To oversample, or not to oversample, that is the question. And do you need it? If you would believe all the stories that have been written about the subject, then we must be stupid that we are still running a CD-94 with oversampling. Non-oversampling is the way to go!
In my opinion, you first would have to start by asking yourself, what is oversampling doing in your beloved Marantz before you disable it. And if you do it, what are the benefits, if there are any. The oversampling in our Marantz CD-94 is done by the SAA7220p/a or p/b (depends on your model). It provides some really useful features as there are:
- Interpolation of data in error
- Finite impulse response transversal filtering with four times increased sampling rate
- A digital audio output
Features of the SAA7220 Filter;
The SAA7220 is a dual-channel linear-phase FIR digital filter chip. The SAA7220 chip employs a four-times oversampling design. The number of filter coefficients has been increased over the first-generation filter by 24 per channel to a total of 120. Passband ripple has been reduced to a 0.02 dB to reduce the possibility of that artifact’s audibility. (Experiments have shown that pass-band ripple of +/-0.2 dB can be heard.) Besides, the SAA7220 mutes the signal smoothly so that no clicks or pops occur when the player starts, stops or pauses. Uncorrected data errors stemming from burst errors are masked with an 8-sample interpolator on-board the SAA7220 to create a smooth bridging waveform over the bad samples. Improved interpolation and error correction results in a significantly better tolerance to burst errors compared to earlier chipsets. The SAA7220 can deliver a digital-format audio output for interfacing to out-board digital signal processors. 
The oversampling in the SAA7220 works by increasing the sampling frequency four times, to 176.4 kHz. This results in a frequency spectrum containing multiples of the original 44.1 -kHz sampling frequency, at 88.2, 132.3, 176.4 kHz and so on. By oversampling four times, the noise power originally restricted to a band from 0 Hz to 22 kHz is now distributed over a band four times as wide, or 0 Hz to 88 kHz. Only one -quarter of the noise remains within the audio band, and the rest will be eliminated by filtering, giving a 6 -dB improvement in performance. 
So why did a lot of people convert their players into nos-mode? Is it the bad reviews that the SAA7220 got? I name a few of them;
- I’ve read elsewhere that the SAA7220 is a very noisy chip and tends to propagate a lot of noise back into the power supply to the detriment of the rest of the components.
- If we look at the situation in an ordinary CD player, we find that the master clock oscillator circuit is part of a larger device, in Philips style players it’s the SAA7220 digital filter chip. The oscillator circuit is almost certainly a standard-cell type. Although performing quite adequately as the clock generator in most digital systems, its jitter performance is terrible. Its power supply rejection and temperature stability are also terrible and remember this is just a tiny circuit stuck on the edge of a much larger chip and will be subject to a large amount of power supply modulation noise since it is fed off the main digital supply.
Or are the reported results by people causing this popularity?
- The sound is truly fantastic and much better than the standard configuration. Technically it isn’t correct but to my ears, it sounds more analog type.
- Have you tried the zero oversampling mod? It’s fantastic!
- After the mod — the music flows somehow easier, in a more liquid, more natural fashion. It does not lose any of the good quality, none of the power or attack of bass — everything becomes just a little bit better. Don’t expect miracles — I said a LITTLE BIT BETTER.
- What you gain is more timing accuracy — because every steep filter (digital or analog) will deliver a sort of pre-ringing (and ripple).
But instead of the ‘beloved audio nirvana’ that every audiophile is looking for, the non-oversampling modification introduces some nasty side-effects. It introduces much more noise in the audio spectrum getting worse at higher frequencies, and harmonic distortion increases. Another side effect that’s audible: a rather sharp roll-off in the higher frequencies.
It is also rumored that better decoupling of the SAA7220 gives much better audible results. Use of chokes and ferrites, or even a separate power supply line for the SAA7220. Recently someone measured the differences with different kinds of decoupling (with and without choke or ferrite) and the results were negligible and probably not audible. The same goes for the use of oscon capacitors on the PD16 board, not audible as well and if you have to change these capacitors because they are too old: a good low-ESR cap will do the trick.
But how does a Marantz CD-94 sounds without all aforementioned modifications? Absolutely fabulous! Last year I burned a concert of Youn Sun Nah on CD and that sounds so beautiful on my cd player. Given the fact that the sample that I used was only a 320kbps! Youn Sun Nah did a memorable performance in the Théâtre du Châtelet recorded by Arte France. I think it is still a wasted chance that “they” never released this on c.d. The accordion played by Simon Tailleu so subtle, really ‘touches’ you and sends shivers up and down your spine! You can hear the acoustics of this beautiful theatre and opera house in the recording! Go listen for yourself and you will be stunned!
Two other good examples of recordings that bring you the best of the Marantz CD-94 are a ‘Waltz for Debby’ and ‘Sunday at the Village Vanguard’ by the Bill Evans Trio. Both recorded on the 25th of June 1961 at the Village Vanguard club in New York City are such rare moments of recording history. If you want to understand these recordings and the way the trio played; Marantz is your perfect partner. The way it is reproducing Scott La Faro playing his double-bass in these sessions is remarkable. Drums sound like drums, with life-like snare drums and hi-hat.
I experimented for some time with op-amps, and in my set-up, the AD711 was not that bad. I was hoping for big results changing them to other types but was a bit disappointed by the way it turned out. When I changed them to 4x OPA627 or 4x OPA604 I missed the way instruments sound. When I changed to the combination of OPA627 and OPA604 all pieces fell in place again. It reminded me of the sound of my first Marantz CD-94 with its original NE5534s. Remember: this is pure personal!
Keep in mind that the engineers who designed the Marantz CD-94 were no amateurs. The engineers all had an engineering degree, and trying to better their designs is an illusion most of the time. Scientists from Philips and Marantz were rewarded with several awards for their work. By just giving the Marantz CD-94 regular maintenance, it will let its qualities shine through! Some minor adjustments can be done, but they will not cost you a fortune!
Loudspeakers: Impulse 54 (Transmission Line)
Amplifier: Copland CTA501, Sansui AU555a
Speaker cable: Duelund DCA16GA, Belden 9497
Interconnects: Belden 8402, Duelund DCA16GA
 From the compact disc handbook by Ken C. Pohlmann
 From Dutch Audio Classics website (must visit!)