideally 2K8 or 2K87.
Sayonara
Konnichiwa,
Dave S said:
I think it's a mix. First, your current Op-Amp's are fairly tolerate to PSU problems. Second the really nasty drive and digital section supply really causes troubles.
Dave S said:
Okay. Now, if you take my recommendation get a simple clock (like the cheap Tent Module) and give it it's own supply, low leakage transformer for the supply and a really nicely low noise supply, then the correct logic level to the Servo Chip.
And for goodness sake give the actual Player logic it's own well regulated and clean 5V supply plus a nice solid regulated (7812/7912 is fine) supply for the servos.
Dave S said:
Using this hookup will maximise all possible problems with ground noise etc to maximise jitter. A typhical example why "Clock" is nowhere near the "cure all digital ills" many claim it to be.
Dave S said:
The Datasheet (it is on page 9 Note 5) suggests two 11K resistors. Why you are at it, the capacitor on the Reference section is also quite critical for sonics. I'd use 47uF/6.3V BG NX-HiQ (I do occasionally use BG's, where they actually work). With the above the feedback resistors for the OPA (as discussed previously) become ideally 2K8 or 2K87.
Sayonara
Thorsten,
How can I easy and quick find the polarity of windings of transformer?
Could you describe more precisely what is 2-Chamber safety construction with a dairly thick plastic in the thing and in general small size? Sorry but it might be language barrier that I don't quite understand what you mean.
Do the two transformers have to be mounted in special way to each other to achive better canceletion?
What about leakage if we apply balanced separate transformer on the mains.I mean the one with dual 115VAC windings for 230V and ground taken from center tap?
Bartek
Konnichiwa,
Dual trace 'scope?
No, that is about as precise as it gets. Many transformers wind secondary and primary over each other. "2-Chamber Safety" Type transformers have seperate chambers for primary and secondary and often fairly thick plastic (and air) between the windings and the core.
Symmetrical to even the capacitances.
If our equipment mains transformer is not balanced too we will have some improvement in leakage, but not much.
Sayonara
zygibajt said:
Dual trace 'scope?
zygibajt said:
No, that is about as precise as it gets. Many transformers wind secondary and primary over each other. "2-Chamber Safety" Type transformers have seperate chambers for primary and secondary and often fairly thick plastic (and air) between the windings and the core.
zygibajt said:
Symmetrical to even the capacitances.
zygibajt said:
If our equipment mains transformer is not balanced too we will have some improvement in leakage, but not much.
Sayonara
Is this what you are talking about?
Hi Tim and Kuei
This is something I am really interested in doing in my CD624. At present, the unregulated +/-10V supply for the servos and the regulated +/-5V digital supply share the same output from the transformer, rectifier and smoothing capacitors (6800uF for +, and 1000uF for -). There is a separate rectifier, smoothing caps (330uf and 470uf) for the +/-15V supplies for the analogue stage. This will be redundant as a new transformer and supplies will be added for the opamp and DAC.
What is the best way to improve and regulate the +/-10V and +/-5V supplies from the original transformer?
Should the entire +/-10 servo section be supplied from the old +/-15 supply - leaving the other rectifier/supply for the +/-5V digital section?
If this is the way to go, what changes need to be made to the capacitors (6800uf, 1000uf, 470uf, 330uf)?
TimA said:
Kuei Yang Wang said:
Hi Tim and Kuei
This is something I am really interested in doing in my CD624. At present, the unregulated +/-10V supply for the servos and the regulated +/-5V digital supply share the same output from the transformer, rectifier and smoothing capacitors (6800uF for +, and 1000uF for -). There is a separate rectifier, smoothing caps (330uf and 470uf) for the +/-15V supplies for the analogue stage. This will be redundant as a new transformer and supplies will be added for the opamp and DAC.
What is the best way to improve and regulate the +/-10V and +/-5V supplies from the original transformer?
Should the entire +/-10 servo section be supplied from the old +/-15 supply - leaving the other rectifier/supply for the +/-5V digital section?
If this is the way to go, what changes need to be made to the capacitors (6800uf, 1000uf, 470uf, 330uf)?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.