shauns@vice.UUCP (Shaun Simpkins) (02/11/84)
A nice posting, Greg...I have no argument but a small clarification of filter
types. If we take into account the stated frequency response of the Sony unit
and the transition band requirements there is NO WAY a 9th order Bessel filter
with its very gradual slopes near cutoff can produce over 70dB of attenuation
with only a 10% frequency change. The Sony units must use a non-linear phase
reconstruction filter to obtain the very steep transition slope needed near
F-3dB. Proof of this can be had by examining the step response pictures of the
CDP-101, Kyocera, and Phillips units. The Kyocera and Phillips units both
have lower overshoots than the Sony unit. They also have a little Gibbs
ringing on the tail end of the step, whatever that signifies (other than it
is an artifact of digital filtering). I've also done some simulations of
9th-order Thompson-Butterworth and Bessel filters which agree with the test
results I've seen.
The Philips unit does use a low order Bessel filter. The constant envelope
delay of this type of filter would imply that the Phillips should perform
better than the Sony (if phase smearing is audible), but curiously,
I've heard several reports (not on this net) that the Philips sounds poor
compared to the other units I've mentioned. I don't know of any salon in this
area that stocks Philips, and so can't vouch for this opinion. I should also
mention that many other things can factor into sound quality, such as buffer
amplifier design, different methods of error interpolation, etc., so one
should not take this as a condemnation of digital filtering.
The latest Yamaha CD-X1 has taken a cue from all this hullaballu and
incorporated 2X oversampling/digital filtering into its 16-bit system to a)
make a cheaper reconstruction filter and b) avoid envelope delay problems.
It's the only 16-bit system that I know of at its price that uses oversampling.
I'll be listening to it sometime soon and will report back on my findings.
In search of the $300.00 quality CD player,
--
Shaun Simpkins
uucp: {ucbvax,decvax,chico,pur-ee,cbosg,ihnss}!teklabs!tekcad!vice!shauns
CSnet: shauns@tek
ARPAnet:shauns.tek@rand-relay