| Title: | Welcome to the CD Notes Conference |
| Notice: | Welcome to COOKIE |
| Moderator: | COOKIE::ROLLOW |
| Created: | Mon Feb 17 1986 |
| Last Modified: | Fri Mar 03 1989 |
| Last Successful Update: | Fri Jun 06 1997 |
| Number of topics: | 1517 |
| Total number of notes: | 13349 |
There have been lively and detailed discussions on the demerits of the
digital sampling process used in the CD: phase shift introduced by the
sharp low-pass filters, inability to record very low-level or very high
frequency signals, and so on. Because of these shortcomings, some
purist audiophiles claim that the conventional phono reproduction system
is far more accurate than the CD.
What is interesting is that no one has yet to give an in-depth analysis
of the conventional phono to compare relative accuracy of the both media.
I dare to bring forward an argument that phono reproduction system may not
be as accurate as some claim. The argument is a simple one:
*****************************************************************
Any magnetic cartridge produces an output voltage which is
proportional to the stylus' velocity not the deviation. It
follows that the magnetic cartridge produces the electrical
signal which is a derivative (as in calculus) of the signal
on the disk's groove.
*****************************************************************
1) Since there is no corresponding integration (again, as in calculus)
during the disk mastering process, the output signal of the
magnetic cartridge does not, in general, resemble the signal
supplied by the master tape. For example, if the master tape
contains a triangular wave, the output from the magnetic cartridge
would be a square wave. Because most musical signals are made
of composite sinusoidal waves of different amplitude and frequency,
the output signal from the magnetic cartridge can SOUND like the
signal supplied by the master tape with RIAA equalization.
2) If a sine wave is recorded on the disk, the cartridge produces
a cosine wave as the output signal. It follows that the CARTRIDGE
ALWAYS PRODUCES A SIGNAL THAT IS 90 DEGREES SHIFTED REGARDLESS
OF FREQUENCY compared to the signal in the groove. What does this
mean? Well, I do not know how human brain processes the phase
information. But if it recognizes the cosine wave as a sine wave
which is advanced by 90 degrees, we have an interesting situation.
Suppose we have a composite of 2 single cycle sine waves, one
50 Hz and the other 5 KHz, recorded on a disk. When the signal
comes out of the cartridge, it will be 2 single cycle cosine waves.
If our brain recognizes the composite as two single cycle sine
waves, then the 50 Hz signal begins almost 5 ms earlier than the
5 KHz signal. Could this be true?
3) I wonder if a dynamic microphone and a magnetic tape head behave
in a similar manner. Is it possible that we are listening to
the third, or even fourth derivative of the original sound wave
in an analog recording. Does this explain differences between
the analog and digital sound?
Since I do not have extensive knowledge of the science of recording,
I will be happy if someone knowledgeable could shed some light on my points.
/Proud owner of SLP-7 with a bum drawer/
| T.R | Title | User | Personal Name | Date | Lines |
|---|---|---|---|---|---|
| 187.1 | ACADYA::PAHIGIAN | Tue Sep 24 1985 11:08 | 42 | ||
I'm not, by any means, a calculus expert, but I do know that filters of any kind introduce phase shift. Before joining Digital, I designed instrumentation for a cardiology lab; one of my jobs was to design a band-limited subsonic amplifier with a 96-dB-per-octave rolloff at 0.5 Hz to be used for observing heart sounds on a storage 'scope. I alerted the cardiologist to the fact that an analog design would cause the waveform to be an approximation at best of the true heart sound due to massive phase shifts approaching f(o). His response was that the approximation was satisfactory for the application. In short, it was imperfect, but it worked, and the customer was pleased with the results. It enabled him to make the correct diagnosis. Phono reproduction is subject to similar imperfections but to a lesser degree due to the gentle slope of the RIAA curve. The delayed sounds we put up with intrude less upon the listening experience than do multi-miking, oxidized connections in the studio, the recording console with God knows how many jellybean op amps and high-DA caps and crummy tone circuits, compression, eq, mixdown (we pass it through that console more than once!), overmodulation during the cutting of the master, and excessive use of a particular stamper during pressing. And by the way, when we master, we use Dolby A or dbx (more compression/expansion, more eq games, more lousy caps) along with tape eq (similar but not identical to RIAA). The points I'm trying to make are, first, that yes, we have problems we've got to fix in the recording process, but the time delays caused by the RIAA curve contribute little to recording's typical drawbacks in light of the condition of the signal chain from microphone to stamper, and second, that even if we were to clean up that signal chain, we could probably live with the inherent imperfections of the RIAA curve. It's imperfect, but it works. With a clean signal chain, the results can be quite pleasing. Considering the electronic and mechanical atrocities that occur between the actual performance and the time the pressing is made, I'd say the phase shifts caused by home RIAA pale in comparison. Just so I say SOMETHING about CD players, (this IS a CD notes file), viva oversampling and digital filtration. And let's use stiff power supplies and make room for Wonder Caps; as we're learning how to get accurate with digital, we shouldn't forget how we got accurate with analog. Craig | |||||
| 187.2 | GRAMPS::WCLARK | Tue Dec 17 1985 16:18 | 17 | ||
.0 How do you suppose info is placed on the master disk ? FM ? All cutterheads are velocity generators. Cartridges are Velocity sensitive. Nice match. Incidently a groove which appears triangular to the eye will produce a square wave at the output of the cartridge. Analog tape probably has more difficulty reproducing squarewaves than a decent phono cartridge. I stopped listening to sinewaves and squarewaves a long time ago, I much prefer music. Walt | |||||
| 187.3 | MIRACL::MATSUOKA | Thu Dec 19 1985 02:02 | 8 | ||
RE:2 Do they "pre-integrate(?)" the signal into the cutterhead with the -3dB per octave equalization to make it a velocity generator? Is it why they must pre-emphasize the high frequency signal (whose groove displacement would be tiny) to reduce the effect of record surface noise? Masamichi | |||||
| 187.4 | EAGLE1::KONG | Wed Dec 25 1985 22:11 | 22 | ||
RE:-.1 A simplistic way of looking at it: RIAA pre-emphasis differentiates the signal. The phono-cutter integrates the signal. The amplitude of the groove resembles that of the signal after the above two processes. A ceramic pickup is amplitude sensitive and therefore requires no further de-emphasis. A magnetic pickup (which is what most of us have) is velocity sensitive and outputs a signal that is the derivative of the amplitude of the groove and hence the derivative of the original signal. RIAA de-emphasis integrates the signal put out by the magnetic pickup. Therefore we get back the original signal. Hope this simple explanation helps. /tom | |||||
| 187.5 | GRAMPS::WCLARK | Fri Dec 27 1985 16:31 | 20 | ||
RIAA pre emphasis came about mainly due to limits of cutter heads. Historically
all cutter heads were inductive devices (an inductive electromechanical device
would move off in a give direction forever if a fixed DC level were applied).
Low frequency information of a given amplitude will cause a greater peak
displacement of the mechanical half of the device (the cutting stylus here)
than the same amplityde information at a higher frequency. To manage this
problem within the mechanical confines of a cutter assembly (and compliance
limits of a playback cartridge) a 6db/octave slope was specified from 500Hz
down. Since they were mucking about with filters anyway the RIAA folks decided
to do something about the hf signal-noise problem of records. The record
surface has a nasty habit of being noisey right where it is both most audible
and where musical information is of the lowest amplitude, the treble reigon,
so they decided to throw in a 6db/octave slope boost starting at 1Khz.
This also coincided pretty close to what a ceramic or strain gauge cartridge
would see as constant amplitude (since they are amplitude sensitive) so they
adopted a neat little package that allowed both velocity and amplitude devices
to reproduce from the same source.
Walt
| |||||
| 187.6 | EAGLE1::KONG | Fri Dec 27 1985 17:05 | 11 | ||
The RIAA time constants used to be 3180uS, 318uS, and 75uS. This corresponds to a 6db/Oct boost between 50Hz and 500Hz, flat between 500Hz and 2122Hz, and 6db/Oct boost from 2122Hz up. I think the low frequency emphasis has been changed a few years back. Anyone know what the current standard is? /tom | |||||
| 187.7 | GRAMPS::WCLARK | Mon Dec 30 1985 10:17 | 15 | ||
Re: .5, .6 I would like to correct an error in .5 as pointed out in .6...... The HF corner is 2Khz not 1Khz as I mentioned. I believe the 'standard' for the low end goes something like this. During record the 6db/octave roll off continues to DC. The playback LF boost is optional with the designer. The choices are to continue the boost down to the Preamp LF cutoff or insert a 6db corner at 50Hz (which effectively freezes the gain of the preamp below 50Hz. Some designers go a step further and add a 6 or 12db corner around 20Hz to help prevent system overload caused by record warp below 10Hz. Walt | |||||