I have a slightly different take on Mark's comments, not to say that his methods didn't serve him
very well in the realm of professional for-release mastering.
I've tested the idea of noise being added by the analog stage of my converters (both Benchmark ADC
and CardDeluxe PCI card). When normalizing from a -6dBfs, or even from a -12dBfs peak to -1dBfs,
there is no audible noise at any reasonable listening level added by the converter input. Yes, the
noise floor of the analog electronics _before_ the coverter can be boosted enough to be audible, but
to me that's a simple case of moving the inherent dynamic range (ie it wouldn't matter whether you
added gain on the way in or normalized the digital result, the noise floor of the playback equipment
relative to content peak is the same both ways). I think Richard is saying basically the same thing.
We may have three different but equally effective MO's going here. Mark's method harkens back to the
days of 1630 mastering, where you needed to find and set your peak levels before you made your A-D
transfer. This was definitely how my mother operated with the Mercury CD's; she and her engineers
knew to set the dcs ADC input level where digital zero basically equalled Audiotape 1960 brown-oxide
saturation, because in almost all cases, there would be an fff peak that would be at or near
saturation (Ampex 350 playback electronics have more headroom than brown-oxide tape, so electronic
distortion was not a consideration). There are technically some very brief digital over-zeros on
some Mercury CDs here and there, but no one has ever returned one because it won't play or it sounds
lousy. The manufacturing plants tolerated them because their own QC people couldn't hear anything
that would lead to merchandise returns. To be honest, I'd be more conservative with levels if I were
doing it today, but I'd be working in 96/24 and because I was working in a DAW, I'd have a chance
for as many "do-overs" as I wanted. When you're transferring to a linear medium and the studio clock
is running to the tune of dozens or hundreds of dollars per hour, you behave differently!
Richard's MO is very modern and I'd say on the cautious end of conservative. However, with a
very-low-noise signal chain, it should work and I'm sure Richard has done his own noise-floor
analysis and is OK with the results. His long list of satisfied clients speaks for itself.
My MO is somewhere in the middle, but only because I came up using linear media and learned to
maximize s/n (ie keep levels as high as possible without electronic distortion or noticeable tape
saturation). Since I started transferring at high-resolution about 10 years ago, I've dialed back on
input levels significantly, landing where I described in my earlier post this morning. Once again,
if I were doing more aggressive DSP, I'd be more conservative on the transfer levels because
aggressive DSP can severely effect maximum level and dynamic range, especially EQ because you are
undoubtedly adding or removing something that has an out-of-phase component somewhere, and that can
greatly effect peak level, ambient noise level and overall levels. Observe how much levels change on
a 78RPM transfer as you experiment with different rumble treatments. Same to a lesser degree with LP
records and also with mass-duped tapes. In theory, master tapes shouldn't have rumble problems that
-- Tom Fine
----- Original Message -----
From: "Richard L. Hess" <[log in to unmask]>
To: <[log in to unmask]>
Sent: Friday, October 19, 2012 2:54 PM
Subject: Re: [ARSCLIST] Analog to digital dBFS standards
> Hi, Mark,
> Your advice, as always, is greatly appreciated. Yes, I do tend to try and optimize when I have an
> album master, but many tapes I receive are long-form (more than a few are 1/4-track) and I think
> I'm doing well if tape saturation occurs at about -1 or -2 dBFS in the transfer process.
> Also, I would never do a transfer where the converter noise is obvious. No matter where we
> increase gain if the gain staging is properly analyzed and addressed the noise source should be
> some combination of tape and first stage of the tape repro preamplifier--normally it is almost all
> off the tape. I worry if I do not hear a large jump in noise when I hit play.
> I am certain you have done your share of tapes with wildly varying levels that go from "in the
> noise" to saturation. Since I normally do not have a budget on a per-tape basis to scan for
> highest peaks, it is with these tapes that I am conservative. Album masters and other "high
> quality" tapes are a different matter (and a different budget).
> It would be interesting to hear the context of the types / content / quality of the tapes the
> original poster is digitizing.
> On 2012-10-19 2:02 PM, Mark Donahue wrote:
>> Couple of quick comments from someone who does a fair amount of re-issue
>> mastering from material that has been "Domain Transferred" as part of
>> 1) Don't normalize levels. spend the time to get the levels right on the
>> transfer. Gain staging is your friend. I regularly see stuff that was
>> transferred at peak levels of -12 to -18 then normalized. While in the
>> digital domain the process is pretty transparent, it is in the analog
>> domain that there is a problem. The analog noise floor of converters is
>> usually somewhere around -100 and typically has an audibly different
>> character when compared to the analog noise from the tape machine. When you
>> add 12 to 18 dB of gain the noise floor becomes audible. It's really not
>> that hard to get the peak levels into the top 6 dB. Get a good quiet
>> attenuator to adjust the levels feeding the converters. This stuff is
>> really not rocket science.
>> 2) Get to know the safe operating levels for your equipment. We've
>> calibrated out converters for the past 30 years to +18dBu=0 dBFS. We tend
>> to use professional equipment and test the devices to see where the device
>> clips. Most professional analog gear has at a minimum peak level of +22
>> before clipping. Many Semi pro and consumer devices are incapable of
>> getting to +18 without clipping or distortion. You need to find out where
>> your equipment clips and leave the appropriate headroom. Again, gain
>> staging is your friend.
>> 3) Operating levels for tape machines are all relative. (Unless calibrating
>> for Noise reduction). It's the same if someone calibrates to 185 and peaks
>> at +3 an the VU or 250 and peaks at 0 VU. On tapes without tones it is
>> most important to get the azimuth correct. Calibrate level and EQ to the
>> MRL and then align the azimuth for peak high frequency on the tape to be
>> transferred. Equalization can be corrected later with very little
>> degradation. For folks dealing with tapes from the EU, learn to hear the
>> difference between NAB and CCIR.
>> These are just a couple of my pet peeves. I could go on for hours.
>> I'll step off my soap box and let the conversation continue.
>> All the best,
>> Mark Donahue
>> Soundmirror, Inc.
>> Boston, MA
> Richard L. Hess email: [log in to unmask]
> Aurora, Ontario, Canada (905) 713 6733 1-877-TAPE-FIX
> Quality tape transfers -- even from hard-to-play tapes.