1. most if not all professional tape machines had a "bias trap" (low-pass filter) to block bias
interfering with playback. Keep in mind that early tape recorders, and especially early magnetic
film recorders, had relatively low bias frequencies (several well under 100kHz), which could beat
with some audible frequencies. So the bias trap was an early known thing in magnetic recorder
2. in the case of a Plangent transfer, a separate electronics path recovers bias, onto a separate
digital track. The audio tracks do not have any bias interaction or "pollution." John Chester can
explain this in better detail, since he built the electronics.
Also, I agree very much with John Chester's point that the digital "recording" system must introduce
as few distortions or errors or anything else as is technically possible, because all of that
compounds on the playback errors from the analog source. One thing we haven't discussed is that
there are many different ADC designs, some offering a more transparent analog front end and also
some being better able to deal with peak-levels from the source than others (ie analog headroom --
you certainly don't want overload distortion to happen before something is digitized). The thing
that I find very impressive about the march of knowledge and technology is that we have these kinds
of tools today, at affordable prices. This has all happened in my lifetime. When I was born (1966),
NHK had not yet developed a reasonable-fidelity digital audio recorder.
-- Tom Fine
----- Original Message -----
From: "DAVID BURNHAM" <[log in to unmask]>
To: <[log in to unmask]>
Sent: Monday, November 23, 2015 10:50 AM
Subject: Re: [ARSCLIST] What is really higher resolution?
Well also, (and I haven't read about this anywhere but it seems plausible), the analog signal on a
reel to reel tape is riding on a high frequency bias. This bias signal is on the tape as an audio
signal, (plainly heard if you are "rocking" the tape across the head at a very slow speed. If you're
sampling at 196khz, I can visualize some interference between the bias frequency and the sampling
frequency unless a low pass filter completely removes the bias frequency.
On Monday, November 23, 2015 8:42 AM, Tom Fine <[log in to unmask]> wrote:
Correction -- I meant to say:
"Then there's the fact that some tapes AREN'T slit perfectly enough to ride through the transport
with relatively even track-tracking (i.e. relatively perfect azimuth throughout the tape)."
For what it's worth, Irish/Ampex/Quantegy battled slitting-imperfection issues throughout its
existence. 3M slitting was generally better, as was BASF and Agfa and, I'm guessing (due to no
experience with the product), EMI. Audio Devices (AudioTape) was also generally slit precisely.
Keep in mind too that age physically distorts some tapes, so the ride across the heads is even less
perfect. We all know about acetate tapes warping and curling and shrinking. There hasn't been much
science done on baking polyester tapes and whether that creates transport-azimuth issues. My own
experience is that some tapes do have issues because, after baking, there is some non-sticky residue
on the tape surface and edges. This problem can probably be mitigated by running a baked tape across
a Pelon wipe before transfer.
My main point is -- and you can test this with sine-wave and square-wave signals on even the most
modern and well-maintained tape machines -- tape is far from output=input, so it's falacious to say
its output is "steady-state" fidelity to the original source. Tape's own mechanical and electrical
imperfections are constantly "chopping bits off" the original signal, and thus reducing resolution.
Same with disk recording and playback.
-- Tom Fine
----- Original Message -----
From: "Tom Fine" <[log in to unmask]>
To: <[log in to unmask]>
Sent: Monday, November 23, 2015 7:27 AM
Subject: [ARSCLIST] What is really higher resolution?
> There is an argument to be made that analog media playback can't possibly offer that many data
> points to be collected. To wit ...
> 1. when you play a tape, you are fighting the laws of physics. For one thing, no transport
> provides a perfect ride across the heads. Resolution is damaged by wow and flutter (time-smear),
> plus imperfect tape-to-head contact cause by anything from uneven head wear to imperfections in
> the tape surface to simple dust and other particles in the air. There's also static electricity
> and other results of friction. Then there's the fact that some tapes are slit perfectly enough to
> ride through the transport with relatively even track-tracking (i.e. relatively perfect azimuth
> throughout the tape). Tape electronics, especially old ones, are prone to what are now considered
> high levels of distortion and noise, and unless they have been thoroughly overhauled, aging
> components compound these problems.
> 2. a commercial disk release has mechanically-lowered resolution from the get-go. There are some
> issues with lacquer "memory" (where the groove shrinks back a little bit when it cools after
> cutting -- this is a controversial topic among cutting engineers, but direct-metal mastering was
> invented as a solution to this alleged problem). Then there is resolution-loss in the plating
> process, because the laws of materials science and physics say it's impossible to make a perfect
> imprint (there is some granularity to all materials, plating cannot be perfectly uniform, etc).
> And, the pressing machines can't be perfectly clean on every press, the vinyl biscuit can't be
> perfectly pure, etc. In fact, if you think about disk-manufacturing, it's quite miraculous that
> the whole system got to where it can sound as good as some records do. With shellac records, it's
> even more so because the technologies hadn't evolved as much and shellac itself is a very
> imperfect carrier material. So, before a stylus even hits the groove, you have stages of materials
> imperfections baked in (literally), which results in at least surface noise if not ticks, pops and
> groove distortion. And then there's the matter that no stylus tracks a groove perfectly, there is
> an inherent noise floor in all mechanical playback (dragging a diamond through a groove), and that
> only the very best preamps offer super-low noise floors (this is even more of an issue with
> low-output moving-coil setups).
> One argument made by the anti-digital crowd of yore (I don't hear this argument made about
> higher-resolution digital, except by ideologue zealots) is, "no matter how much you sample a sound
> wave, you're still breaking it into chunks and it's not a steady-state wave." But, see above. The
> output from the analog playback system itself is not really a steady-state wave. Physics and
> materials science prevent that from being so. So the question is, which system is actually
> capturing more "resolution"? Let the debate begin!
> -- Tom Fine
> ----- Original Message -----
> From: "Corey Bailey" <[log in to unmask]>
> To: <[log in to unmask]>
> Sent: Monday, November 23, 2015 3:14 AM
> Subject: Re: [ARSCLIST] Hi-Rez symphony recordings
>> Data points per bit should have read: data points per dB of dynamic range (I changed the math,
>> but not the description).
>> So, the comparison should read:
>> Lets take a look at the available data points for each dB of dynamic range for CD quality digital
>> audio: 65,536 (data points) divided by 96 (dB of dynamic range) = 682.6 data points for each dB
>> of dynamic range of a given sample.
>> Compare that to 24 Bit/ 96K digital audio: 4,294,967,296 (data points) divided by 144 (dB of
>> dynamic range) = 29,826,161 data points for each dB of dynamic range of a given sample. And,
>> there are more than twice a many samples taken! Now, we are talking about some decent resolution.
>> Also, my hand typed chart got wacked by the forum formatting. Hopefully, you get the idea.
>> Corey Bailey Audio Engineering
>> On 11/22/2015 7:47 PM, Corey Bailey wrote:
>>> Mr. Kevil:
>>> Let's take a look at Bit Depth as applied to digital audio: "Bit Depth divides a given sample
>>> by its value."
>>> BIT DEPTH DATA POINTS DYNAMIC RANGE
>>> 8 256 48dB
>>> 16 65,536 96dB
>>> 24 4,294,967,296 144dB
>>> Now, lets take a look at the available data points per bit for CD quality digital audio: 65,536
>>> (data points) divided by 96 (dB of dynamic range) = 682.6 data points for each bit of a given
>>> Compare that to 24 Bit/ 96K digital audio: 4,294,967,296 (data points) divided by 144 (dB of
>>> dynamic range) = 29,826,161 data points for each bit of a given sample. And, there are more than
>>> twice a many samples taken! Now, we are talking about some decent resolution.
>>> So, even though you have only 70 dB of dynamic range available for those old tape recordings
>>> (before the introduction of Noise Reduction), you will obviously capture much more of that
>>> available dynamic range using the archival standard 24Bit/96K (Hi Rez) sample rate and bit
>>> The same logic applies to any of the old audio carriers.
>>> Plus, when restoration takes place, the results are less artifact prone when using higher bit
>>> depths and sample rates.
>>> Corey Bailey Audio Engineering
>>> On 11/22/2015 3:03 PM, L. Hunter Kevil wrote:
>>>> A transfer of a 1960s tape marketed in a 24/96 wrapper is what? Doesn't the resolution of the
>>>> tape correspond to the equivalent of an 8- or 12-bit word? If so, what does the wrapper do?