In response to Mike Fradis’ of Waves email posing questions about a proposed Waves project to model a Tape Emulation plug-in, I answered Mike’s initial questions and also provide additional thoughts on the subject. This White Paper is a highly personal and subjective response to the topic. When I wrote this I was providing as much information as possible to allow Waves can select those aspects and points of view that suited their product development and marketing agendas.
Please note. As I am not an employee of Waves Inc. and I am not being compensated for this document, the copyright of this document, it’s intellectual property and it’s verbiage remains with me. Reproduction of this document is not permitted without the written permission of the Author. Waves does have my permission to use any of the intellectual property contained in this White Paper, at no cost, for the product development discussed in this document.
Please describe the sound you are seeking to achieve when choosing to record or mix to tape.
Sweetness and musicality. Some say warmth, which is close, but warmth can also be construed to mean extra bottom end and a somewhat subdued top end, some might say a ‘laid back sound’. When I say warmth, with regard to the ‘tape sound’, that is not what I mean. When I say warmth I mean the opposite of coldness of sound (which is the supposed ‘digital’ problem). I define warm in a recording medium as a ‘human’ sound’, a ‘musical’ sound, an ‘accessible’ sound, a sound that does not fatigue the ears or the mind, a ‘soulful’ sound, a ‘natural’ sound.
Assume you have a great sounding tape modelling plug-in, and you had to choose only three adjustable controls from the following list. Which would you choose? Please rank only three in the order of importance, the first being most important
• NAB/IEC adjustable filter
• NAB/IEC fixed selector
• Tape Speed
• Machine Type
• Tape Type
• Playback using play head as well as play/record heads (slap delay)
It’s impossible to provide a simple trio of answers. It is actually a large ‘suite’ of these items that effects my choices for a particular analogue tape sound.
This decision is also based on musical requirements (genre’ being a big one). But, that said … if I had to make a list it might read like this:
1.) Tape Machine Type.
This is a crucial decision because the distinctive sound of the various tape machines electronics coupled with the influence of their transports (i.e. wow, flutter, scrape filter effect, etc.).These items can have a massive impact on the character of an analogue tape recording. Following is a list of tape machines I have had extensive personal experience with. My detailed comments and preferences are detailed in my answer to your final question.
(Tape Speed is also what effects the slap delay when feedback is used … we used a VSO [Variable Speed Oscillator] to drive the tape machine for finer control of this delay.)
There has always been a struggle between 15ips and 30ips, with much disagreement within the engineering community on this topic. 30ips had an extended top end (plus the tape hiss was an octave higher, thus less intrusive) and a slight reduced bottom end while 15ips had a ‘fatter’ bottom end and a slightly softer top end. It was always a compromise but again this decision was almost always made based on the musical requirements (again genre’ rears it head).
3.) It’s impossible to select just three individual elements since the ‘analogue sound’ was comprised of a group of controls working together which effected the variety of sounds possible with any analogue tape recording.
a.) Tape Type (comments below on various tape types)
• I started with 3M Scotch 131, which was the standard magnetic tape for decades. This tape operated at the industry standard operating level of 185 nWb/m = 0db (Ampex 0 or “Ampex Standard Operating Level”).
• 3M Scotch 207 was one of my all time favorites. Sweet and musical. Of course 207 was a 1 mil stock as compared to Scotch 206, which was the standard 1.5 mil tape. The recommend operating level of this tape was 250 nWb/m = +3db, although I always rated it at 185 nWb/m, ran at 30ips and used Dolby A Noise Reduction (arguable for some, but it worked well for me). Scotch 206/207 never had very little of any shedding problems. I have a library of 206/207 some well over 40 years old and the majority is still in pristine condition. It should be noted that I have taken good care of my library, but have not resorted to any special procedures.
• I also did quite a lot of work on AGFA PER 525. It was a great sounding tape, but ultimately there were supply and consistency problems so I abandoned using it. As I remember this tape was rated at 250 nWb/m = +3db.
• Ampex 456 was not a great tape in my opinion. It was rather ‘square’ and ‘flat’ sounding. I have also seen nearly disastrous shedding problems with this tape. It is rated at 370 nWb/m = +6db, making it one of the first truly high output tapes.
• Further down the road the QuantegyGP9 series was quite nice. No tape was ever quite as sweet as the older, classic Scotch tapes but for a modern tape GP9 was quite acceptable. You could rate it well under the +9 (502 nWb/m) it was designed for, enhancing the transient response and then use noise reduction for more esoteric recording work. It was however a 1.5mil tape so the top end was always a little wanting. For R&R without noise reduction running it at +9 (520 nWb/m) and hitting it pretty hard resulted in a nice solid sound. Being a 1.5mil stock helped with the print through as well. But again, GP9 was a ‘modern’ tape and perhaps not the sonic standard one thinks of when one envisions the classic ‘tape sound’. For marketing reasons though, it must be included as a choice since is was so popular and in the end it was one of the only tapes you could buy when 3M and Ampex shut down tape manufacture. For that reason the younger engineers never experienced some of the older ‘classic’ tape stocks.
• Therefore, like the Digidesign (Avid) Reel Tape, the tape types that the user should be able to select should include Scotch 207, Ampex 456 (mostly because of it’s popularity) and Quantegy GP9.
b.) Tape Drive
Operating level was one of the reasons you picked a certain tape, so again, it’s hard to separate Drive/Level from Tape Type. The standard tape operating reference levels I am referring to are: 185 nWb/m = 0db (Ampex 0 or “Ampex Standard Operating Level”), 250 nWb/m = +3db, 320 nWb/m = +5db, 370 nWb/m = +6db and 520 nWb/m = +9db. Of all of the options, drive level had one of the most profound effects on the sound of analogue tape recording.
c.) Thickness of the Tape
Although not on your list, the thickness of the tape was a very important consideration. The thinner the tape the greater the problem with print through but the tighter the head wrap thus the smother and more extended the top end…usually I used the thinner tapes (1mil), recorded at lower tape drive levels [say Ampex zero 185 nWb/m, again gaining transient headroom] and using noise suppression to handle the tape noise resulting from the more conservative levels. Additionally Dolby Noise Reduction could help reduce the print through by a small degree! Conventional operators used the 1.5mil tapes, which didn’t not have the same top end, but also did not suffer from the potential print through problems. Certainly when the high output tapes arrived (above +3db) the ticker backing became necessary … although there was always the option of rating the tape very conservatively, gaining the additional transient headroom and then using noise reduction to deal with the tape noise issues. I know this works, because I resorted to this technique often as the more classic tape stocks became less and less available.
Bias was really critical with regard to the noise level of the tape and the quality of the high end of the tape. (There is a good definition of Bias in the KMT White Paper located here.) Typically I over biased 3db at 10k and 30ips on Scotch 207 (which was 1 mil tape).The published standard was to over bias by 1db at 700Hz or 1Khz but we found that at 10Khz we had a better resolution of adjustment. I understand that in Europe the standard was an over bias of as much as 10db at 10Khz, again claiming improved resolution of the bias adjustment. I do not believe these tactics of bias alignment were peculiar to any tape type (provided you did your bias alignment on the tape stock you were intending to use, and in some cases, on a roll by roll basis), but were more related to the eccentricities of analogue tape recording technology. That said, some experimentation with various tape types provided different results with different bias adjustments. Frequently this work was done by the technical department at the studios and I often followed their suggestions.
A thought! Although the Bias frequencies were in the range of 75kHz to 175kHz and mixed with the signal, the theory was that this mixed ultra-sonic signal would be inaudible upon playback. That said the purpose of this Bias signal was to “modulate the amplitude of the recorded signal to a higher average flux level and move the recorded signal away from the nonlinear zero-crossover range and into the linear portion of the curve”. Who is to say that, although theoretically inaudible, the Bias frequency did not modulate lower audible frequencies, thus having an effect on the sonic quality of the recordings? A form of inter-modulation distortion perhaps? Is this another less understood reason why analogue sounds like analogue and has been so hard to duplicate in the digital domain for so long?
e.) Pre-Emphasis Curves.
As an American Engineer my experience with pre-emphasis curves other than NAB is limited (fixed only). For a very short time, Ampex introduced AME (Ampex Master Equalization). Most of us did not like this curve’s side effects and Ampex rapidly abandon producing machines with this option. I would have no doubt that different pre-emphasis curves would have a powerful impact on the sonic quality of the recording, but it must be said that for these pre-emphasis curves to be implemented in a plug-in one would have to model the entire chain of emphasis>recording>playback>de-emphasis. One of the side effects of these curves was to extend the frequency response of the recordings while suppress the inherent noise of tape recording…with the side effect of increased, or at the very least, changed distortion characteristics.
Certainly CCIR, which was a European standard pre-emphasis curve was very important and must also be considered. Perhaps some other members of the team will be in a better position to comment on this curve. Of course all pre-emphasis curves had side effects and their advantages always had to be balanced against their disadvantages.
Do you usually follow the factory calibration procedure?
Other than the pre-emphasis curves, you must be referring to the Bias adjustment specification of various tape emulsions. The short answer is NO, but there is some additional information included in what I wrote on Bias earlier in this document.
On the other hand, if you are referring to tape drive, then great liberty was taken with that specification. In my case I tended to rate tapes conservatively to gain transient response, but for others, they increased the saturation and compression characteristics of the analogue recording process as a tool for using negative sonic behavior of tape overload to their advantage. Again, these choices were generally genre’ specific. Conservative rating of tapes for acoustic recording was the way I went. But for R&R I was as happy as anyone to ‘slam’ the levels on the tape (I must clarify, that even in that camp, I was still on the conservative side).
Do you have a strong preference for the sound of a particular tape machine?
To some degree this would be a bias based on the machines available to a variety of engineers. I had to opportunity to work with quite a few machines and I’ll relate them to you and indicate my order of my preference separated by mastering recorders Master-#X(less than 4 tracks) and multi-track recorders Multi-#X(over 4 tracks):
300 valve machines(Up to 4 Tracks)
Glorious but impractical!
351 valve machines(Up to 4 Tracks)
The Best! If it was still available it might very well be my first choice as a stereo mastering machine. I once owned one in a portable case and for the life of me I can’t remember what I did with it. Oh…how I wish I still had it! Master-#1
601-2 Portable Stereo Machines
Quite amazing for a small portable tape machine, but again, impractical.
Only a so-so machine. I have extensive experience with this machine and I would avoid it if I could!
Same as the AG-440…both machines were transitions from the great valve machines to the first of the really good solid state recorders.
MM-1000 – 16/24 Tracks
Nicknamed the “Master Muncher”. This was one of the very first large scale multi-track (16 track) build on the chassis of the original Ampex Helical Scan Video Recorders. Pretty dreadful!
MM-1100 / MM-1200 – 24 Tracks
This is what we used at Sunset Sound. They were, arguably, the best sounding multi-track recorders (24 track) that I ever used. The rumor was always that Sunset’s MM-1100’s were modified. I can’t confirm this, but if they were, my guess is that the modification was improved input transformers, perhaps by Dean Jensen. Multi-#1
ATR 100 Series (101/102) – 2 Track
The best Ampex solid state mastering machines. The ATR101/102 was the standard for modern ¼” machines in the US for many years until the Studer’s became more available. Master-#4
MTR-90 MkII/MkIII – 24 Track
A lovely machine. When I was at Todd-AO we were running, I believe, 65 machines facility wide. They had very stable electronics, sounded great and had one of the best and quickest (also quietest) 2” transports ever built. They performed flawlessly as slave machines driven by a master timecode system. I had somewhat limited experience using them for music, but I would not hesitate to use one for a music project. Multi-#2
MTR-12/MTR-15 – 2 Track
Every bit as handsome a machine as it’s 24 track big brother. I have recently purchased a MTR-12 for the transfer, digitization and restoration of my 1/4″ tape library. Master-#3
Stock 3M machines
This was the equipment purchased for the Elektra Records West Coast Studios I designed, built and operated while Chief Engineer for Elektra Records.
M-56 – 8 Track – 1” – A lovely machine. A good, unique, Iso-Loop transport. Perhaps the electronics were a bit touchy and the transport alignment had to be absolutely correct or they could ‘eat’ a tape in the blink of an eye, damaging it beyond repair. I wouldn’t rate this machine as many modern engineers have never used it.
M-79 – 2 Track
This was the mastering machine we used at the Elektra Studio. It sounded quite good, but it’s not rated here because the unstable electronics were a total pain. I also suspect that, because of this, finding one in proper operating condition for modeling would be nearly impossible … that said, I would still question if the results would be worth the effort.
These machines were based on the 3M transport but were highly modified with custom John Stephens electronics as well as a capstan free drive system. They also came in a 32 and 40 track configuration. “Constant tape tension is just one advantage of our servo-operated motors. The usual capstan and pressure rollers have been eliminated, and with them went a major source of flutter and oxide wear.” They were quite an esoteric machine that had a reputation for being extremely transparent. When they broke, John Stephens himself would personally show up and fix them (clearly a bit of an eccentric). The most notable thing about the Stephens machine was the drive system. I witnessed the Stephens machine save a very famous album and quite expensive album from total disaster (Ringo Star’s “Photographs”) when it was recorded on a early batch of Ampex 456 and the oxide started falling off before they got to overdubs. Only the Stephens machine, with it’s capstan free drive system, could play the master tapes so they could be transferred to another tape stock (as you would have it, 3M Scotch 206). I would suggest that the Stephens machine is way too esoteric to consider modeling.
A-80 / A-800 / A-820 – 24 Track
One of the best transports in the industry. This is a very popular machine that would have to be modeled and included in any Analog plug-in. It is included in the Digidesign Reel Tape along with the solid state Ampex (perhaps an ATR-102). It was, however, not my favourite machine and I would only use it if there were no other choices. In recent years I have had cause to use it again, I must say that my opinion has mellowed. That said I would still only rate it as Multi-#3.
A-80 MkII – 2/4 Track
Actually I liked this machine much better than it’s 24 track big brother and have mixed numerous albums to two track ½” with excellent results. The A-80 MkII would be my practical first choice for a mastering tape machine, especially given the difficult of finding, maintaining and aligning an Ampex 351-2. Master-#2
JH-24 – 24 Track
For reasons I have never fully understood the MCI is a very popular machine, especially in Australia (perhaps MCI had a great sales team, or perhaps the other competing manufactures didn’t think the small market in Australia warranted their effort). The sound doesn’t impress me, the electronics are unstable and the transport is a disaster. If you model it I would suggest it should only be because you find a strong consensus amongst the rest of the Waves team to include it.
JH-110B – 2 Track
Same as above!
Would you be using a tape plug-in primarily on individual tracks or your master bus?
Well, that depends on how much system overhead the plug-in takes. I suspect your ultimate product could be quite DSP hungry.
• If it sounds truly amazing it is quite possible I would use it on the Master Bus.
• I think it highly unlikely that I would use it on individual tracks.
• I think it mostly likely that I would use it on sub-mix busses or stems.
For example I work with quite a lot of sampled strings and orchestral instruments. Especially for strings I will create a sub-mix bus and then use Pro Tools Reel Tape across the entire bus. In addition I may also use the Phoenix/Crane Song plug-ins on the individual tracks (they are very low overhead and have a broad range of timbres available that suit the individual sections of a string orchestra quite nicely).Again this depends on the quality and texture of the samples and sample library used.
Especially with sampled strings it is almost always necessary to try to offset the artifacts of ‘digital’ technology in an attempt to capture some of the warmth, texture (rosin and horse hair) of a well recorded string section.
Additional Considerations Not On Your List of Questions.
Also a great deal of extraordinary analog tape recording was done using noise reduction (this is probably a highly arguable and perhaps even a contentious comment) so one must consider whether or not to include this as an option in any Analog Tape Emulation plug-in. Whenever any form of Noise Reduction was used it had a very strong influence on the sonic quality of the recording. I suggest these are some of the devices that could be considered:
• Dolby A (301) – I did the very first commercially successful recording using Dolby A Noise Reduction (Elektra Records – Judy Collins – “Wildflowers”). I was one of many who found the gains of using noise reductions dramatically offset any down sides and I used it extensively on most projects. Many others would disagree with me on this, some vehemently! Most of them would make the case that Dolby A Noise Reduction tended to ‘eat’ the snap and crispness of drum recordings…I wouldn’t argue with them on that, but it worked for me on the kind of music I specialized in recording
• Dolby SR – Dolby SR didn’t appear until I moved from the Music Industry to the Film and Television Industry. Dolby SR (Spectral Recording) technology clearly improved the Dolby A technology, but it also contributed unique and arguable artifacts of it’s own. That said, I like Dolby SR and would have used it in place of Dolby A if it would have been available when I was doing the bulk of my popular music recording.
• DBX Type 1 – DBX Noise Reduction was quite popular amongst a few for a short time. It was a one band noise reduction system and for my money, I wouldn’t get within a mile of it … never used it. I though it sounded like ‘crap’ because of the way it wiped out the high end and the fact that I could hear the noise reduction ‘pumping’ … but many would disagree with me.
• TelCom/Telefunken C4D Noise Reduction – Highly esoteric. I only used this on one project but I must say it was a totally amazing sound. “…the system was called TelCom in it’s professional version and that was quite popular with European broadcasters since it solved all of the tracking and calibration problems of the older systems and to boot has a somehow “sweet” character to it’s sonic imprint.” The problem was, in the US, these systems were as rare as ‘hens teeth’ so it was not practical to follow up on this experience. To this day the TelCom Noise Reduction remains my all time favorite.
Tape compresses gradually, beginning with the high frequencies first, which prevents the high frequency peaks from sounding harsh. This is perhaps where the analogue tape recording has gained the reputation of sounding ‘sweet’ or ‘better than live’. In this regard 3M Scotch 206/207 was the epitome of this sound. Like tape distortion, by over-driving analogue tape the natural saturation and compression characteristics of analogue recording can be manipulated for sonic advantage.
In and of itself, the inherent distortion of the analogue recording medium contributed to the ‘tape sound’. I would say there were two classification of this distortion that needs to be considered in an emulation plug-in.
• Inherent Specification Distortion – Firstly would be understanding and emulating the standard or inherent distortion included in the basic specification for analogue tape recording (as I recall it was 3% THD at Zero VU or +4dbm)…it started there and became much more complicated (excluding the electronics or transport of the tape recorder itself). I can go on but if you visit the MRL site (makers of the best standard alignment tapes) you will find a magnificent library of technical papers discussing every single issue that contributes to the ‘analogue sound’ from azimuth, to bias, to pre-emphasis, to flux sensitivity to the effects of the scrape filters on the sonics of recoding. All of it well over my head, but in the hands of Waves engineers this information is a gold mine! The URL is http://home.comcast.net/~mrltapes/
Magnetic tape recording was a compromise and a magnificent one at that. Fully understanding how these compromises were manipulated buy some of the best minds in the industry, thus creating artefacts of their own, all in the quest of making something impossible work, is the key to the successful emulation of this technology.
• Intentional Abuse of the Standard – Second, and perhaps this is what quite a few engineers mean when the talk about this sound so fondly, was what happened when the basic standard was abused, driving the tape into saturation and compression (a bit like the ‘all buttons in mode’ on a UA 1176 Limiter). This effect is the natural characteristic of analogue tape carried to an extreme, and in some cases to an outrageous extreme. This ‘effect’ primarily found favour in the hands of engineers and producers making Rock and Roll, especially some of the ‘heavier’ forms of this musical idiom. So how do you emulate this effect…well first you must capture the ‘standard sound’ then just wind it up to various extremes. Turn up the input and turn down the output to compensate. Obviously Digital doesn’t want to be overloaded, but that’s not really an issue here, as you are really dealing with the internal behaviour of the modelling, say such as really winding up the input of a compressor to make it ‘pump’. All doable by Waves engineers, I am confident.
Comments about Tape Noise
It is impossible for me to make a comprehensive contribution to this development without have a rather significant argument with myself regarding the pros and cons of tape noise, coupled with the thermal noise inherent in the electronics used in the classic analogue tape recording equipment. As this argument progresses you will see that I am of two mind. I must acknowledge that I carry a 40 plus year bias against tape hiss…which was the ‘bane’ of a long and prosperous career. Clearly with the suggested development it can not be as simple as a single control marked ‘analog’ that switches on the noise as the entire application is about modelling various forms and degrees of analogue recording artefacts, of which noise is only one small component.
The Case Against Tape Hiss – It feels as though I spent the better part of 30 years managing tape hiss. Whether it was through noise reduction (which, as it says, only reduces the noise) or spending much of my energy during the mix cuing tracks in and out to reduce noise. Additionally there was always noise from sources other than the tape recording process to manage (thermal noise in microphones, hum and buzzes in guitar amps, air conditioning, environmental noise such as rumble from passing trucks or subways, pre-amp noise in the consoles, patched in outboard gear) and on and on and on.
So clearly I’m not a great fan of noise in general. For my entire career and even to this day … noise is my enemy! I believe that music is ‘mystical’ and that any sound that is not part of the music tends to break the ‘suspension of disbelief’ created by the music. Certainly when I’ve attended live performances of classical music I question how either the players or the audience can tolerate all the noise in the auditorium (coughs, sneezes, chair squeaks, etc.). Minus extraneous noises (such as hiss, hum, surface noise, etc.) music should exist in an ethereal space which allows the mind and the soul to be free … which is where it belongs when in the presence of music. (OK…so I’m a little eccentric on this topic…I’ll accept that. It doesn’t, however, preclude me from being right!)
For me digital technology was a God send and all of a sudden most of the forms of noise I had been battling for years were gone. The CD, whatever it’s limitations, eliminated the dreadful surface noise of the vinyl record (plastic molecular reform, scratches, ticks and pops, etc.). Digital Audio Workstations with their almost non-existent noise (in a practical sense) has always made me wonder why anyone would leave the analog button on with of Waves vintage plugs. Others I’ve talked to, like myself, upon opening one of these plugs head directly for the ‘analog’ button and switch it off, scratching our heads and wondering why it’s always defaulted to ‘on’. We’ve talked behind your back (including with some of your own team) that if one does a mix and uses a number of these plugs with the ‘analog’ switched on, it’s not long before the accumulation of noise becomes unbearable. Once again, the focus becomes the noise, not the music … which is, of course, completely backwards!
I would like to restate my very strong opinion that one of the key misunderstandings in our industry, especially amongst my younger colleagues,
is that the ‘analogue sound’ equates to noise. It does not … noise is the problem, not the solution!
The Case for Tape Hiss – So after that rant, here is the other side of the story. Although, perhaps impossible to prove, there could be a modulation of the signal by the tape hiss that contributes to the sound of analogue magnetic tape recording. I wonder if this effect can be quantified? I strongly suspect that anyone who took the absolute position I stated in the previous section during product development and interface design could be ‘throwing the baby out with the bathwater’.
The Best of Both Worlds – Perhaps it’s as simple as giving the user a bit more control over the noise beyond a simple on/off button:
• Possibly a On/Off button preceded by a volume control for the noise only.
• A selection of the octave of the tape hiss (remembering that 15ips hiss is an octave higher than 7 1/2ips hiss and that 30ips hiss is an octave higher than 15ips hiss). I am not suggesting that this be automatic based on the recording speed selected on the GUI by the user (as is the case with other tape modeling products), as this recording speed selection has a massive impact on the ‘sound’ totally independent of the quality of the tape hiss.
• Also one should remember that ‘hiss’ is not just a high frequency ‘white noise’. Hiss or tape noise is quite a complex broadband signal.
Another consideration would be to have Off-On / Selector controls for:
- Tape Hiss/Noise (with a volume control)
– The selectable thermal noise of valve tape electronics or the thermal noise of solid state tape electronics completely separate from the Hiss/Noise controls.
– Perhaps one also wants to have control over some of the other tape noise components such as the bias ‘rocks’. Bias rocks were usually a by-product of problematic tape formulations. It was a sound, below the level of the tape noise rather like large soft rocks rolling down a dirt hill. To make sure you had everything right, on a roll by roll basis, one would do a bias adjustment, frequency alignment and then short the input of the recorder (dead patch), wind up the monitor gain and make sure the tape noise was a steady state sound. If you had bias rocks the only option was to reject that roll of tape and start over again.
Many of these are the things that I don’t miss for one single heartbeat when working in the Digital Environment.
I hope you will find this of use. I suspect it is way over the top for what you expected and perhaps even for what you require. I also suspect that much of this information is already known by your team … but that said, I simply can not assume anything when it is my intention to offer you every little bit of knowledge I can in the hope that you will take what you need and discard that which you either …
• already know
• don’t agree with
• or find of no use for your current development
… all in the hope that perhaps you will find a few ‘gems’ that will contribute to this much needed development. At the very least this has been a charming, of not somewhat labor intensive, wander down memory lane.