Mods to Naim CDS3
- Thread starter rontoolsie
- Start date
rontoolsie
pfm Member
Why go for a 10H choke when you can get a 30H one (with a threefold extra reduction in ripple)? The cost of the choke is mainly the time spend destroying a drive unit, picking the ceramic magnet of it, machining the T yolk, screening and potting. The time spent winding it is nominal and there would be no significant difference in price for a 10 or 30H piece.
If you are at all interested, e-mail me at [email protected]
If you are at all interested, e-mail me at [email protected]
Sparts155 said:
rontoolsie
pfm Member
Here are the effects of what a choke regulator does to the ripple. The choke is placed after the rectifiers and before the filter caps. The below tracings are the ripple going into the filter caps, and show a 50-fold reduction by the simple insertion of the choke (510mV vs 10.9mV). The shape of the after-choked ripple is also rather more benign. Although I do not have the traces of the ripple coming out of the voltage regulator boards, it shows a 30-fold reduction (somebody can figure out what that translates to in dB).
Here is the ripple pattern at 120Hz without any choke regulation
And after (note that since it is of much lower magnitude, the vertical scale is different from the picture above.
Note, both heat and ripple cause degradation of filter caps. By greatly reducing the ripple going into them, the projected lifespan before needing a recap should be considerably lengthened.
Here is the ripple pattern at 120Hz without any choke regulation
And after (note that since it is of much lower magnitude, the vertical scale is different from the picture above.
Note, both heat and ripple cause degradation of filter caps. By greatly reducing the ripple going into them, the projected lifespan before needing a recap should be considerably lengthened.
Ron,
Interesting work you're doing there. The second trace looks like what I would expect a capacitor filtered power supply to look like, but the first puzzles me with its little downward spikes. I'd be interested if others know how to interpret this. Do you know what power supply circuit the traces show?
Many have found the MKT 2XL a good alternative to tants, but this involves getting at the back of the PCB. It occurs to me that you did not have to do that for the Cardas caps.
David
Interesting work you're doing there. The second trace looks like what I would expect a capacitor filtered power supply to look like, but the first puzzles me with its little downward spikes. I'd be interested if others know how to interpret this. Do you know what power supply circuit the traces show?
Many have found the MKT 2XL a good alternative to tants, but this involves getting at the back of the PCB. It occurs to me that you did not have to do that for the Cardas caps.
David
rontoolsie
pfm Member
The upper trace is the output from the diode rectifiers of a 552PS BEFORE the filter caps (yes, I also choke regulated that)...it is split rail-giving a a fairly high ripple entering filter caps. The ripple after the voltage regulator boards is somewhere in the order of 5mV....but with a choke upstream becomes closer to 0.2 mV, so not only is it 50x lower, the spectral content is also more benign. And the choke gives extra low impedance energy storage...with a constant current draw of say 1 A (not unreasonable given the large numbers of LM317 internally, a 40H choke would store 20J of energy and contribute to the dynamics and general solidity of sound. Which is exactly what I hear.
rubaiyat_23
pfm Member
Actually, Ron, your remarks about switching the ground via the relays are incorrect–there are four relays, one of which is used to switch in the make-up gain required by terms of licensing the HDCD circuit.
A cursory look at the ground plane of the board would show this quite clearly.
Did you solder the wires yourself? You may actually be missing some resolution in this configuration–draw out a schematic for yourself, HDCD discs will be -6dB too quiet.
Also–no reason why you couldn't mount the bypass capacitor for the filter caps on lugs (as the other connections to the reservoir capacitors)–fairly easy to do, really, solder component to lugs ahead of time (no heat sinking into capacitor).
Which part of the power supply is the choke used on? As it looks as if you're only using it on the positive-going section of the analog supplies, how does this affect offset at the analog circuits (discrete bi-polar op amp design)?
Lastly–have you looked at the spectral analysis plots for the two ripple traces shown above? Under load (e.g., signal through output stage)?
A cursory look at the ground plane of the board would show this quite clearly.
Did you solder the wires yourself? You may actually be missing some resolution in this configuration–draw out a schematic for yourself, HDCD discs will be -6dB too quiet.
Also–no reason why you couldn't mount the bypass capacitor for the filter caps on lugs (as the other connections to the reservoir capacitors)–fairly easy to do, really, solder component to lugs ahead of time (no heat sinking into capacitor).
Which part of the power supply is the choke used on? As it looks as if you're only using it on the positive-going section of the analog supplies, how does this affect offset at the analog circuits (discrete bi-polar op amp design)?
Lastly–have you looked at the spectral analysis plots for the two ripple traces shown above? Under load (e.g., signal through output stage)?
rontoolsie
pfm Member
I did initially just choke regulate the +24V for a number of reasons...
1. The value of the choke (30H, <0.25 ohm DC resistance, heavy screening) made it impossible for an off-the-shelf component, so one had to be hand built (including coil winding and machining a T-yolk) just to see if this was possible. The initial measurements and more importantly the effect on subjective performance spurred me to move ahead.
2. I like to implement changes one thing at a time.
3. Cost.
I have now individual chokes on each of the +24, -24 (analog) as well as the +15, -15 (digital) for a total of four chokes, each of the same 30H spec. This takes the weight of the XPS2 over that of a 555PS. The result is that the CDS3 sounds turbocharged as compared to a standard XPS2 with a sense of dynamics, purity, timing and tonal accuracy that has got to be heard to be believed- and yes, I did have another unmodded XPS2 unit side by side to compare it with. Several weeks ago I took it over to a friends who is using a CDX/XPS1 into a 52. I also took along my 552. Power amp was the NAP500.
The improvement of the XPS1 to the quad choked XPS2 was even larger in degree to swapping out his 52 for my 552 which was done in the same listening session. The CDX was transformed in quite an unexpected way transforming into a source worthy of the 552.
Given the high current draw of the CDS3 head unit, the chokes also provide a greatly increased energy storage (as the energy stored by an inductor is proportion to the square of the current) that by my calculations are several times higher than the filter caps themselves. In conjunction with a 50-fold reduction of ripple the overall effect is striking and only advances the abilities of the XPS2 which is already a great design.
I also have looked at ripple traces with and without a load. If memory serves me correctly, the load can result in as much as a threefold reduction in the ripple, but is still much higher than the capacitor-choke combo.
1. The value of the choke (30H, <0.25 ohm DC resistance, heavy screening) made it impossible for an off-the-shelf component, so one had to be hand built (including coil winding and machining a T-yolk) just to see if this was possible. The initial measurements and more importantly the effect on subjective performance spurred me to move ahead.
2. I like to implement changes one thing at a time.
3. Cost.
I have now individual chokes on each of the +24, -24 (analog) as well as the +15, -15 (digital) for a total of four chokes, each of the same 30H spec. This takes the weight of the XPS2 over that of a 555PS. The result is that the CDS3 sounds turbocharged as compared to a standard XPS2 with a sense of dynamics, purity, timing and tonal accuracy that has got to be heard to be believed- and yes, I did have another unmodded XPS2 unit side by side to compare it with. Several weeks ago I took it over to a friends who is using a CDX/XPS1 into a 52. I also took along my 552. Power amp was the NAP500.
The improvement of the XPS1 to the quad choked XPS2 was even larger in degree to swapping out his 52 for my 552 which was done in the same listening session. The CDX was transformed in quite an unexpected way transforming into a source worthy of the 552.
Given the high current draw of the CDS3 head unit, the chokes also provide a greatly increased energy storage (as the energy stored by an inductor is proportion to the square of the current) that by my calculations are several times higher than the filter caps themselves. In conjunction with a 50-fold reduction of ripple the overall effect is striking and only advances the abilities of the XPS2 which is already a great design.
I also have looked at ripple traces with and without a load. If memory serves me correctly, the load can result in as much as a threefold reduction in the ripple, but is still much higher than the capacitor-choke combo.
PigletsDad
My intelligence test came back negative.
30-fold in amplitude is 30dB.
1Theotherone
Member
Ron Toolsie's mods to my XPS / CDS3 has to be the best VFM upgrade I have ever made, taking the CDS3 to a whole new level for a fraction of the cost of a CDP555.
I recently went from a CDX2/XPS2 to a CDS3/XPS2. I found the CDX2/XPS2 was like a teenager on steriods. Lots of energy and attack, though occasionally I experienced too much edge and digital sheen. Regardless, I found it an excellent player for the price though wanted a more analogue sounding player. The upgrade to a CDS3 certainly put to rest any stridency or harshness experienced with the CDX2. The presentation was also more detailed though I found the sound too laid back for my taste. I therefore decided that I needed to upgrade things.
I had previously successfully dealt with Ron and had no qualms in going the modification route. Unlike Naim dealers, Ron has no financial interest in the modifications, he is purely motivated in bringing the best music to the enthusiasts ears. It is his hobby, not his business. I am not an equipment buff and consider myself fortunate to have benefited from Rons extensive knowledge of all Naim equipment and other high end gear plus all the R&D he has undertaken.
The whole modification process went very smoothly and took less time than it took to purchase my CD players from Naim !
FYI, all the above CDPs (except the CD5X) have been played through a NAP 500 and NAC 552 and so comparisons have been on a level playing field.
One album I have played on each of the forementioned CDPs is Abbey Road. The difference between the unmodded CDS3 to the modded CDS3 was massive, so much so, I could have been excused in thinking I had magically obtained the remastered version to be released on 09/09/09, it was that good !
The modded sound is extremely well balanced and organic and this is prior to the end of the burn in. Every instrument is well presented and lyrics, which I often have difficulty following, are clear though more impressive is that vocal inflections are extremely well conveyed. The feel is of coherence and body with excellent bass grip. Every day I think how beguiling the sound is and cd after cd presents me with music that I was unaware existed. It really is like having a whole new collection.
Hats off to Ron for realising that a few modifications, which often involves replacing cheap components with quality components, can and does reap the listener incredible rewards. I spoke briefly with the engineer that does the work and he succinctly summarized things, by stating the Naim gear is well designed and built though there is room for improvement and the improvements make it more analogue sounding. He was spot on re: the modded player/power supply sounding analogue, in fact I never thought any cdp could sound so analogue. If analogue is not your taste then I would strongly guide the reader to the CDX2 in the Naim line.
Let's face it, I have always thought that when purchasing a high end piece of Naim gear part of the comfort in the purchase is that it really cant be improved upon. Well Naim in the past couple of years have themselves shown that there is always room for improvement; case in point is the introduction of the Hi Line and now the Powerline with the general concensus being that these minor improvements reaped significant rewards. Rons modifications are far more significant than the hi line / powerline and so are the rewards.
Highly recommended.
I recently went from a CDX2/XPS2 to a CDS3/XPS2. I found the CDX2/XPS2 was like a teenager on steriods. Lots of energy and attack, though occasionally I experienced too much edge and digital sheen. Regardless, I found it an excellent player for the price though wanted a more analogue sounding player. The upgrade to a CDS3 certainly put to rest any stridency or harshness experienced with the CDX2. The presentation was also more detailed though I found the sound too laid back for my taste. I therefore decided that I needed to upgrade things.
I had previously successfully dealt with Ron and had no qualms in going the modification route. Unlike Naim dealers, Ron has no financial interest in the modifications, he is purely motivated in bringing the best music to the enthusiasts ears. It is his hobby, not his business. I am not an equipment buff and consider myself fortunate to have benefited from Rons extensive knowledge of all Naim equipment and other high end gear plus all the R&D he has undertaken.
The whole modification process went very smoothly and took less time than it took to purchase my CD players from Naim !
FYI, all the above CDPs (except the CD5X) have been played through a NAP 500 and NAC 552 and so comparisons have been on a level playing field.
One album I have played on each of the forementioned CDPs is Abbey Road. The difference between the unmodded CDS3 to the modded CDS3 was massive, so much so, I could have been excused in thinking I had magically obtained the remastered version to be released on 09/09/09, it was that good !
The modded sound is extremely well balanced and organic and this is prior to the end of the burn in. Every instrument is well presented and lyrics, which I often have difficulty following, are clear though more impressive is that vocal inflections are extremely well conveyed. The feel is of coherence and body with excellent bass grip. Every day I think how beguiling the sound is and cd after cd presents me with music that I was unaware existed. It really is like having a whole new collection.
Hats off to Ron for realising that a few modifications, which often involves replacing cheap components with quality components, can and does reap the listener incredible rewards. I spoke briefly with the engineer that does the work and he succinctly summarized things, by stating the Naim gear is well designed and built though there is room for improvement and the improvements make it more analogue sounding. He was spot on re: the modded player/power supply sounding analogue, in fact I never thought any cdp could sound so analogue. If analogue is not your taste then I would strongly guide the reader to the CDX2 in the Naim line.
Let's face it, I have always thought that when purchasing a high end piece of Naim gear part of the comfort in the purchase is that it really cant be improved upon. Well Naim in the past couple of years have themselves shown that there is always room for improvement; case in point is the introduction of the Hi Line and now the Powerline with the general concensus being that these minor improvements reaped significant rewards. Rons modifications are far more significant than the hi line / powerline and so are the rewards.
Highly recommended.
Perhaps a CDS2 might have been the better choice?
Unfortunately anything manufactured can be improved upon and anything modded can too, you can never make THE best, its all compromise.
I would never mod a CDS3/XPS2, I think its a step too far in modding. It may have being cheaper to build a replacement PSU. It is of course YOUR CDS3 and if you fully intend to keep it for a long time then thats ok. Do buy a spare transport though.
This isn't meant to sound as harsh as it reads.
Really glad you are enjoying it.
1Theotherone
Member
Trancera; thank you for the response. You are not being harsh and I respect your point of view. Indeed, I am keeping this player for the long term and I agree that one day I should probably purchase a spare transport.
1Theotherone
Member
I have many HDCDs and have been delighted with the replay of them on the modded system.
rontoolsie
pfm Member
The mods the the CDS3 head unit are several fold
1. The reed relays are 'hotwired' while still allowing switching. There is no compromise with HDCD disks.
2. Wiring to output jacks is replaced with Cardas litz.
3. A total of twelve capacitors in series with the analog output are bypassed (not replaced) with Cardas capacitors.
4. The transport is then adjusted for the best eye pattern using an oscilloscope.
The XPS2 receives a more thorough treatment.
1. The filter capacitors to the analog and digital supplies are replaced with higher quality Xicon ones. In the case of the digital supply (which requires a higher current draw from the CDS3 head unit), the amount of capacitance is doubled from 9400 uF to 18800uF. This results in even less ripple getting into the D/A chip with a consequent lowering in jitter.
2. Each of the +24V, -24V, +15V and -15V rails that go to the analog and digital supplies are individually choke regulated, with the chokes located between the rectifier diodes and the replaced filter caps. This results in a very significant reduction not only ripple entering the filter caps, but also after the voltage regulator boards, even when they are loaded by a powered on head unit. Yes, I checked. And they confer a huge amount of extra energy storage, probably of the order of 5-10 times greater than that of the filter caps alone (remember that the energy stored by chokes is not proportional to the voltage across them like capacitors, but rather to the square of the current going through them, which in the current sucking CDS3 head unit- is surprisingly large).
3. The wiring from the rectifiers to the boards is replaced by Cardas litz, as is the wires from the chokes to the PCB.
I have had the opportunity to compare three different CDS3/XPS2 units to my modded CDS3. They each suffered greatly in the comparison, sounding spacially and dynamically flat and greatly lacking in harmonic richness. The lower frequencies were also subjectively rolled off and with a bloated, sluggish and lackluster presentation. True stereo imaging and soundstaging were also greatly compromised. One CDS3 owner was convinced that his unit just had to be broken as it sounded so insipid next to mine.
I know this sounds like a harsh condemation of the CDS3 player, and my subjective impressions sound like wishful hyperbole. But I assure you that they are not. The CDS3 redone is by a very large margin the best player I have ever heard, and then some.
I have also compared the rebuilt XPS2 with a 555PS and much, much prefer the former. However there are things that can be done with the 555PS that turns this
....into one of these... [
But that is a different story, and may be best told by a different person.
1. The reed relays are 'hotwired' while still allowing switching. There is no compromise with HDCD disks.
2. Wiring to output jacks is replaced with Cardas litz.
3. A total of twelve capacitors in series with the analog output are bypassed (not replaced) with Cardas capacitors.
4. The transport is then adjusted for the best eye pattern using an oscilloscope.
The XPS2 receives a more thorough treatment.
1. The filter capacitors to the analog and digital supplies are replaced with higher quality Xicon ones. In the case of the digital supply (which requires a higher current draw from the CDS3 head unit), the amount of capacitance is doubled from 9400 uF to 18800uF. This results in even less ripple getting into the D/A chip with a consequent lowering in jitter.
2. Each of the +24V, -24V, +15V and -15V rails that go to the analog and digital supplies are individually choke regulated, with the chokes located between the rectifier diodes and the replaced filter caps. This results in a very significant reduction not only ripple entering the filter caps, but also after the voltage regulator boards, even when they are loaded by a powered on head unit. Yes, I checked. And they confer a huge amount of extra energy storage, probably of the order of 5-10 times greater than that of the filter caps alone (remember that the energy stored by chokes is not proportional to the voltage across them like capacitors, but rather to the square of the current going through them, which in the current sucking CDS3 head unit- is surprisingly large).
3. The wiring from the rectifiers to the boards is replaced by Cardas litz, as is the wires from the chokes to the PCB.
I have had the opportunity to compare three different CDS3/XPS2 units to my modded CDS3. They each suffered greatly in the comparison, sounding spacially and dynamically flat and greatly lacking in harmonic richness. The lower frequencies were also subjectively rolled off and with a bloated, sluggish and lackluster presentation. True stereo imaging and soundstaging were also greatly compromised. One CDS3 owner was convinced that his unit just had to be broken as it sounded so insipid next to mine.
I know this sounds like a harsh condemation of the CDS3 player, and my subjective impressions sound like wishful hyperbole. But I assure you that they are not. The CDS3 redone is by a very large margin the best player I have ever heard, and then some.
I have also compared the rebuilt XPS2 with a 555PS and much, much prefer the former. However there are things that can be done with the 555PS that turns this
....into one of these... [
But that is a different story, and may be best told by a different person.
rontoolsie
pfm Member
The last image is supposed to be a smaller, resized version but for some reason does not pick up the link from my hosting site. EDIT...it does now!
The 555PS 'massive loss of capacitors' is actually a capacitor bank of identical value uF as the original Felsic/SAFCO CO ones, but in a smaller package with lower ESR. And each of these have been bypassed (underneath their circuit board) by Cardas caps. And again, the energy storage is greatly increased by the massive inductors as compared to its original storeage abilities.
The inductors are chokes each circa 40 Henries, identical to those used in the XPS2 rebuild.
You may also notice the smaller motherboard has been hotwired (some of which is unseen as it is on the underside of the board)...the secondaries from the main transformer are obliged to go through angel-hair skinny traces in this motherboard, which are pretty much guaranteed to be current limiting. These skinny traces were jumped with much lower gauge value wires.
The end result is the 555PS sound literally about twice as good as it did in the stock form. And it is just not me who thinks so!
The 555PS 'massive loss of capacitors' is actually a capacitor bank of identical value uF as the original Felsic/SAFCO CO ones, but in a smaller package with lower ESR. And each of these have been bypassed (underneath their circuit board) by Cardas caps. And again, the energy storage is greatly increased by the massive inductors as compared to its original storeage abilities.
The inductors are chokes each circa 40 Henries, identical to those used in the XPS2 rebuild.
You may also notice the smaller motherboard has been hotwired (some of which is unseen as it is on the underside of the board)...the secondaries from the main transformer are obliged to go through angel-hair skinny traces in this motherboard, which are pretty much guaranteed to be current limiting. These skinny traces were jumped with much lower gauge value wires.
The end result is the 555PS sound literally about twice as good as it did in the stock form. And it is just not me who thinks so!