advertisement


Mods to Naim CDS3

rontoolsie

rontoolsie

pfm Member
There has been much written about modifying the Nac 32/62/102, but nothing at all about tweaking/modding their higher end gear. In the last few months I have had mods done to the CDS3, the XPS2 and the 552 PS.

For now I will only address the mods done to the CDS3 head unit. For now.

The CDS3-unlike its 1/2 predecessors has the luxury of both DIN and RCA outputs that are independently switchable on/off. This is done by a pair of reed relays for each output, one of which switches between the hot signal and the other shunts selected input ground to ground.

Now there is no need to switch BOTH the signal and ground by independent relays... this subjects the critical signal to possible non-linearities of the relays, which in this particular instance is -2dB at 20Hz and -.2dB at 20kHz (or is it the other way around).

So, the first step was to identify where in the PCB the signal hot entered its relay and to hotwire from there directly to the DIN and RCA outputs.

Like this.

CDS3_reed_relay_bypass_4.jpg


The result... a cleaner, sweeter and more articulate top end and a more tuneful and firmer bottom end with less hash in the background. Well worth the price for the cost of a few inches of Cardas litz chassis wire.

The next step is to look at the output capacitors in series with the signal...
 
These are 10uF 35V blue tantalum bead caps to sited to the left of the reed relays pictured above. Good sounding caps for sure, but no immune to the benefits of being bypassed with high quality ones. I have found that the Cardas caps are unsurpassed in this application, so next in go a pair of 0.1 uF Cardas caps, which are about 20x as large in size but only 1/100 times the electronic value of the tants. This required a little topological exploration to get them best sited.

Like this.

CDS3_bypass_output_caps_2.jpg


The result.... even larger than bypassing the reed relays. Everything is more organic sounding, cleaner, smoother, more detailed and better timed. Well worth the cost of the capacitors (circa $12 USD each). Not for the faint of heart due to the very small space within which to solder the leads.
 
Ron,

Hats off to you - you must have the biggest pair of cochonnes on the planet to be messing around inside a CDS3 !

Looking forward to more pron!


Jim.
 
The bypass capacitors have been in situ for 48 hours now...like the other Cardas caps they require a full 200 hours to hit their peak. What I am hearing with in place is a clarity and harmonic correctness that makes the unbypassed output caps to sound a tad tonally threadbare. To be sure the modded CDS3/modded XPS2 do not have the startling realism that the 555 player I had inhouse today did, but still signficantly ahead of where it used to be.

Please note that my cohones are not large enough to put my untrained fists inside the CDS3...I merely assumed the financial risk if these tweaks were to have backfired. Instead I used the same person to effect these mods who also designed and built my custom DBL crossovers-which have with these latest additions revealed how natural and articulate the DBLs can actually sound.
 
Are you sure this is wise Ron.? You could be up to your neck in sheep droppings before too long. Anyway, congratulations on having escaped from the box buying treadmill and welcome to the wonderfully satisfying world of bodgineering....
 
Definitely not a unit to be modding on after a date with Stella Artios ;)

Thanks Ron, the more details the better IMO - I for one would expect that there could be some useful ideas falling out of this for those with older Naim items.

*takes sideways glance at the relays in his pre and starts searching for info on Cardas caps*
 
The xps mods effect an even larger improvement. One other person has had the same thing done and we both agree the result is bigger than moving from a 52 to 552, a move both of us have earlier made. Details will follow shortly. Beware though that the vital part is not commercially availble and has had to be handcrafted to spec. Now, how is that for a tease!
 
rontoolsie said:
Beware though that the vital part is not commercially availble and has had to be handcrafted to spec. Now, how is that for a tease!
Click to expand...


An inductor...?
 
Spot on!!! The trick is to get a 30 Henry inductor with 0.3ohms DC resistance and capable of handling 3 Amps of current (there are like 15 LM 317s in the analog boards of the CDS3) to enable a choke regulated power supply. Hand winding on a very high permeability T-yolk is the answer, followed by intensive screening. Pictures and subjective details to follow in the next day.

Jo Sharp said:
An inductor...?
Click to expand...
 
Looking forward to the details of the implementation. Have you also looked at replacing the 317 regs with something better? I'm thinking especially of the analog regulation and also the dac/clock/buffer areas, too.

Cheers,
Carl
 
The insides of the XPS2 have been shown a time or two before. As found in another thread there is a +/- 24V for the analog circuits (feeding some 15 LM317 inside the CDS3), +15V for the digital chips and +/- 10V for the transport/logic. Although the 24 is split rail, this is achieved by taking in a single + voltage then using a voltage splitter and inverter to come up with the negative leg, instead of using different transformer windings.

Putting the XPS2 on a scope and meter showed about 2.5mV in AC ripple/noise that was finding its way both into the digital and the analog voltage feed. Relatively low, but still higher than the zero tolerance.

So we decided to experiment with a choke regulator, initially just on the analog voltage supply which would not only smooth out the ripple but also act as an extra resevoir of energy storage. The fact that the +/- legs are derived from a single + feed meant that only a single choke would be needed in this application. EDIT...THIS IS NOT ACCURATE. THERE ARE IN FACT SEPARATE SECONDARY WINDINGS FOR EACH OF THE +24 and -24 VOLT RAILS. SUBSEQENT CHOKE REGULATION OF THE -24V RAIL PROVIDED ANOTHER REALLY LARGE IMPROVMENT, AS DID ANOTHER TWO CHOKES (FOUR TOTAL NOW) ON EACH OF THE +15V and the -15V. WAAAAAY, WAY BETTER THAN A 555PS I HAD AVAILABLE TO DIRECTLY COMPARE IT TO.


Given the 15 LM317 in the analog boards, each using 30-40 mA of current it was decided that the choke needed a current rating of at LEAST 3A. And instead of having a few Henries, why not try for 20-30 Henries. And...if wishes can come true, why not go for a DC resistance of less than 0.5 ohm. A value of 30 Henries would give an impedence of over 20,000 ohms to 120Hz ripple, while a low DC resistance would not drop the voltage beneath the abilities of the voltage regulators and keep a low output impedence.

To give you an example how seemingly impossible these parameters in conjunction are, the 'high value' choke that I had fitted into my Epona SPS power supply is 15 Henries but with only 150 mA of current and about 150 ohms of DC resistance, and is a goodly size.

After 2 weeks of searching we found out that there is not a single company who sold such rated chokes. One company who custom winds to 'any' specification said (by the engineer that designs all of their chokes) that this could not be done within a 3.5 inch diameter coil.

The only way was for Daryl (a driver/speaker/amplifier designer who has hand wound more than his share of prototype voice coils) to hand wind such a choke using 12 gauge copper (instead of 22 ga found in most chokes) around a very high permeability metal T-yoke that had to be obtained from sacrificing one of his own design drivers. This required about 2 hours of labor chipping off the ceramic magnet that was glued onto the yoke by some super-industrial adhesive that refused the intrusion of any of the solvents that was tried. FYI a ceramic magnet is like glass and its removal resulted in sharp shards flying across the room.

Once a value of 30 Henries was obtained in a relatively compact package the unit was then heavily screened, and potted. The center was also tapped out to allow it to be screwed into a chassis.

The next step was to instert said choke into the analog voltage feed of the XPS2....

See below for pictures and subjective evaluation.
 
top_view.jpg


The choke is the cylindrical thing above the transformer with two wires going in and out. The bar tag was lifted from something entirely unrelated and serves only to cover a central tapped hole and maintain some sort of semblance that this is a legitimate product.

The sound......

Within 10 minutes of staring up the CDS3 I knew that this was one of the biggest black box upgrades I had ever made. Yes, larger than the 300 to 500 and larger than the 52 to 552.

There was a great increase in frequency extension at both ends, a massive firming up of the bottom end and a yet unprecendented degree of crystalline purity. And a dynamic performance that made the unchoked unit sound dull and insipid.

I had to have my impression independantly cofirmed, so I overnighted it to a friend in California ($110) where he too found this to be the single biggest jump he had yet made. So much so that he sent his XPS2 overnighted here to have the same thing done. We both agree that there is absolutely no subjective downside of the choke.

If anybody else would like to have this mod effected, Daryl has agreed to do it for a price that is way fair given the time involved in constructing the choke and the need to sacrifice a perfectly good drive unit.

I may also be persuaded to lend my tweaked XPS2 to somebody in the continental United States if they pay freight both ways.

But wait...there is so much more that can/has been done to the XPS2. Details to follow.
 
Given the success of bypassing the output caps on the CDS3, why not try bypassing the filter caps of the 552PS with similar Cardas caps??

Why not indeed.....
552PS_with_Cardas_bypass_1_ringed.jpg


The front pair of filter caps are tied together and hence only one bypass cap is needed. The rear cap had a busbar that soaked up all the heat of the 100W soldering gun. So this was bypassed where it feeds into the slew of voltage regulators. As can be seen this is nestled between the rear and the next-to-rear voltage regulator.
 
Ron,

Do you have the choke in one DC rail only?
Isn't there numerous DC rails coming out from the t-former and the rectifiers?

Presumably in the analogue amp feed. But isn't there minus and plus power inputs going to the analogue side?

Aha, you have explained it earlier. Sorry!

Great photos and information! Thanks!

Oz
 
Im finding it difficult to watch this thread!

I do wonder if the extreme nature of these mods financially is a step too far. May of us here would love to be able to afford this equipment and here it is being 'improved' because the Bass needed firiming up at the bottom end (ok you didnt say it needed it but claimed it as the result)

I also wonder if there would be a much better way to do all of this. Buy a NAP second hand in good nic and mod it that such that it does what the XPS2 does. Sell the XPS. Im not sure of the prices but I would have thought it would be way cheaper, you can of course sell the innards of the NAP for a decent sum anyway.

You also get the benefit of testing the new psu against the original XPS until you are happy it betters the original.

I realise thats its yours to do as you wish, but for such costly and admired equipment i find it strangely cruel!

</shutsup>
 
The problem is the transformer in a NAP is completely wrong (not enough feeds, wrong voltage, insufficient number of rectifiers) and would have to punch the back chassis and insert a Burndy socket. So in effect a NAP would only offer you a chassis that would need extensive modification, and you would probably be better off crafting an entirely new design de novo.

The DBL speakers I use are hugely sensitive to the quality and quantity of bass and these mods have taken them to an entirely new level. I did try my modded XPS2 on a friends Tukan-based system and the improvement in tune at the LF was quite audible too.

Remember that an older XPS can be purchased for not too much over $2000, and taken to a level way above a new XPS2 for almost $2000 less than the price of a new XPS2. Cruel to be kind maybe?

Buy a NAP second hand in good nic and mod it that such that it does what the XPS2 does.
Click to expand...
 
You must log in or register to reply here.


advertisement


Back
Top