This whole conversation has me intrigued about the technical limitations of vinyl. For example, is there a minimum or maximum frequency that can be "encoded" in vinyl (akin to the ~20Hz and 44.1kHz cutoffs on digital CDs)? Does this differ for 33s, 45s, 78s? Is there a way (based on the needle tip and equipment setup) for a knowledgeable person to precalculate the bands that the most noise interference will likely be in, and thus minimize or remove it? Etc.
It's particularly pertinent as I'm planning to attempt a vinyl-to-FLAC encoding soon.
You might want to review the vinyl myths page (http://wiki.hydrogenaudio.org/index.php?title=Myths_(Vinyl)) in the wiki, especially where it talks about frequency limits.
You will find that different records have a different-sounding "whoosh" of noise imparted by the rush of the groove vinyl along the stylus, so I can't imagine any way to predict that noise exactly. Speed is one of many factors in that.
You might want to review the vinyl myths page (http://wiki.hydrogenaudio.org/index.php?title=Myths_(Vinyl)) in the wiki, especially where it talks about frequency limits.
You will find that different records have a different-sounding "whoosh" of noise imparted by the rush of the groove vinyl along the stylus, so I can't imagine any way to predict that noise exactly. Speed is one of many factors in that.
Ah! That makes for an interesting read, and does indeed answer many of my questions. Thanks for pointing it out.
High frequency vinyl noise is strongly broadband, so there's little use shifting that around, per se.
The arm+cartridge resonance defines the low frequency limit. If you're lucky, it will be 8-12hz; if unlucky, 5-20hz.
The HF limit is dependent on many factors, but at the very least, CD4 records go all the way to 45khz. Typically, you'll know if you have a CD4 record. Otherwise, the useful frequency range limit might be as low as 15khz, or as high as 20-30khz.
Hypothetically, on arm+cart combinations which are too low (<8hz), it might make sense to e.g. play a 33rpm record at a much faster speed, so as to shift the signal spectrum away from the resonance, because spot frequency SNR is ludicrously low around resonance. But AFAIK, that's probably more trouble than it's worth.
The arm+cartridge resonance defines the low frequency limit. If you're lucky, it will be 8-12hz; if unlucky, 5-20hz.
Very few records are cut with any signal lower then about 20-30Hz. Also, all LPs will have their bass summed to mono below about 100Hz to prevent extreme vertical groove modulation.
The thing about the resonant frequency is that you want it to be below the lowest signal frequency on an LP (ie. less than 20Hz) to prevent exciting the resonance. But you also want to keep it above the frequency of the micro-warps that exist on even the seemingly flattest LP (ie. above about 10Hz). This means that the arm/cartridge will ride up & down over warps - if the resonant frequency is too low then the warps will be read as very low frequency signal.
The HF limit is dependent on many factors, but at the very least, CD4 records go all the way to 45khz. Typically, you'll know if you have a CD4 record. Otherwise, the useful frequency range limit might be as low as 15khz, or as high as 20-30khz.
The ultra high frequencies on a CD4 record can only be read by specialised stylus profiles (eg. Shibata). Play them with anything else (eg. a standard elliptical or (heaven forbid) a conical), and the carrier frequency is destroyed. But CD4 is a historical footnote - nobody actually uses it these days (do they?).
Mainstream LPs are often cut with a low pass filter around the 18kHz mark to prevent overheating of the cutting head. Any kind of signal that comes off an LP above 20kHz is pretty much entirely noise and distortion.
Hypothetically, on arm+cart combinations which are too low (<8hz), it might make sense to e.g. play a 33rpm record at a much faster speed, so as to shift the signal spectrum away from the resonance, because spot frequency SNR is ludicrously low around resonance. But AFAIK, that's probably more trouble than it's worth.
I don't see how this would work. Play a 33rpm record at 45rpm and you'll shift the warps higher up the frequency range, and therefore still above the resonant frequency. So they will still be incorrectly read as signal. If anything, the way to avoid this would be to reduce the speed to 16rpm - but then the bass signal is in danger of dropping low enough to excite the resonance.
If your arm/cartridge combination has a resonant frequency in the wrong place, you have these choices:
1. If it's too high, you can add mass to the arm - stick a penny on the headshell. This will drop the resonant frequency. (Don't forget to reset the tracking force correctly. This is NOT the same as the old trick we did on Dansettes to stop them skipping!)
2. If it's too high, replace the cartridge with a lower compliance model.
Unlike CD, vinyl has no inherent high frequency limit, in theory.
In regards to frequency response, FR, don't worry about the overall bandwidth limits, worry about the number of dB that the overall playback system (LP record condition, phono cartridge, phono preamp) deviate from being perfectly flat in the audible region. That's what matters in terms of audibility.
The ultra high frequencies on a CD4 record can only be read by specialised stylus profiles (eg. Shibata). Play them with anything else (eg. a standard elliptical or (heaven forbid) a conical), and the carrier frequency is destroyed. But CD4 is a historical footnote - nobody actually uses it these days (do they?).
Mainstream LPs are often cut with a low pass filter around the 18kHz mark to prevent overheating of the cutting head. Any kind of signal that comes off an LP above 20kHz is pretty much entirely noise and distortion.
Just out of curiosity, didn't the Beatles record an ultrasonic signal at the end of
Sgt Peppers's -- supposed to make your dog's ears perk up? And if so, has anyone ever checked a needledrop to see how 'high fidelity' that tone was on playback?
The ultra high frequencies on a CD4 record can only be read by specialised stylus profiles (eg. Shibata). Play them with anything else (eg. a standard elliptical or (heaven forbid) a conical), and the carrier frequency is destroyed. But CD4 is a historical footnote - nobody actually uses it these days (do they?).
Mainstream LPs are often cut with a low pass filter around the 18kHz mark to prevent overheating of the cutting head. Any kind of signal that comes off an LP above 20kHz is pretty much entirely noise and distortion.
Just out of curiosity, didn't the Beatles record an ultrasonic signal at the end of Sgt Peppers's -- supposed to make your dog's ears perk up? And if so, has anyone ever checked a needledrop to see how 'high fidelity' that tone was on playback?
The frequency which appears on the British lead out groove of Sgt. Pepper, just before "Lucy abbey all the way" [at least that's what I hear] is very high frequency but
not ultrasonic. It is perhaps 15 kHz, I'm guessing.
I was close. Geourge Martin says it was 18 kHz.
http://www.youtube.com/watch?v=0WkUgNKOtSE (http://www.youtube.com/watch?v=0WkUgNKOtSE)
EDIT: Wikipedia says "15 kilohertz"
http://en.wikipedia.org/wiki/Sgt._Pepper&#...earts_Club_Band (http://en.wikipedia.org/wiki/Sgt._Pepper's_Lonely_Hearts_Club_Band)
should be able to verify off the CD then, yes?
Here's a 'Blue box' needle-drop:
(http://i50.tinypic.com/otn3wx.png)
Unlike CD, vinyl has no inherent high frequency limit, in theory.
This is wrong, i believe. Vinyl has inherent theoretical high and low end frequency limits due to the material medium. It also has inherent limits in other areas such as, among other things, distortion and noise.
What it lacks is a theoretical upper frequency limit
due to the encoding. Digital overcomes the inherent physical limits of the medium and replaces these with an inherent high frequency limit due to the encoding.
In practice this allows digital to exceed the inherent limits of vinyl in every respect. In frequency response terms even CD vastly exceeds vinyl in the low end, and clearly exceeds it in the high end in practice. Though a vinyl record system can put out higher frequencies than a CD, this is always in practice about 100% noise. Below it's upper limit CD digital is flatter and cleaner in every way than vinyl recordings.
More importantly, CD is good enough to exceed the limits of human hearing, if only slightly. Once you have exceeded the limits of our ears then further improvement is theoretically pointless so far as playback is concerned.
should be able to verify off the CD then, yes?
15.080 kHz, according to my rip of Sgt. Pepper's that I bought about 10 years ago, i.e. not from the newer remasters.
Looks like that jives with the spectrograph of the needle drop that bandpass posted.
should be able to verify off the CD then, yes?
There are two different things here. The spiral lead out groove on some pressings of the vinyl. including the original British one (but not US) has the ~ 15 kHz tone. When it then clicks into the central, inner, locked groove circle you get the "Lucy abbey all the way" part and (as we see from bandpass' spectral analysis) the annoying tone ceases.
The "Lucy abbey all the way" part has been released on some CDs, such as Rarities, however the annoying tone part that precedes it has never been released on any CD that I know of.
Unlike CD, vinyl has no inherent high frequency limit, in theory.
This is wrong, i believe....
What it lacks is a theoretical upper frequency limit due to the encoding.
I have no idea where you got that I was talking about anything
other than the encoding of vinyl, compared to that of the CD format, which I named, so I wasn't wrong.
Unlike CD, vinyl has no inherent high frequency limit, in theory.
I wasn't wrong.
This is wrong and you should feel wrong.
The big question is precisely how you define that "high frequency limit", as vinyl keeps getting worse the higher you go, unlike digital's perfect bandlimited reproduction.
due to the encoding[/i].
I have no idea where you got that I was talking about anything other than the encoding of vinyl, compared to that of the CD format, which I named, so I wasn't wrong.
Well I have re-read the message I was replying to and I didn't see any mention of encoding in it.
is there a minimum or maximum frequency that can be "encoded" in vinyl (akin to the ~20Hz and 44.1kHz cutoffs on digital CDs)?
I was responding to this, the OP's opening question.
He used the word "encoding" and made it clear he was talking about the mediums of CD and vinyl, not their playback.
Also, I made it a point to emphasize I was speaking theoretically, by italicizing the words "
in theory" in my sentence:
Unlike CD, vinyl has no inherent high frequency limit, in theory.
---
The CD format has a hard limit in that only frequencies below 22.05 kHz can be encoded. This is not the case with vinyl. There is no hard limit, upper frequency limit like that with vinyl,
in theory.
. I stand by this.
In theory, it's still wrong. There is a theoretical upper bound to vinyl's frequency reproduction. Like I said, precisely how you define that limit will change where you place that theoretical upper bound.
It's not as clearly defined as CD, but that does not imply that there is no theoretical upper bound!
... The HF limit is dependent on many factors, but at the very least, CD4 records go all the way to 45khz. ...
The 4 channels of a CD-4 disc were limited to 15 KHz. Each rear channel difference signal was modulated onto a 30 KHz carrier. The cartridges had significant distortion and frequency response variations above 20 KHz or so, but because the modulation method was FM based the demodulated signal was acceptable provided the disc and stylus were in good condition.
... The HF limit is dependent on many factors, but at the very least, CD4 records go all the way to 45khz. ...
The cartridges had significant distortion and frequency response variations above 20 KHz or so,
Just a few dB off, even up to 45 kHz, isn't very significant in my book.
... Just a few dB off, even up to 45 kHz, isn't very significant in my book.
I agree, so long as you're talking about the frequency range above 20 KHz rather than below it.
Yeah (if I'm getting your humor), I guess when neither the test signal nor the deviated version is audible to human ears, terms like, um, "significant" don't really apply.
In theory, it's still wrong. There is a theoretical upper bound to vinyl's frequency reproduction. Like I said, precisely how you define that limit will change where you place that theoretical upper bound.
It's not as clearly defined as CD, but that does not imply that there is no theoretical upper bound!
Well, in very strict theory, every time-limited signal has infinite bandwidth, so...
The "Lucy abbey all the way" part has been released on some CDs, such as Rarities, however the annoying tone part that precedes it has never been released on any CD that I know of.
I think it's on both stereo CD releases (1987 and 2009). I haven't got the mono ones yet.
Cheers,
David.
here it is ripped from the CD (1987 edition)
But if you look at the full size pic, you can see the tone is straighter and cleaner than the needledrop...and of course the silence before it is blacker. Interesting to see to edit right into the 'lucy abby' bit compared to the needledrop. I presume the CD reflects what's on the the master tapes?
The cartridge I was using before I sold my vinyl rig claims a frequency response up to 70kHz (http://www.denon.co.uk/uk/Search/Pages/ProductSearchResult.aspx?k=dl304). No limits are given so it's pretty meaningless but I can't helpwondering how they tested it to come up with that figure. A test record that goes up to and beyond 70k?
Interesting to see to edit right into the 'lucy abby' bit compared to the needledrop. I presume the CD reflects what's on the the master tapes?
According to this, the 15 kHz tone on the original UK LP version was added manually, and is not on the master tape:
"dog cut"
basic recording- none
additional recording- none
master tape- none
cut into first mono master disk 28 Apr 1967.
cut into first stereo master disk 1 May 1967.
generated 1987.
On LP this 15 KHz tone is not a recording at all, but was cut mechanically into the master disk after A Day in the Life. The dates shown are of the first disk cutting for Parlophone in the UK; it was recreated at unknown dates for other master disks, but was never cut into disks by Capitol for US release. For CD it was created electronically in 1987 and put into the master digital tape used for CD production. "
http://www.norwegianwood.org/beatles/disko/uklp/pepper.htm (http://www.norwegianwood.org/beatles/disko/uklp/pepper.htm)
On LP this 15 KHz tone is not a recording at all, but was cut mechanically into the master disk after A Day in the Life. The dates shown are of the first disk cutting for Parlophone in the UK; it was recreated at unknown dates for other master disks, but was never cut into disks by Capitol for US release. For CD it was created electronically in 1987 and put into the master digital tape used for CD production. "
I've read that before, but my ears (not to mention the graph above) tell me something is wrong or missing from this description - it's a fairly hissy/noisy and slightly distorted tone on the CDs. In its deficiencies, it seems at least comparable to some of the LPs (though that's a tough call to make).
My early 1970s pressing of Pepper has the reversed-speech run-out groove starting at a different point in the loop from my 1987 CD.
I fear we're heading off topic here
Cheers,
David.
I suspect the CD's version is simply an audio recording of the LP's 15 kHz tone section, record noise and all, hence the reason they are comparable to your ear.
The Beatles insist the random chatter, in this instance, is neither reversed nor has any secret message. It doesn't sound like backward speech to me unless played backward (http://www.youtube.com/watch?v=erROQrJqzhw). [My turntable at the time could play backwards and I examined all of their stuff that was reversed, BTW]
There are several overlapping voices and in addition to the main voice I hear a secondary voice saying "[unintelligible] five by five".
I suspect the CD's version is simply an audio recording of the LP's 15 kHz tone section, record noise and all, hence the reason they are comparable to your ear.
That 'looks' reasonable, since a pure tone would not have all that noise around it if generated and recorded digitally. But the one on the CD is still more stable-looking than the one in the needledrop posted. Difference in analog playback rigs?
Yes. We are not only seeing two different turntables/cartridges/phono preamps, we are also most likely seeing two different pressings, from two different cutting lathes, with different center spindle hole alignments (a notorious source of wow), from two different records which may have huge differences in wear. [And very high frequencies, especially recorded in the central area, are the first to go, or at least distort, from needle damage.]
This whole conversation has me intrigued about the technical limitations of vinyl. For example, is there a minimum or maximum frequency that can be "encoded" in vinyl (akin to the ~20Hz and 44.1kHz cutoffs on digital CDs)? Does this differ for 33s, 45s, 78s? Is there a way (based on the needle tip and equipment setup) for a knowledgeable person to precalculate the bands that the most noise interference will likely be in, and thus minimize or remove it? Etc.
It's particularly pertinent as I'm planning to attempt a vinyl-to-FLAC encoding soon.
Since you still have not received a reply to this question, I will give it a shot, but truthfully you should be asking someone who is currently mastering vinyl and has the technical background to answer in practical rather than theoretical terms.
The information you have been given so far is mostly wrong. Infinite frequency can only be recorded with an infinitely small stylus, which does not exist in practical terms. Other factors, such as the grain size of the vinyl plastic molecules, and the tensile strength of the vinyl at high playing temperature for any given thickness of groove wall also enter into the equation in a way I cannot address, but keep in mind that as frequency goes up, the groove wall shifts orientation from side to side to front to back. It goes through a 90 degree rotation, and thins out too because the waves crowd closer together. All of these mathematically inescapable physical limitations play a part in the upper frequency limit. Complex geometrical calculations are required to determine the absolute theoretical limits under any given condition such as stylus geometry and groove velocity.
This example below assumes the viewpoint of cutting the master, because that is where the ultimate limits of the technology lie. Duplication errors reduce the performance from the master theoretical limit. Once the master is cut there is absolutely no way to improve the result on playback. Turntables are already optimized for playback at baseband.
The groove velocity (linear velocity of the groove as it traverses across the stylus from front to back) and the dimension of the stylus that lies along the groove (rather than across it) pretty much dictate the upper frequency limit. The reason can be summed up like this. If the frequency is infinite, the groove is essentially motionless, and the stylus retraces its path across groove that has already been cut on cycle n as it cuts cycle n+1.
Stated alternatively, for a fixed groove velocity and stylus dimension (length, let's call it, and keep in mind the length is longer at the top of the groove than at the bottom because the stylus is pointy) the maximum frequency that can be cut is the highest frequency for which the groove displacement is larger than the wavelength of the audio plus the stylus length. If the groove displacement is smaller than the wavelength plus the stylus length, the stylus will recut material that has already been cut, destroying the features of the cut groove.
Remember, as the wavelength gets shorter, so does the groove displacement, because the frequency goes up, shortening the time between peaks. At some limit the wavelength will shorten until it is equal to the length of the stylus, and at that point the leading edge of the stylus cannot cut any waves into the groove because it has already cut that part of the groove on the previous cycle with the trailing edge of the stylus.
Long before that point is reached, the amplitude of the wave that is cut into the vinyl will begin to go down because the stylus starts partially (as opposed to completely) retracing the cut of the wave into the groove when any part of the prior wavelength is still in contact with the stylus. This means that in addition to the absolute high frequency cutoff point where no signal at all can be recorded, there is also a rolloff point where the amplitude begins to decline at lower frequencies too, and a somewhat linear trace connects the two points together (in the dB world, anyway), assuming that the rolloff follows some usual curve. Then again, this particular rolloff is not due to resonance, but rather to overlapping waves, so who knows what it really looks like...
The dimension of the stylus must be added to the wavelength because half the stylus sticks out past time zero on the leading and trailing edge of the cut, so the leading edge of the stylus adds to the length of the trailing edge of the cut wave, and the trailing edge of the stylus adds to the length of the leading edge of the cut wave.
This is why shorter stylus of the elliptical or shibata profile, where the stylus is wide to contact the groove, but short to contact in as small a profile as possible, is the preferred stylus. It allows a stylus to trace higher frequency waves without mistracking or obliterating detail. Conical stylus essentially runs through your record like a snowplow, permanently destroying the high frequency content on the very first pass. Shibato or elliptical stylus concentrates the contact force into a smaller area, causing more rapid wear on the stylus that also damages grooves, so checking and replacing the stylus regularly is the single most important factor (after cleaning stylus and record, that is) to preserving fidelity.
This is why elliptical and Shibata stylus is preferable to conical. Shortening the stylus from front to back allows it to track a higher frequency accurately.
As already stated elsewhere on this forum, 70KHz may be the upper limit of vinyl, but no one I know specified either the stylus dimension or the groove velocity. Remember, single masters used full 12 inch LP form factor running at 78 rpm to maximize fidelity. I presume that is the format that will get you 70KHz but I do not know if that is measured at the outer edge where groove velocity is maximum, or at the inner edge where groove velocity is minimum. The average 33 1/3 rpm LP has noticeable high frequency rolloff and distortion on the innermost track fresh out of the wrapper, but at the outer edge where the groove velocity is higher, the noise has higher frequency components also, giving it a nice breezy whooshing character.
As far as filtering a commercially released LP is concerned, I would expect that your sound equipment already filters out anything outside the audible range of 20Hz to 20KHz and as such you should not need to add any additional filtering. Your preamp will handle it for you. The only additional filtering I would recommend would be a lowpass at 10KHz or 15KHz to denoise a particularly damaged LP, or a highpass at 60Hz or so to denoise a particularly badly warped LP that is making your woofers fly out of their baskets and thumping your pictures off the walls. However, I would apply the filtering at playback rather than at recording, because you might change your preferences depending on the listening environment. If you filter your recordings there is no way to go backward to the original sound...
There are a dozen or so other factors to consider also such as cartridge resonances and cutting amplifier limitations that also limit the frequency response. Take it from me, an engineer, what engineers do is start with an impossible specification in terms of time, budget, performance, and price, and massage it all until it sort of fits into the design space, then ship it quick before the market window closes. If vinyl does 70KHz in the best case, and 45KHz for quadraphonic frequency modulation, then we can assume that it just barely fills the frequency range of perfect human hearing, namely 20Hz to 22KHz, with some degree of linearity, and from what I recall, almost every cartridge specification except the very most expensive fail that specification in either the low or high end, or both, by more than 6db, when actually installed in a turntable. If you get 15KHz out of old vinyl you should consider yourself lucky. If you are over the age of 40 you should also consider yourself lucky if you can still hear 15KHz, especially if you are an audio junkie that warms up your rack of amplifiers every evening with a 6 pack of beer and a pizza and your best friend.
Good luck.
... The information you have been given so far is mostly wrong. ...
Regrettably, so is the information you have given, for the cutting part of the process at least. (The reproduction side info is OK.) Your explanation appears to assume that the cutting stylus has the same shape as a reproduction stylus, which is far from the case.
Cutting is done with a "chisel" profile stylus. It is basically a shallow triangle in profile as viewed from above. The major flat face faces the direction of cutting. There are minor facets cut on the "corners" to burnish (smooth) the walls after cutting, and often other slight modifications to the profile. The maximum frequency that can be cut by the stylus is determined by the size of the burnishing facets, and by the frequency / excursion / tracing speed product - if you move the stylus side to side too fast and too far, the back sides of the triangular stylus will slam into the previously cut walls.
In practice, the cutting HF limit is usually determined by the design of the stylus driving system. Remember that the signal being fed to the head has a reverse RIAA response curve - the HF energy is incredibly boosted. The drive coils in the head are very small and have a low thermal inertia. A second or two of excess HF energy will burn them out. CD-4 records, with frequencies up to 45 KHz, were actually cut at half speed (16 2/3 RPM). This only required 1/4 of the HF energy that full speed cutting required. They still had to be cut about 6 DB lower in level than a standard stereo disc.
For the original question, I would suggest applying whatever processing is required to make the result sound acceptable to the user. In my limited experience, it's not worth saving it "warts and all" unless there will be no opportunity to re-rip it later. The user may well find that a turntable or cartridge upgrade will result in reproduction better enough that he will want to re-rip anyway. I've found that most high-end turntable/arm/cartridge combos do a competent job of reproducing the music. They tend to differ more in how well they de-emphasise vinyl shortcomings such as surface noise.
... The information you have been given so far is mostly wrong. ...
Regrettably, so is the information you have given, for the cutting part of the process at least. (The reproduction side info is OK.) Your explanation appears to assume that the cutting stylus has the same shape as a reproduction stylus, which is far from the case.
Cutting is done with a "chisel" profile stylus....[snip]
I think you may possibly be maligning CherylJosie. I got the impression that her discussion of stylus shapes (elliptical, Shibata, etc) was during a section talking about playback, but when she got to the bit about cutting, she makes no mention of stylus shape. For sure, as I read it I mentally noted that she failed to mention that the cutting head is not the same profile, but that doesn't necessarily mean she thinks that it is.
The discussion about cutting limitations applies in a general sense. The limits will vary according to cutting head profile, of course, but the fact that there are limits determined by basic geometry holds for any profile. I don't think it really matters that she didn't explain the cutting head profile when all she was trying to establish is that there are such limits.
... I think you may possibly be maligning CherylJosie. ...
On the technical points, I disagree. I could quote and rebut the specific points, but it's not that big a deal. I accept that I could have left out the first sentence of my post. I'm getting old, I need to fight the tendency to become grouchy as well.