24-bit precision gives you about 16.77 million values. Assuming a total groove width of 50 x 10^-6m, the maximum movement of the cutter is physically bounded at about half that. Much more and the cutter will be in the space for an adjacent groove. Thus, 50 microns width divided by 16.77 million gives us about 3 x 10^-12m, i.e. ~0.03 angstroms.
The diameter of a hydrogen atom is 1.0 angstroms (1 x 10^-10m). That would make the resolution of a 24-bit digital signal equivalent to an analog cutter whose resolution is just about 1/30 the width of a hydrogen atom. Sadly, this seems to be physically impossible, as none of the particles smaller than atoms are stable enough to be used in records.
Of course, records aren't made of hydrogen, they're made of the polymer pvc. One molecule of pvc is about 100,000 angstroms. This means that, if the cutters were actually removing single pvc molecules the vinyl records would have about 11 bits of resolution. Sadly, they don't get even that precise, though I'm not sure the actual precision. To get down to a record made of hydrogen atoms (possible under very low temp/very high pressure I suppose) one would need 19 bits. Anything beyond that is useless as long as the laws of physics hold.
Therefore, all other things being equal, digital is superior to vinyl. That said, mastering on CDs is often terrible while the mastering on records is often made somewhat better. This varies from CD to CD and record to record, and CDs are technologically far superior to records.
Full Version: Vinyl vs Digital and 24 bit vs 16 bit from vinyl.
Dig that crazy math man! Speaking as a huge vinyl fan; you're absolutely right! As much as I really love the LPs, funky sleeve notes and great graphic cover art - I'll take CD every time if given a choice of which ‘sounds better’ – mastering / productions caveats aside.
Andrew D.
www.cdnav.com
Not to be too nit-picky, but vinyl records aren't cut directly but are molded from a metal master, so the limit would be the diameter of a metal atom, which is much smaller than 100,000. Also, if they were cut directly in vinyl, the cutter needn't remove whole molecules but would readily break bonds as needed to remove part of a molecule.
QUOTE(pdq @ Mar 5 2008, 12:11) 
Not to be too nit-picky, but vinyl records aren't cut directly but are molded from a metal master, so the limit would be the diameter of a metal atom, which is much smaller than 100,000. Also, if they were cut directly in vinyl, the cutter needn't remove whole molecules but would readily break bonds as needed to remove part of a molecule.
But the metal master has to have been created in some way....
Best first post in years.
QUOTE(Nick.C @ Mar 5 2008, 08:22)
QUOTE(pdq @ Mar 5 2008, 12:11)
Not to be too nit-picky, but vinyl records aren't cut directly but are molded from a metal master, so the limit would be the diameter of a metal atom, which is much smaller than 100,000. Also, if they were cut directly in vinyl, the cutter needn't remove whole molecules but would readily break bonds as needed to remove part of a molecule.
But the metal master has to have been created in some way....
Indeed, and that's where the diameter of a metal atom comes in.
But the contact area of the stylus covers many thousands of molecules, and PVC deforms elastically. So the actual traced waveform could have a resolution that is considerably more accurate than that implied by the size of a "single" PVC molecule, at the expense of some time resolution (which due to tracing distortion really isn't there anyway).
What exactly is a "single" PVC molecule, anyway? PVC is a chain polymer. It might be 100000A on its long axis, but each member of the chain is what, H3C2Cl? That can't be more than 50A. So even ignoring the contact area argument, you may be off by a few orders of magnitude on the size calculations.
I'm all for bashing vinyl, and I agree that you're unlikely to get more than 12 bits out, but I think this math is too sloppy to support that.
What exactly is a "single" PVC molecule, anyway? PVC is a chain polymer. It might be 100000A on its long axis, but each member of the chain is what, H3C2Cl? That can't be more than 50A. So even ignoring the contact area argument, you may be off by a few orders of magnitude on the size calculations.
I'm all for bashing vinyl, and I agree that you're unlikely to get more than 12 bits out, but I think this math is too sloppy to support that.
QUOTE(Axon @ Mar 5 2008, 13:39)
But the contact area of the stylus covers many thousands of molecules, and PVC deforms elastically. So the actual traced waveform could have a resolution that is considerably more accurate than that implied by the size of a "single" PVC molecule, at the expense of some time resolution (which due to tracing distortion really isn't there anyway).
What exactly is a "single" PVC molecule, anyway? PVC is a chain polymer. It might be 100000A on its long axis, but each member of the chain is what, H3C2Cl? That can't be more than 50A. So even ignoring the contact area argument, you may be off by a few orders of magnitude on the size calculations.
I'm all for bashing vinyl, and I agree that you're unlikely to get more than 12 bits out, but I think this math is too sloppy to support that.
PVC consists of alternating CH2 and CHCl units, so in theory when cutting through the vinyl molecule you could cut to this small a unit by severing the molecule at a C-C bond. In reality none of this matters because the cutting is actually done on a metal master and not the vinyl.
Edit: SOME masters are cut directly into metal, but most are cut into lacquer-coated metal.
QUOTE(pdq @ Mar 5 2008, 10:27)
QUOTE(Axon @ Mar 5 2008, 13:39)
But the contact area of the stylus covers many thousands of molecules, and PVC deforms elastically. So the actual traced waveform could have a resolution that is considerably more accurate than that implied by the size of a "single" PVC molecule, at the expense of some time resolution (which due to tracing distortion really isn't there anyway).
What exactly is a "single" PVC molecule, anyway? PVC is a chain polymer. It might be 100000A on its long axis, but each member of the chain is what, H3C2Cl? That can't be more than 50A. So even ignoring the contact area argument, you may be off by a few orders of magnitude on the size calculations.
I'm all for bashing vinyl, and I agree that you're unlikely to get more than 12 bits out, but I think this math is too sloppy to support that.
PVC consists of alternating CH2 and CHCl units, so in theory when cutting through the vinyl molecule you could cut to this small a unit by severing the molecule at a C-C bond. In reality none of this matters because the cutting is actually done on a metal master and not the vinyl.
And the vinyl is pressed, not cut.
The metal master is not cut, either, the lacquer is cut, then plated, and then the plated master is cast, and turned into a stamper, and then that presses the actual vinyl.
While I'm not a huge fan of LP either, the OP is actually too dismissive, but only a little. Left out is surface noise (it's a physics thing, you have to have it), elastic deformation, plastic deformation, mistracking, the effects of equalization ...
True, I simplified a lot. But, as I also showed, even if you could get it down to angstrom resolution you would still only have the equivalent of 19 bit audio. I ignored surface noise and such because it made the calculation easier, not because it was accurate. Since it makes things less accurate my calculations make it seem a bit better than it is.
The real point is that, physically, there is no such thing as analog. You can't (easily) move less than 1 atom, you can't detect changes that small with a needle (well, a STM can, but that's a very different needle) and thus digital truly can be equivalent to vinyl. Everything is quantized, so with enough bits of data stored anything can be accurately digitized up to the point where the uncertainty principle begins to matter.
That said, vinyl has big album art. NIN recently offered some good extras with their newest album, but I'm waiting for more artists to offer digital downloads + shipping you a poster/lyrics book/etc instead of trying to cram it into a CD jewel case.
The real point is that, physically, there is no such thing as analog. You can't (easily) move less than 1 atom, you can't detect changes that small with a needle (well, a STM can, but that's a very different needle) and thus digital truly can be equivalent to vinyl. Everything is quantized, so with enough bits of data stored anything can be accurately digitized up to the point where the uncertainty principle begins to matter.
That said, vinyl has big album art. NIN recently offered some good extras with their newest album, but I'm waiting for more artists to offer digital downloads + shipping you a poster/lyrics book/etc instead of trying to cram it into a CD jewel case.
QUOTE(SoAnIs @ Mar 5 2008, 20:13)
The real point is that, physically, there is no such thing as analog. You can't (easily) move less than 1 atom, you can't detect changes that small with a needle
To be correct, needle movement is analog. Medium is not, but if we think of tracking needle, not groove per se. So we're always end up with interpolation here when decreasing time delta to get to the "fraction" of molecula.
QUOTE(Vitecs @ Mar 6 2008, 07:46)
QUOTE(SoAnIs @ Mar 5 2008, 20:13)
The real point is that, physically, there is no such thing as analog. You can't (easily) move less than 1 atom, you can't detect changes that small with a needle
To be correct, needle movement is analog. Medium is not, but if we think of tracking needle, not groove per se. So we're always end up with interpolation here when decreasing time delta to get to the "fraction" of molecula.
Well OK, but that's no different in result to the reconstruction filter in a DAC; it's still using "quantised" information to create a smoothed out signal.
Hello, SoAnIs!
There's one point I'm missing here. The signal coded on a vinyl disc is time-continous and not time-discrete. That means that any "molecule error" spreads over a very large frequency range. The audible band is only a small subset of it. So, without the knowledge about the noise's PSD it's difficult to estimate the signal-to-noise ratio you'll get after filtering out everything above 20 kHz.
Still, it's a fun thought experiment.
Cheers,
SG
There's one point I'm missing here. The signal coded on a vinyl disc is time-continous and not time-discrete. That means that any "molecule error" spreads over a very large frequency range. The audible band is only a small subset of it. So, without the knowledge about the noise's PSD it's difficult to estimate the signal-to-noise ratio you'll get after filtering out everything above 20 kHz.
Still, it's a fun thought experiment.
Cheers,
SG
QUOTE(SoAnIs @ Mar 5 2008, 03:20)
24-bit precision gives you about 16.77 million values. Assuming a total groove width of 50 x 10^-6m, the maximum movement of the cutter is physically bounded at about half that. Much more and the cutter will be in the space for an adjacent groove. Thus, 50 microns width divided by 16.77 million gives us about 3 x 10^-12m, i.e. ~0.03 angstroms.
The diameter of a hydrogen atom is 1.0 angstroms (1 x 10^-10m). That would make the resolution of a 24-bit digital signal equivalent to an analog cutter whose resolution is just about 1/30 the width of a hydrogen atom. Sadly, this seems to be physically impossible, as none of the particles smaller than atoms are stable enough to be used in records.
Of course, records aren't made of hydrogen, they're made of the polymer pvc. One molecule of pvc is about 100,000 angstroms. This means that, if the cutters were actually removing single pvc molecules the vinyl records would have about 11 bits of resolution. Sadly, they don't get even that precise, though I'm not sure the actual precision. To get down to a record made of hydrogen atoms (possible under very low temp/very high pressure I suppose) one would need 19 bits. Anything beyond that is useless as long as the laws of physics hold.
Therefore, all other things being equal, digital is superior to vinyl. That said, mastering on CDs is often terrible while the mastering on records is often made somewhat better. This varies from CD to CD and record to record, and CDs are technologically far superior to records.
As far as I know there are an an infinite amount of numbers between the number 1 and the number 2. But I guess you could measure the amount of pixels in film if you measured each ray of light.
If you are recording digitally the the bit depth would measure how high or low the wave is at any given sample. 16 bit, if I am not mistake, only has 9 values in either direction. What I don't understand is why people think 9 is good enough to say any more is a waste.
QUOTE(A Dawg @ Jun 2 2008, 23:58)
As far as I know there are an an infinite amount of numbers between the number 1 and the number 2. But I guess you could measure the amount of pixels in film if you measured each ray of light.
Depends on the set of numbers you're operating with.
QUOTE(A Dawg @ Jun 2 2008, 23:58)
If you are recording digitally the the bit depth would measure how high or low the wave is at any given sample. 16 bit, if I am not mistake, only has 9 values in either direction. What I don't understand is why people think 9 is good enough to say any more is a waste.
2^16 = 65536, so yes, you are mistaken. 18 total values would be just past 4 bits, which is obviously insufficient.
I personally feel that 16 to 24 bit is a small, but noticeable step up. I have a soundcard which lets me change the bit depth while music is playing. I have played around with it and I would say that it was about as noticeable as going from 720p to 1080p on a 35 inch tv. Few would notice. But at close glance, you can see the extra detail. But when you go from 44.1k to 48k BIG difference. Most noticeable when a quiet passage is played at a high volume. CD's suffer a digital hiss, which are basically the steps in between samples. DVDs sound clearer than cd's to me because of these extra "steps." More steps in the same amount of space = smaller step. But the sound is of a lower quality than a cd, if that even makes sense. I am not very good at expressing my thoughts through words sometimes :'(. And I was joking about the lower.
Here is my stance. I am a guy who still buys records and loves to flac em 96k times a second with 24 bits in them samples. You could have the normal 1(16 bit) lock on your doorknob. But I have a deadbolt, locking knob, and a chain.(24 bit) And instead of a wooden or plastic (44.1 khtz)door, I have a steel door.
Am I actually safer(sound better) than you just because of that, maybe. But I definitely feel safer.
Was that too dumb
Here is my stance. I am a guy who still buys records and loves to flac em 96k times a second with 24 bits in them samples. You could have the normal 1(16 bit) lock on your doorknob. But I have a deadbolt, locking knob, and a chain.(24 bit) And instead of a wooden or plastic (44.1 khtz)door, I have a steel door.
Am I actually safer(sound better) than you just because of that, maybe. But I definitely feel safer.
Was that too dumb
I would argue that a closer analogy than a wood door vs. a steel door would be going vegetarian because you're afraid of food poisoning. Just because you feel safer about it doesn't mean you really are safer, or that the risk was all that important in the first place.
You should read up more on sampling theory. There is no "space" between samples.
DVDs are usually lossily encoded on their audio tracks, but their DACs are generally of the same garden variety that are used in computers. There's no difference in how they decode audio - merely in the choice of lossy encoder used.
That said, I still rip vinyl to 24/96, but I make no justification for it nowadays, and I'm running low enough on disk space that I am considering switching to 16/44 for good.
You should read up more on sampling theory. There is no "space" between samples.
DVDs are usually lossily encoded on their audio tracks, but their DACs are generally of the same garden variety that are used in computers. There's no difference in how they decode audio - merely in the choice of lossy encoder used.
That said, I still rip vinyl to 24/96, but I make no justification for it nowadays, and I'm running low enough on disk space that I am considering switching to 16/44 for good.
QUOTE(A Dawg @ Jun 3 2008, 06:41)
I personally feel that 16 to 24 bit is a small, but noticeable step up. I have a soundcard which lets me change the bit depth while music is playing. I have played around with it and I would say that it was about as noticeable as going from 720p to 1080p on a 35 inch tv.
Whereas, with all other things being equal, in a correctly controlled blind test, passing an ABX test of 16 vx 24 bits, or 44.1kHZ vx 48kHz is near impossible, except with extreme samples and/or faulty equipment.
Cheers,
David.
QUOTE(Axon @ Jun 3 2008, 09:34)
That said, I still rip vinyl to 24/96, but I make no justification for it nowadays, and I'm running low enough on disk space that I am considering switching to 16/44 for good.
I have always felt that 16 bits reproduces the vinyl's surface noise quite well so 24 bits would be overkill, at least for final playback. For recording 24 bits is better if any manipulation will be done to have more data to work with.
Some calculations for you and a proof that the LP has a resolution of 32000 bits!
http://www.st-andrews.ac.uk/~www_pa/Scots_...rt12/page2.html
http://www.st-andrews.ac.uk/~www_pa/Scots_...rt12/page2.html
From the article...
Cheers, Slipstreem.
QUOTE
The effect is to divide the microns swing of a 0 dB 1 kHz sinewave into 32,000 steps...
Steps, not bits. That's slightly less than 15 bits. That equates to 1600 steps (less than 11 bits) at 20kHz. Cheers, Slipstreem.
QUOTE(SoAnIs @ Mar 5 2008, 03:20)
Therefore, all other things being equal, digital is superior to vinyl. That said, mastering on CDs is often terrible while the mastering on records is often made somewhat better. This varies from CD to CD and record to record, and CDs are technologically far superior to records.
Sure, in theory. Thankfully, music reproduction depends on many factors, of which bit depth plays only a small part. From my perspective it seems clear that the analog versus digital methodology argument is logically moot. People are simply comparing apples to oranges. Personally, I believe that analog music reproduction has its place, as does digital music reproduction. These two distinct methods are clearly not in any way equal, and personally I believe, not even comparable. Why people waste their time going on about which is better is simply beyond me!
Here, let me try using an example of what I'm trying to get at; which is better, an apple or an orange? Boy doesn't that answer sound obvious! Now, lets reword it; which is better, analog or digital? See, just as silly, in my humble opinion of course.
But a digital recording can sound like anything you want - including an exact and perfect reproduction of the sound of an analogue (vinyl) recording, including all the noise, distortion, clicks, pops etc.
A vinyl recording will only sound like digital when the content is such that all the faults are masked. In other words, you can listen to a digital recording and think you're listening to vinyl; but you can't listen to vinyl and think you're listening to digital.
Therefore digital is superior as a delivery format because it's doesn't impose its own character on the audio. "Special effects" should be added because people want to, not because they're part of the delivery format.
Cheers,
David.
A vinyl recording will only sound like digital when the content is such that all the faults are masked. In other words, you can listen to a digital recording and think you're listening to vinyl; but you can't listen to vinyl and think you're listening to digital.
Therefore digital is superior as a delivery format because it's doesn't impose its own character on the audio. "Special effects" should be added because people want to, not because they're part of the delivery format.
Cheers,
David.
QUOTE(tot @ Jun 3 2008, 14:15)
QUOTE(Axon @ Jun 3 2008, 09:34)
That said, I still rip vinyl to 24/96, but I make no justification for it nowadays, and I'm running low enough on disk space that I am considering switching to 16/44 for good.
I have always felt that 16 bits reproduces the vinyl's surface noise quite well so 24 bits would be overkill, at least for final playback. For recording 24 bits is better if any manipulation will be done to have more data to work with.
I'm using a 36dB/oct HP filter for to get these noises off from vinyl recordings. Actually I even use a software based RIAA filter so that's why 24-bit is my choice.
Juha
Oh, nice! Another flat transfer partisan. Welcome to the club.
QUOTE(Axon @ Jun 3 2008, 12:20)
Oh, nice! Another flat transfer partisan. Welcome to the club.
I have to wonder, what does all that RIAA curve do to your overall gain structure and dynamic range, then.
The gain structure and DR are completely f*cked up, but the RIAA curve has nothing to do with that. It's entirely because the sound card I chose (an E-MU 0404 USB) appears to have an inferior dynamic range at 60db gain compared to its competitors. (I'm told that other integrated preamp/ADCs have no problem punching 80db SNR at that gain.)
Even then, the sound card noise floor is 10db lower than the noise floor of a quiet groove, at all frequencies. It still sounds pretty good.
Are you aware of Rob Robinson's AES preprint from the 123rd convention? ("Filter Reconstruction and Program Material Characteristics Mitigating Word Length Loss in Digital Signal Processing-Based Compensation Curves used for Playback of Analog Recordings").
Even then, the sound card noise floor is 10db lower than the noise floor of a quiet groove, at all frequencies. It still sounds pretty good.
Are you aware of Rob Robinson's AES preprint from the 123rd convention? ("Filter Reconstruction and Program Material Characteristics Mitigating Word Length Loss in Digital Signal Processing-Based Compensation Curves used for Playback of Analog Recordings").
QUOTE(Woodinville @ Jun 3 2008, 22:26)
QUOTE(Axon @ Jun 3 2008, 12:20)
Oh, nice! Another flat transfer partisan. Welcome to the club.
I have to wonder, what does all that RIAA curve do to your overall gain structure and dynamic range, then.
AFAIK, the filter does nothing more than what a hardware RIAA filter does since ...
- overal gain can be controlled by adjusting the gain coefficients (mine is set as no extra gain)
- I've never measured if there is something strange with the DR
... ABX between hardware RIAA and this method is not necessary here since the result is ~different (this depends on how good RIAA stage you have for your cartridge ... is it optimized by the cartridge specs, etc.)
... IMO, you need to hear the difference to tell which one is more natural sounding (I like the result I get a lot since low frequency area is much deeper/sharper (there's a sound there) and mid/high areas are clearer w/ lots of details I can't get out of the hardware path of my stereo system ... but as said, it's the hardware altogether involved in this)
These software filters I have programmed (by calculations mentioned in quote below) are just a bit more accurate in reproduction of the RIAA EQ (accuracy depends on selected sample rate ... I normally use 3th-4th order IIR filters so the maximum error @ 0Hz to 20kHz is calculated between ~0.0006dB (44.1kHz - ~0.000004dB (96kHz) and the phase is OK.
Here's the procedure how these filters been calculated by Robert Orban:
QUOTE
>Robert Orban wrote:
>
>> I can't see the first part of this thread, but if you are trying
>> to do an IIR simulation of the RIAA phono de-emphasis curve
>> (assuming s-plane poles at 50.5 and 2122 Hz and an s-plane zero
>> at 5005. Hz), here are some good minimum-phase magnitude
>> approximations.
>
>Neat, thanks! How did you make them?
I used a program I wrote (in ye olde Fortran :-). The outline goes as
follows:
Given a desired magnitude response in the z-plane, there exists a
response in a frequency-warped u-plane that, when bilinear-transformed
to the z-plane, creates the desired z-plane magnitude response.
-Compute the [magnitude response]^2 of the s-plane prototype on a grid.
This is the square of the desired z-plane response.
-Warp the frequency axis by using the bilinear transform, recognizing
that we are approximating using omega^2 as our frequency variable. The
warp maps Nyquist to infinity.
-Make a least-squares rational approximation (i.e., ratio of
polynomials) to the values on the frequency grid. (I used the Numerical
Recipes routine RATLSQ, which uses Chebychev polynomials.)
-Refine the approximation to make the fractional error minimax by using
Remez's Second Algorithm [which applies to rational functions; it's not
the same as the Remez algorithm used in the classical MPR FIR design
program; see Forman S. Acton, Numerical Methods That Work (Revised
Edition), Washington D.C., American Mathematical Society, 1990, pp 310-
314]
-Transform the result into the z-plane in two steps. The first
recognizes that we have been approximating using the magnitude square
function, so we must take the square roots of the poles and zeros of the
approximated rational function, taking the negative real parts to
guarantee a stable, minimum-phase function. The second step is to apply
the bilinear transform to the result of the first step. This yields the
final z-plane poles and zeros.
There are some "interesting" numerical issues in making this procedure
work, mainly because the Remez update formulas require solving a system
of mildly nonlinear equations that tend be ill-conditioned.
The nice thing about the algorithm is that the frequency-warping moves
Nyquist to infinity and thus increases the resolution of the
approximation close to Nyquist, which is where difficulties often occur.
Robert Orban
>
>> I can't see the first part of this thread, but if you are trying
>> to do an IIR simulation of the RIAA phono de-emphasis curve
>> (assuming s-plane poles at 50.5 and 2122 Hz and an s-plane zero
>> at 5005. Hz), here are some good minimum-phase magnitude
>> approximations.
>
>Neat, thanks! How did you make them?
I used a program I wrote (in ye olde Fortran :-). The outline goes as
follows:
Given a desired magnitude response in the z-plane, there exists a
response in a frequency-warped u-plane that, when bilinear-transformed
to the z-plane, creates the desired z-plane magnitude response.
-Compute the [magnitude response]^2 of the s-plane prototype on a grid.
This is the square of the desired z-plane response.
-Warp the frequency axis by using the bilinear transform, recognizing
that we are approximating using omega^2 as our frequency variable. The
warp maps Nyquist to infinity.
-Make a least-squares rational approximation (i.e., ratio of
polynomials) to the values on the frequency grid. (I used the Numerical
Recipes routine RATLSQ, which uses Chebychev polynomials.)
-Refine the approximation to make the fractional error minimax by using
Remez's Second Algorithm [which applies to rational functions; it's not
the same as the Remez algorithm used in the classical MPR FIR design
program; see Forman S. Acton, Numerical Methods That Work (Revised
Edition), Washington D.C., American Mathematical Society, 1990, pp 310-
314]
-Transform the result into the z-plane in two steps. The first
recognizes that we have been approximating using the magnitude square
function, so we must take the square roots of the poles and zeros of the
approximated rational function, taking the negative real parts to
guarantee a stable, minimum-phase function. The second step is to apply
the bilinear transform to the result of the first step. This yields the
final z-plane poles and zeros.
There are some "interesting" numerical issues in making this procedure
work, mainly because the Remez update formulas require solving a system
of mildly nonlinear equations that tend be ill-conditioned.
The nice thing about the algorithm is that the frequency-warping moves
Nyquist to infinity and thus increases the resolution of the
approximation close to Nyquist, which is where difficulties often occur.
Robert Orban
The Emu specs say the preamp noise is approaching the minimum possible, which is quite good at that price, even without consideration of the rest of the box's contents. You find that its noise is unusually high near full gain? Is this full spectrum noise? Do RMAA tests tell you anything interesting?
QUOTE(Axon @ Jun 3 2008, 22:40)
The gain structure and DR are completely f*cked up, but the RIAA curve has nothing to do with that. It's entirely because the sound card I chose (an E-MU 0404 USB) appears to have an inferior dynamic range at 60db gain compared to its competitors. (I'm told that other integrated preamp/ADCs have no problem punching 80db SNR at that gain.)
Even then, the sound card noise floor is 10db lower than the noise floor of a quiet groove, at all frequencies. It still sounds pretty good.
I'm using the same E-MU 0404 USB which don't necessarily need another pre-amp for input ... it gives good enough quality for playback purposes (in this case I gain the filter 15-20dB and the RIAA EQ is done realtime) but, I also have prepared matched pre-amp too for recording purpose mainly (I'm using Technics turntable w/ Technics 205CMK3 cartridges ... which is quite capable cartridge by the factory measures they include into bundle).
Juha
Are you using a moving coil cartridge? Normal gains for MM is around 35dB, if I remember correctly. Or is there some reason you need more gain with MM using that approach (quite a bit more gain!)?
QUOTE(AndyH-ha @ Jun 3 2008, 23:18)
Are you using a moving coil cartridge? Normal gains for MM is around 35dB, if I remember correctly. Or is there some reason you need more gain with MM using that approach (quite a bit more gain!)?
If this was for me then, I'm using MM cartridges.
Juha
No, Axon is the one turning the gain to maximum.
QUOTE(Axon @ Jun 3 2008, 12:40)
The gain structure and DR are completely f*cked up, but the RIAA curve has nothing to do with that. It's entirely because the sound card I chose (an E-MU 0404 USB) appears to have an inferior dynamic range at 60db gain compared to its competitors. (I'm told that other integrated preamp/ADCs have no problem punching 80db SNR at that gain.)
Even then, the sound card noise floor is 10db lower than the noise floor of a quiet groove, at all frequencies. It still sounds pretty good.
Are you aware of Rob Robinson's AES preprint from the 123rd convention? ("Filter Reconstruction and Program Material Characteristics Mitigating Word Length Loss in Digital Signal Processing-Based Compensation Curves used for Playback of Analog Recordings").
Yes, it's a simple exercise in filtering.
You don't see the 20dB in slope causing you any overload problems, etc. Oh, say, with Dark Side of the Moon? (JJ squints at a spectrum he just measured and wonders...)
It's entirely a matter of the gain used. I can clip if I actually set the gain to 65db and then play, say, the spot frequencies on STR151. (Note that this is a LOMC, an AT-OC9, I'm talking about.) But it's fine at 55db, I can always turn it down further if it overloads there, and I don't think I'll see many 20khz tones at those kinds of levels anyway... At 55db, normal classical LP levels hover at around 13-18db, and pops/ticks peak at -3. (Since I've never encountered that relatively high level of pop with my 440ML, I guess you could take that as some sort of empirical evidence for an MC having a tighter phase response than an MM. The transient when the stylus lands on the record is razor-sharp.)
I haven't recorded DSOTM yet (and well, my pressing is a well-work original Capitol, so it's not like it has much treble to begin with). But I did record Bernstein's Carmen, and while the opening cymbals are piercingly loud when recorded without RIAA, they are notably distortion-free. (It's worth noting that right now I'm recording at 24/44 and I'm using a naive bilinear IIR implementation of the RIAA filter, but once I actually implement the filter correctly, and perhaps record at 24/96, I'm not anticipating any surpises.)
It's also worth noting that the noise profile of both the sound card and the vinyl background hiss when flat-transferred (that I've measured so far) is white.
I haven't recorded DSOTM yet (and well, my pressing is a well-work original Capitol, so it's not like it has much treble to begin with). But I did record Bernstein's Carmen, and while the opening cymbals are piercingly loud when recorded without RIAA, they are notably distortion-free. (It's worth noting that right now I'm recording at 24/44 and I'm using a naive bilinear IIR implementation of the RIAA filter, but once I actually implement the filter correctly, and perhaps record at 24/96, I'm not anticipating any surpises.)
It's also worth noting that the noise profile of both the sound card and the vinyl background hiss when flat-transferred (that I've measured so far) is white.
QUOTE(Juha @ Jun 3 2008, 15:35)
QUOTE(AndyH-ha @ Jun 3 2008, 23:18)
Are you using a moving coil cartridge? Normal gains for MM is around 35dB, if I remember correctly. Or is there some reason you need more gain with MM using that approach (quite a bit more gain!)?
Yeah, like I said above, I'm running MC. The OC9 has a sensitivity of around 0.4mV IIRC. Obviously, if I was still running with my 440 it would all be running like gangbusters, but I was running into a lot of EM interference issues running balanced into the 0404, and going fully balanced on the 440ML requires physically cutting the ground tab that's hard wired to one of the signal wires, and I was really enamoured of the thought of not requiring any loading, just a straight shot to the XLR inputs... I'm having to rethink that now.
QUOTE(AndyH-ha @ Jun 3 2008, 15:11)
The Emu specs say the preamp noise is approaching the minimum possible, which is quite good at that price, even without consideration of the rest of the box's contents. You find that its noise is unusually high near full gain? Is this full spectrum noise? Do RMAA tests tell you anything interesting?
Creative Labs is full of shit. What else is new?The 0404 USB's specs are awesome on paper, as are the RMAA results. The problem is that the EIN/SNR numbers, while accurate, only apply for 0db of gain applied. As soon as you apply gain, you lose that EIN/SNR. So while you do in fact have that 105+db SNR with no gain, at 65db, you're literally down to 60db.
That is not how it's supposed to work. The Focusrite Saffire has an EIN of 120db at 60db of gain. Rob has tested the Saffire and it runs like a champ wired up to a LOMC, with no noise issues.
BTW Juha, that's really good info on using Remez to optimize IIR filter design. I've been having a lot of trouble coming up with high-accuracy RIAA coefficients.
Of course, is it possible for you to just post the coefficients for 44/48/96/192khz?
QUOTE(Axon @ Jun 4 2008, 00:35)
BTW Juha, that's really good info on using Remez to optimize IIR filter design. I've been having a lot of trouble coming up with high-accuracy RIAA coefficients.
Of course, is it possible for you to just post the coefficients for 44/48/96/192khz?
http://www.dsprelated.com/showmessage/73300/3.php
EDIT:
BTW, which one would be the better place to put the (subsonic) HP Filter in signal path ... before or after RIAA Filter?
Juha
EDIT2:
BTW, Axon, would it be too much to ask from you if you mail me or link here a short sample (~0.5-1 min) of some of your flat recorded vinyls you feel is a good for testing this software RIAA filter in action. As I'm using VST based technology, I could render the sample w/ filter added and send it back (or link here) so you (or anyone) could make some comparisons too. Sample file bit-resolution could be any and the sample-rate 44.1/48/88.2 or 96kHz but, the format should be better to be lossless (FLAC, WAV).
Juha
QUOTE(2Bdecided @ Jun 3 2008, 11:04)
Therefore digital is superior as a delivery format because it's doesn't impose its own character on the audio. "Special effects" should be added because people want to, not because they're part of the delivery format.
Sure, but what if the music was originally an analogue recording? Are you suggesting it all should be digitally remastered? Personally, I'd try to preserve the original sound by not messing with it. By the way, imho, there's more than enough music for both analog and digital reproduction, just in case you were worried about missing out on something. QUOTE(2tec @ Jun 7 2008, 22:31)
QUOTE(2Bdecided @ Jun 3 2008, 11:04)
Therefore digital is superior as a delivery format because it's doesn't impose its own character on the audio. "Special effects" should be added because people want to, not because they're part of the delivery format.
Sure, but what if the music was originally an analogue recording? Are you suggesting it all should be digitally remastered?Actually, even if they studio _is_ willing to send me the original master tape, I'd rather they didn't. I'd feel quite guilty wearing out all those Beatles master tapes just for my own private listening pleasure. Much better to make a digital copy that can be duplicated without loss as many times as necessary.
Of course I'm asking for a _good_ digital "remaster" - a careful transcription of the original - none of this dynamic range compression smashing against 0dB FS, and none of these terrible "drag a third generation dub off the shelf, play it on any old machine, ignore the EQ that was supposed to be applied at playback, and digitise it" efforts. But yes, digital every time. I even like noise reduction, and digital re-mixing to avoid a tape generation loss inerrant in an analogue master - though both are difficult to do in a way which pleases everyone, and good "modern" analogue mixing and mastering is already more than good enough.
People get very upset with the use of digital remastering to create vinyl. I mentioned the Beatles - many of their CDs and "remastered" LPs sound nothing like the original LPs, but "digital" isn't to blame - I can copy the original LPs onto CD just fine!
Cheers,
David.
QUOTE(Juha @ Jun 3 2008, 23:10)
QUOTE(Axon @ Jun 4 2008, 00:35)
BTW Juha, that's really good info on using Remez to optimize IIR filter design. I've been having a lot of trouble coming up with high-accuracy RIAA coefficients.
Of course, is it possible for you to just post the coefficients for 44/48/96/192khz?
http://www.dsprelated.com/showmessage/73300/3.php
QUOTE
BTW, which one would be the better place to put the (subsonic) HP Filter in signal path ... before or after RIAA Filter?
Well, if it's in the software domain, does it really matter? As long as you're working in floating point, it shouldn't. QUOTE
BTW, Axon, would it be too much to ask from you if you mail me or link here a short sample (~0.5-1 min) of some of your flat recorded vinyls you feel is a good for testing this software RIAA filter in action. As I'm using VST based technology, I could render the sample w/ filter added and send it back (or link here) so you (or anyone) could make some comparisons too. Sample file bit-resolution could be any and the sample-rate 44.1/48/88.2 or 96kHz but, the format should be better to be lossless (FLAC, WAV).
Yeah, I guess I could put something together. I'm out of web space at the moment, though (my old provider disappeared in the middle of the night - I'm not even kidding). I could probably email a small FLAC or send it through gmail?QUOTE(2Bdecided @ Jun 9 2008, 06:03)
Actually, even if they studio _is_ willing to send me the original master tape, I'd rather they didn't. I'd feel quite guilty wearing out all those Beatles master tapes just for my own private listening pleasure. Much better to make a digital copy that can be duplicated without loss as many times as necessary.
Heh, heh, heh.
QUOTE(2Bdecided @ Jun 9 2008, 05:03)
QUOTE(2tec @ Jun 7 2008, 22:31)
QUOTE(2Bdecided @ Jun 3 2008, 11:04)
Therefore digital is superior as a delivery format because it's doesn't impose its own character on the audio. "Special effects" should be added because people want to, not because they're part of the delivery format.
Sure, but what if the music was originally an analogue recording? Are you suggesting it all should be digitally remastered?This is a "lo-fi" version of our main content. To view the full version with more information, formatting and images, please click here.