Sorry in advance for very-very long post!
... I see (based on your explanation of the 3 strips @ freq/db amplification) where the high end might get distorted. 40db typical gain on the bugle + the extra 25db that hasn't been padded down via the RIAA curve. But I still dont "get" why the low end would have problems as well.
It's amazing - I think questions would be about high end, not low.
The amplification you are talking about is just
relative to the total power of the signal. The total power should be chosed in the way the peak of the signal does not exceed max input signal capability of the ADC. In that point of view the distribution of the power amongst bands is not important. So "25dB up in high freq." means only that the signal would have more high frequencies power and less middle and even more less - low frequencies. Taking into account that:
- the ADC (even the best) has finite analog part's noise; (then you will digitize the high frequencies with excellent signal-to-noise values, middles - worse than usual (part of middles in the total power is less as we agreed before), lows - the worst)
- the ADC (even best) has finite conversion noise; (then after the conversion the low-band conversion noise will rise according to the RIAA correction)
- the ADC (even best) has finite digitizing error; (that explains more distortions in the low-end: the less "power" fit the low band - the more the finite digitizing error affect the signal in the band)
- most ADCs has statistically optimised digital filter as part of the digitizing process; (that explains problems in high end - typically the amount of high frequencies in the signal is much lower then low, the filter is optimised for that situation, in our case it would be almost equal which adds the filter errors due to non-optimum setup)
- you cannot get more information (in digital) then you have after digitizing (second problem for low frequencies - it will use tiny number of bit-depth comparing to "normal" situation; the tiny bit-depth will have small amount of "information" after the digital amplification you'll save "rudeness" of low-informative lows).
My thoughts here on the process are: "Hey this makes sense... I should pursue the feasibility of it."
I've already written a lengthy answer, I'm afraid if I analyse the whole idea it takes several pages to read. First of all the idea is not new. It has been used mainly for restoration of old records in the low-signal end for decades and for high-signal end - more than 10 years in many receivers. Initially, the idea go back to multiband equalisers - at the time - analog. And it has the same problems as the initial idea: more distortions and phase variations.
To be more precise I copy the list of problems of "old approach" they give us:
•Analog EQ's produce Noise because of the op amps used.
•The left and right channels track poorly because of component variations in terms of tolerance.
•Aging has a substantial effect on capacitors, so what you hear today will be different than what you hear a year or two from now from an analog system.
•Analog circuits pick up some hum because of the physical loop areas which can not be avoided in the circuit layout.
•Analog circuits display crosstalk due to stray capacitance which can not be eliminated between the channels.
•Analog circuits are somewhat microphonic picking up low levels of room sound or feedback.
They use the same "op amps" in their RIAA-less devices. Most of the analog stages are the same in the "new" and "old" circuits... including caps. Pure analog circuits doesn't have problems with high digital frequencies which lay nearby the analog circuits, so pick up rather less junk than the combined ones. And you don't have to add ADC-DAC process which is not ideal at all!!!
•Analog EQ's have sloppy frequency and phase calibration because of resistor and capacitor tolerances.
ADC-DAC conversation is more vulnerable to several variations including jitter, variations in driving oscillator... Many studies says the amplitude variations is not so bad comparing to other type (distortions, jitter ...) inconsistencies.
•Analog components such as resistors, capacitors and transistors exhibit a temperature dependency referred to as tempco (temperature co-efficient). Therefore, the performance of an analog circuit (gain and break point frequencies) will change as a function of ambient temperature.
For such circuits you can use precise parts with low or stable pps. On the other hand increasing gain (because before ADC there is "temperature dependent analog circuits") in the conversion process to digital could lead to clipping easily.
•Physical capacitors exhibit such anomalies as DA (Dielectric Absorption), ESR (Effective Series Resistance) ESL (Effective Series Inductance), Voltage dependent capacitance or incremental capacitance (dC/dV) (which creates non-linear capacitance vs. signal level), and leakage resistance. All of these parasitics cause them to behave in a less than ideal manner when used in an analog circuit.
Scarecrows! Very-very big!
If I start to list all problems in ADC-DAC conversion they will occupy many-many pages!
