From: Patent Tactics, George Brock-Nannestad


the discussion turned on reverse reproduction of an analog tape and also half-
speed mastering--the latest was a comment by Tom Fine.

Reverse reproduction of analog tape when making a copy (or cutting to disc) 
was well-know. As usual, because this subject is not my main line of 
activity, I have difficulties in picking the actual references from my 
shelves/piles. The procedure was independently invented, and I know that 
Marvin Camras took out a US patent in the 1940s, related to magnetic sound on 

The rationale behind it is that whenever you reproduce a tape on the recorder 
that made it, you may have amplitude errors in dependence of frequency 
(linear distortion) that are added from the recording and reproducing 
process. And this happens, whichever direction you reproduce in. However, the 
time delay errors are added in the forward direction but annihilated in the 
reverse direction. The delay on the tape at a particular frequency during 
recording is won back on backwards reproduction. This means that transients 
are better defined than if it were a forward reproduction. And I should 
mention that azimuth error has no influence at all.

I used this knowledge when I recorded gritty noise from "silent" shellac 
grooves for mixing in variable quantity to the sound from an LP. I recorded 
full track and then I reversed the tape for reproduction and mixing. However, 
in that case the "programme", the noise, was indistinguishable as to 
direction, only I wanted well-defined transients, because it is those that 
overload the input stages of preamplifiers. Ha, I used a NAGRA, and only 
later did I discover that the NAGRA (I forget, III or IV) had a time delay 
compensating filter that made my exercise superfluous. 

However, when transferring 1950s tapes on the full-track tape recorder they 
were recorded on, then reversing the tape might be relevant, even today.

As to half-speed mastering, that is going to create problems. It is only 
something done when mechanical elements, like a cutting head are involved. 
The bandwidth of electromechanical elements is not high, and in particular 
the strange behaviour at high frequencies make half-speed mechanical 
recording attractive. And if the surface of the mechanical record is dented, 
your tracking at replay improves. But in magnetic recording everything 
depends on the relative velocity between the small magnets on the tape and 
the head. Merely changing the speed does not change the relationship between 
gap width and wavelength. So, if you go half-speed, you get a lower voltage 
induced in you tape head, and the distance to the electronic noise floor is 
reduced by 6 dB. 

Kind regards,


> I forgot to mention ... what about half-speed mastering? So there you're
> having the tape machine 
> treat the transients at half the original velocities (that might be the
> wrong word). How does that 
> compare to original-speed fidelity? My point is, it's an interesting debate
> and I'm not convinced 
> there's a "right" answer as long as you can have original-source azimuth
> alignment and make sure 
> polarity is maintained. I don't consider a magnetic tape playback system
> capable of perfect fidelity 
> from tape to tape and machine to machine, so I think you can't set hard and
> fast rules and, joyfully 
> for some of us and dreadfully for others, we must use our ears and
> aesthetics as the final decider.
> -- Tom Fine
> ----- Original Message ----- 
> From: "Tom Fine" <[log in to unmask]>
> To: <[log in to unmask]>
> Sent: Thursday, October 08, 2009 7:32 PM
> Subject: Re: [ARSCLIST] Reverse engineering
> > This backs up what I have experienced here. The reverse-copied tracks can
> (but don't always) sound 
> > "crisper." And yes, Richard is right -- reverse polarity. You will
> probably hear a difference with 
> > that, too.
> >
> > As I said in my previous post, I'm not sure what's "right" vis-a-vis
> fidelity since you're not 
> > playing it back on the machine on which it was recorded and all
> head-playback 
> > electronics-alignment combos have some sort of sonic signature, be it mild
> or not.
> >
> > Finally, Richard is 100% right about low-fidelity oral histories. It gets
> more debate-worthy when 
> > you get into an instance like well-made music recordings that are
> 1/4-track 7.5IPS and you want to 
> > ingest all 4 tracks at once.
> >
> > Oh, and regarding duplicated tapes ... the unfortunate truth is that there
> is almost never an 
> > azimuth match in the two "sides." The reason, especially with
> quarter-track 7.5IPS reels duped 
> > back in the "golden age" days of the late 50's and early 60's, is that
> there were two record heads 
> > that were almost never in perfect alignment. I think SOME but not all
> later cassette dupers had 4 
> > stacked tracks, but most formats from the days of Ampex 3000 series dupers
> had pairs of record 
> > heads, one for the "side A" tracks and one for the "side B" tracks.
> 8-tracks would be four and 
> > four, from the heads I've seen and dupers that have been described to
> me.
> >
> > The practice I ended up adopting for quarter-track tapes with music is
> just charge for the time 
> > required and ingest one side at a time with an azimuth tweak each time.
> With home-made tapes, if 
> > the azimuth was _really_ well aligned at the factory, you can sometimes
> get away with a 4-at-once 
> > ingestion, but a sharp-eared client can hear the difference with
> transients and may not like his 
> > two sides sounding different.
> >
> > -- Tom Fine
> >
> > ----- Original Message ----- 
> > From: "Robert Cham" <[log in to unmask]>
> > To: <[log in to unmask]>
> > Sent: Thursday, October 08, 2009 6:03 PM
> > Subject: Re: [ARSCLIST] Reverse engineering
> >
> >
> >>I should think that it isn't relevant to digital copying at all, but back
> in the analog '70s we 
> >>used to high speed dub reel to reel tapes backwards because seeing a decay
> as a transient was much 
> >>easier on the electronics than an attack, especially a percussive one. 
> That was the 
> >>recommendation of Ampex, who built our duplicators.
> >>
> >> Bob Cham
> >>
> >>>Back in the day when duplicating tapes was a day job for me, they said
> that side 2 of cassettes 
> >>>duped ay high speed both sides at once would sound better than the side 1
> would. Never made much 
> >>>difference on cassettes, especially at 64 or 128IPS dupe speed, but some
> people told me copying 2 
> >>>tracks worked better in reverse too... They said the electronics could
> respond to transients 
> >>>backwardsbetter than forwards. I have no empiric evidence of this
> though.
> >>>
> >>>Just old tape tales by now, but this had me thinking back... or
> backwards!
> >>>
> >>>I have transferred some quarter track tapes doing all four tracks at once
> top a four channel A/D, 
> >>>and not noticed a significant difference, but it is easier to do them one
> side at a time as then 
> >>>they end up tails out, as long as it is an hourly job and not a mass flat
> fee transfer project.
> >>>
> >>>Hope this isn't irrelevant!
> >>>
> >>>Lou
> >>>Lou Judson
> >>>Intuitive Audio
> >>>415-883-2689
> >>>
> >>>On Oct 8, 2009, at 1:59 PM, Richard L. Hess wrote:
> >>>
> >>>>
> >>>>>It's not the digital realm, its the way the reel electronics handle
> transients and phase
> >>>>
> >>>>There appears to be waveform differences between playback in the two
> directions after accounting 
> >>>>for the polarity flip. To my ears, this is an acceptable tradeoff for
> copying oral history tapes 
> >>>>in half the time. This is especially true of mid-to-low-fi recordings
> such as some 3.75 and most 
> >>>>1.88 in/s reels.
> >>>>
> >>>>Cheers,
> >>>>
> >>>>Richard
> >>
> >