Jimmy The Clam wrote:
"Arny Krueger" wrote in message
...

I've got some concerns about the fact that the DMX's RIAA
equalization is done with software. Historically, this has involved
some compromises in terms of dynamic range. I can't find any
detailed specs or independent tests that show how this unusual
feature actually works.

Do you know if that dynamic range would be reduced to less that 80db?

At some frequencies in the audio range, probably yes.

There's 40 dB of equalization in the RIAA curve. It's basically a giant
roll-off that levels off for a while in the middle.

A RIAA preamp stage generally has about 36-40 dB gain in the midrange,
about 56-60 dB gain at 20 Hz, and 16-20 dB gain at 20 KHz.

If you implement this as a broadband amplifier with flat frequency response
followed by the equalization implemented in the digital domain, You need to
have 56-60 dB gain at all frequencies. IOW. to have the equivalent of 1 volt
of output at 20 KHz, you need to have close to 100 volts and attenuate it by
40 dB. The usual RIAA preamp designs attenuate the signal before it gets
that big.

Consumer sound cards generally have maximum input in the range of 1-2 volts.
If you attenuate that by 40 dB, you have 10-20 millivolts. So the output of
the flat response preamp at 20 Hz, is now limited to 10-20 millivolts, which
is a very small signal. If the card starts out with 95 dB SNR, then the
equivalent SNR at 20 Hz is 55 dB. The good news is that 20 Hz is a
frequency where the ear is insensitive and the turntable is producing a lot
of noise of its own.

Over the whole audio range, the SNR limit due to the flat response preamp
design might average out to be more like 75 dB in this implementation, which
is a somewhat better than you can get with a good turntable and a very clean
record. It's not that much worse than conventional high quality RIAA preamp
designs.

The plus is that the equalization can be implemented in the digital domain
and therefore be highly precise. Expensive high-tolerance capacitors for
equalization (usually 2 per channel) aren't needed.

I think that with the needle up and gains matched, you might be able to hear
a difference in the sound of the background noise, if you compared a flat
response phono preamp followed by a digital equalizer implemented in the
computer, as compared to a standard RIAA preamp. It would be an interesting
experiment. Needle on the record, it could be a horse race.

People including myself have thought about implementing a preamp this way
for years. The plus is that it eliminates the cost of some expensive
precision capacitors and some modest-priced but not cheap precision
resistors.

I know the card has a greater dynamic range than vinyl, but are you
saying that what would come off the record would be reduced further
or just that the full range that the card has to offer would be
reduced?

I just don't know the actual details of how they do what they provide a
summary description for. I'm presuming a solution here that I think is a
logical surmise. It's an interesting approach that a lot of people have
dismissed over the years. Sound cards are getting good enough (close to 100
dB dynamic range) that it might work. Clearly, this approach would have
been a disaster in the days when sound cards only had only 65 dB dynamic
range, or marginal with current cheapo-cheapo sound cards that have more
like 80 dB dynamic range.