advertisement


Mods to Naim CDS3

The beginning...in the picture you can see a small silver banded resistor on the right side-this is the stock one. And on the left is the much larger gold-banded tantalum resistor. These double sided boards make it a slow and arduous process. It takes a minimum of 20 minutes to pull one resistor out, clean up the pads, check its value compared to the schematic and solder the replacement in...and there are over 50 to do!
CDS3_board_2.jpg
 
rontoolsie said:
And here is a picture of the choke regulated XPS power supply. There is one 30H choke for each of the 15+, 15-, 24+ and 24- supplies. In addition to a fifty-fold reduction in noise, the combined 120H of inductance also offers huge extra energy stores. The filter caps for the digital and analog feeds are also bypassed with their own individual Cardas capacitor. The circuit board had to be moved to allow the 4th choke to fit, requiring a lot of countersunk holes to be drilled into the chassis.
XPS2.png
Click to expand...

What chokes did you use please?
 
The chokes are addressed much higher up in this thread-but they are proprietary due to their unique trifecta of very high inductance, very high current capacity and very low DCR. Normally you can get to pick two of those parameters and forfeit the third. Nothing remotely like them are commercially available.
This is sort of what they do...
The feed out of the rectifiers before the filter caps...500 mV of spikey ripple.
beforechoke1.jpg


And the output from the choke looks like this (which is what now enters the VR boards)...note that the scale is very different to allow the residual ripple to appear. The original 500 mV is now around 10mV. A similar order of ripple reduction was also seen AFTER the voltage regulators, with the final ripple being under 0.1mV, down from a handful of mV before the choke. The 317 are very good at rejecting LF noise, whereas the chokes work higher up, and the combination has a very low noise floor and a pretty benign pattern. Similar chokes have been used on Supercaps, 555PS, 52PS and even in the 500PS, all with great benefit. As mentioned above I think its a combination of much lower noise and higher energy stores above what the filter caps alone can allow.
afterchoke1.jpg
 
rontoolsie said:
'Other front runners' were CDS2, CDS3 and I believe CD555. All in the context of an active DBL system, owned not by Naim-but by an individual who had some design input into the original CDS
Click to expand...
Hi Ron, I suspect the CDS prototype you mention resides in a friend's system, which I'm very familiar with! Indeed, the prototype has ventured out into my system for a listening session, and very good it is too - certainly better than my then 555. Although said player did appear to give up the ghost a short while ago, it was revived and AFAIK still does its excellent thing.
 
Yes, Tony...this is a unique product and if your friend lives in the London area it is the same. Do you know what are the main differences between the prototype and the production models of the original CDS? The CDS3 is also a wonderfully designed machine, and let down in several areas in fairly predictable ways, all of which were due to simple cost constraints. If these modded players were production, they would have had to be sold for at least what a CD555 goes for and probably would exceed the 555 performance (that has many of the same constraints as the CDS3).
I wish I had held on to my clapped out CDS1..I'm sure it could have been revived and modded to within a nanometer of its life. There was that 'je ne sais quois' that the CDS1 had, that the later players did not. I will be comparing my modded CDS3 to the latest 'uber-modded' one in the next few days.
 
Not an easy job...the plus-size of the tantalum caps and their close proximity required liberal use of stand-offs. Now it is at least 30 minutes per resistor swap them out. A two day job has turned into a 4 day one. Also note the Cardas bypass caps in situ.
CDS3_board_resistor_standoff.png


CDS3_board_resistor_standoff_2.png
 
rontoolsie said:
Yes, Tony...this is a unique product and if your friend lives in the London area it is the same. Do you know what are the main differences between the prototype and the production models of the original CDS?
Click to expand...
Well, it's not in a case! First time I saw it I assumed it was just another, very dusty, ciruit board. Sorry, I don't know how it differs from the CDS, but I do know it would have been totally impractical, as well as hugely expensive, to productionise it. I do have some photos, but the owner's previously requested I don't reproduce these.
 
Suffolk Tony said:
Well, it's not in a case! First time I saw it I assumed it was just another, very dusty, ciruit board. Sorry, I don't know how it differs from the CDS, but I do know it would have been totally impractical, as well as hugely expensive, to productionise it. I do have some photos, but the owner's previously requested I don't reproduce these.
Click to expand...

Shame - I understand (I am not allowed to show internal shots of my SE hybrid monos...but that is mostly because there is very little in there, and easy to work out weird implementation), but I'd love to see them...any chance of checking again? :)
 
Yes, I think there would be a historical interest in something that probably was fabricated close to 30 years ago, and is beyond the cloning abilities of this populace. I for instance had the schematics of the CDS3 for ages, but did not share them until the CDS3 was no longer a production model. But once it was voluntarily discontinued and the commercial level of interest in making CD-players is near zero, then there is neither harm nor foul.

I could post 1000 pics of CDS3 circuit boards, but without a schematic, nobody would be able to figure out the topology.
 
So the player was re-assembled this afternoon, and I had a 45 minute listening session. The latest round of tantalum resistors, junking some of the electrolytic input coupling caps and improving some of the RC networks made an overall massive change in presentation. Dense musical threads are now very separate and yet cohesive. Bass sounds like you have gone from a 250-2 power amp to a 500, with texture and intent, and vocals just float in their own Hilbert space. The Cardas caps in particular need many hours of run in to sound half decent, and at this point they have had around 5.

But I am putting money where my mouth is....I am also having my player upgraded to the same specification. Its been a twelve year journey and maybe finally I have reached a point (with the CDS3) that is a final destination. As much fun as the journey was, the final port is even better.

One thing to mention....the player at first would play a disk all the way through, but could not random access. Since we have a half-dozen NOS transports we came very close to replacing the old one without question. But a bit of digging around showed what the culprit was. Excessive slop (maybe 3-4mm of play) due to two worn plastic gears. They were readjusted such that fresh plastic now contacts fresh plastic and it now tracks like a champ.
 
Ron

I would have sniped the resistor in half and pulled out with the help of a soldering iron each end, solder sucker the holes and dropped in the new resistor and soldered it in under a minute tops.

Now change those signal rants to film caps.

Pete
 
There were a few reasons it was not done that way...

1. first we wanted to measure the value of the resistors to see if they matched the schematic (they did)...

2 . Resistor leads were minutely wider in cross section than the holes...so it was not a simple drop in.

3. We wanted to reform all solder joints with the superior sounding Cardas eutectic solder.

4 . about changing the signal tants to film caps...many signal caps were bypassed with cardas. We are entertaining replacing all the bead caps with brand new ones as the players are all over 10 years old.

So it was 4 days of hard slog...but the results were totally worth it.
 
Then snip one end of the resistor lead and measure, we used to do this all the time in TV repairs, if it was o/k then a blob of solder to reconnected it.

How did you get over the too small hole/to large resistor problem?

The tants give the "NAIM" sound IMHO, film caps sound so much nicer.

Pete
 
MJS said:
Those 47uF caps are bipolar electrolytics. Naim like the sound of them, there's no need for their bipolar capability in pretty much all the circuits they're used in. Other caps are available, YMMV.
Click to expand...
Some more follow up on these Promisic CO31 47uF lytics. The four of them in the CDS3 form part of a noise shunting/current source. Here is one that is 50% of the nominal value.
47uf23.jpg
even worse.....
47uf3-1.jpg

This one is is over 90% off target. This very low value means that only the highest frequencies of the noise spectrum will be shunted to earth. Everything else will be faithfully transmitted. And LF noise is what can interfere with 'timing' as we know it.

The two remaining ones each measured over 60uF!


Now for the oil-filled lytics that will replace these gold caps
image000003-2.jpg

These four are each within +/- 4% of the specified 47uF.

The 552 and Snaxo are filled with shed loads of these gold caps.....

So-out with the olde, and in with the new!
image000000.jpg


Here is where they are used in the CDS3
47uf.png
 
Last edited:
If you are that far in - see R16? split that into 2 x 22K resistors; connect their centre connection to the -12v rail with another 47uF cap. (4x 22K, +2 x 47uF reqd in total)

It'll make a bigger difference.
 
It works any time you see this CCS style used in Naim - which is everywhere in the output end of their line and dedicated sub-circuit boards.

It really 'stiffens' the CCS and the values are non-critical: take the ccs tail resistor, split it into two convenient and equal values and tie the centre point of that split to the rail the current source is referenced-to. A simple if incomplete summary would be - what you now have is a brutal RC filter on the supply to the ring-of-two transistor pair which attempts to maintain the set current as constant.
22k/10uf is a filter knee <1hz. It's all you need, and values can be flexed to suit what you can fit.
 
You must log in or register to reply here.


advertisement


Back
Top