shauns@tekcad.UUCP (09/01/83)
Ah, yes. Mr. Francois has found out the problem with extreme single-band
compansion. I have used a dbx 228 with my Nakamichi for 5 years, and it
works well except on source material with radical rapid changes in amplitude
and a high noise level, such as piano music on LPs. Getting into the Pink
Floyd discussion, dbx also breathes on the heartbeat at the beginning and end
of Dark Side of the Moon.
Outboard dbx units or better tape decks do not solve the breathing problem; the
cause lies in the fact that dbx uses a 2:1 compression ratio independent of
frequency (which translates to >40dB of noise reduction).
If the recorded signal does not mask the noise by being in the same
frequency range, the result is perceptible noise floor modulation.
Dolby addresses this problem by limiting the amount of compansion to <20dB and
by using a sliding cutoff filter to constrain compansion to the frequency
range of the source. Gain is thus modulated only in the band that the signal
occupies, which reduces breathing. Because of this, on the source material
mentioned above Dolby C will have less breathing than dbx with comparable
noise reduction.
Dolby B and C, however, are absolute level sensing variable compansion ratio
systems and therefore are extremely sensitive to tape deck frequency response
aberrations and
Dolby reference level mismatch. Dolby C, with its greater noise reduction, is
correspondingly more sensitive than Dolby B. This is one reason why it's
difficult to transport tapes between decks and obtain the same frequency
response.
Since dbx's compansion ratio is constant over amplitude,
one can trade a few dB of ultimate S/N for high frequency headroom
by moving the compressed input signal around within the recorder's
dynamic range.
Dolby can't do this; its fixed threshold makes it strictly a noise reducer.
To pull off this trick, processing must occur before record levels are set,
easy with an outboard dbx.
I suspect that most onboard dbx systems process the signal after record levels
are set, thereby eliminating what is for me dbx's most useful feature.
dbx does not breathe if source noise levels are low, such as in the recording
of live music. For LP transcription, Dolby C is completely sufficient.
What I would like to see from dbx is a 2-band compander along the lines of their2bx dynamic range expander. This would eliminate breathing and negate Dolby
C's claimed advantages. Unfortunately, for most source material and most
listeners dbx works fine; dbx isn;t about to fix something that ain't broke.
Oh, well - nothing's perfect...
--
Shaun Simpkins
uucp: {ucbvax,decvax,chico,pur-ee,cbosg,ihnss}!teklabs!tekcad!shauns
CSnet: shauns@tek
ARPAnet:shauns.tek@rand-relaygregr@tekig1.UUCP (09/07/83)
Your problem with DBX is characteristic of this type of noise reduction unit.
It is caused by the fact the DBX uses a much higher expansion/compression ratio
than Dolby, and is not frequency or signal level selective. This method used bydBx produces a much better signal-to-noise ratio improvement than Dolby, but
at the cost of producing "breathing". At low signal levels (or between record
cuts) the dBx system compansion (expansion/compression) reduces tape hiss, or in
the case of encoded records, background noise almost entirely. However, should
loud solo passages occur which have narrow frequency content the system opens
up and reveals the background noise as well. If sufficient frequency content
in the midrange-high end is included in the passage it masks the noise level,
otherwise you hear the noise which appears to be modulated be the varying signal
level. This is the "breathing" you hear. In many types of music, particularly
popular music, the masking effect caused by a constantly wide frequency
spectrum can give very acceptable results. In music that features a lot of solo
instruments the breathing can often be unacceptable.
Unfortunately Dolby B doesn't offer a complete solution either. This system
depends on accurate tracking of absolute signal levels (the reference level
has to be set) and precise matching of circuit parameters between the
encoding and decoding circuits. And IF everything works perfectly the amount
of noise reduction is still insufficient to reduce tape hiss completely as
you have seen. My subjective opinion is that the system you will be happiest
with depends on both the type of music you listen to, and on your sensitivity
to the side effects of both systems.
There is no free lunch! CD disk players cost about $800-$1000 and PCM digital
tape encoders cost about $1500-$2500. These are the only acceptable solutions
for me.
Greg (still pushing CD's)