Adding a custom DAC to a CD player

I've been working on a external DAC design, but I'm convinced that a custom DAC and clock upgrade inside of an existing CD player is a better way to go. Avoid all that nasty SPDIF stuff altogether.

I have a Philips CD920, an early 90s unit, that I'd like to use as a test platform. I have the repair manual for it, and it uses a SAA7431 servo/decoder/DAC chip. I don't see any obvious pins for I2C data, only SPDIF out. I attached the schematic that shows this chip. I cannot find the datasheet on the web anywhere, even at Philips.

Any thoughts?

I also have a Marantz CC-67, but it is a relatively recent purchase that still has a warranty...



  • cd920.jpg
    55.3 KB · Views: 1,500
You need to find a I2s line carrying the 44.1/16 audio data and associated clocks and route that to your custom DAC pcb which I assume will contain a suitable oversampling filter and DAC chips.
You may also want to consider placing the clock on your DAC board and sending that back to the player. This should solve any jitter issues.
Since you probably have to get a new CD-player, if you still want to go with your idea, take a look at cambridge audio D300. It's a 350$ unit and is buildt to be modified. Seperate psu, servo controll unit and dacboards. It even has a shielded toroid transfomer. If you want to improve a cheap CD-player, this is it. The factory will introduce their own upgrades later. I can spot a unused HDCD segment in the display just waiting...

I don't know if the drive (Sony) is high enough quality to reach true audiophile quality, though. - Anyone???
Thanks for the replies!

Pin 32 is the SPDIF output.

The Cambridge Audio D300 sounds like a good machine to start with, especially if it could be found used!

There are a number of test connections on the chip (Test1-9, pins 66-78) which I think are worth a look at, but haven't had the time to dig into yet. The chip is on the backside of the main board, so probing is a bit of a challenge.

I did find that the supply lines are TRASH. How about 250mVpp of high frequency hair on the +5V line? AND, the +/-10V lines to the analog output stage are regulated with a zener/resistor only. I don't need to tell you how bad those lines look.

So far, I think my upgrades to this player will be limited to a better clock (making a Kwak Clock board now), and adding more supply regulation.

Well, I finished the following mods to this player:

-Built and installed a Kwak Clock
-Replace all power supply diodes with fast recovery types
-Replace most supply and local chip electrolytics with much larger values
-Replace output opamps with OPA2604
-Remove output muting transistors
-Add +/-8V regulators to analog output stage

For this effort, this player now sounds noticeably better than my Marantz CC-67. The highs are much clearer, cymbals sound much more distinct, and sibilance on voices is much lower.

BUT, there is still some sibilance on vocals that I would like to get rid of. Anybody have any thoughts on what might be the culprit?

Also, there is noticeable amount of 80kHz signal at the output, even when there is no disk playing. This signal is also present at the output of the DAC chip. Any ideas on this?

My suggestion is not to redo the whole CD in order to get better sound.

Think once or twice if you ought to do something about the speakers? What type do you use?

I would say that the biggiest improvement is changing the speakers! I changed my small Infinity to Martin Logan SL3 when I moved (and got better space). My music listning is before martin Logan and after. It's really more fun with cool speakers and not "spitzenklasse" of the signal sources and the other way around (good source and poor speakers).
I'm using Vandersteen 2ce's, which are not quite at the Martin Logan SL3 price point, but are certainly no slouch. I heard them at my dealer through a Musical Fidelity A3, CD player I don't recall. They were absolutely smooth and clear through the high end.

My main amp/preamp is a Denon AVR-3300, but I have tried going straight from the CD player into a Harmon Kardon Citation 16 power amp with just an attenuation network. Same exact sibilance and graininess in the high end as the Denon. (Harmon Kardon was more effortless, however) Since I was able to partially tame the high end by CD player tweaking, my thinking is that there is more room to go. I may very well have reached the point of diminishing returns with this player, however!

Reclocking and zero-oversampling are certainly valid things to try, but the SAA7431 in the Philips 920 handles all input conversion, oversampling, D/A, and I/V conversion, so not much I can do there! BTW, your clock circuit works very nicely, and I believe it made the greatest sonic improvement.

as everything is inside the decoder chip, further effort is pretty much wasted unless you bite the bullet and install an SPDIF receiver and a new DAC. As long as you have a good master clock and use this to clock both decoder and DAC, you don't have to worry about jitter being induced by the receiver chip. If there is a way to run the receiver in slave mode, i.e. without its internal VCO, it will work even better.
I'd check first if the DOUT pin is really active because in most Philips decoder chips, it can be deactivated by software (but this is rarely done).

I have a cheap, cheap Kenwood player with a simple Sony drive. The supply and grounding scheme is just terrible. In fact, just increasing the main electrolytic made it blow its soldered TO-92 fuse.

Clocking this player in slave mode, i.e. from an external DAC, I got superb sound. So as long as the drive reproduces the digital signal correctly (no missing data), I doubt there is any need to get into optimizing the drive and drive electronics (appealing as it may be).