Home -> Magazines -> Issues -> Articles in this issue -> View
Digital Overdubbing | |
Article from Home & Studio Recording, December 1986 |
Mike Skeet explains how to bring sound on sound recording into the 20th century using two video recorders and a Sony PCM processor.
Folk legend of the development of recording techniques always seems to include references to Sergeant Pepper being done on 4-track and that Les Paul and Mary Ford did their earlier work in stereo between analogue tape machines. Now these old techniques are being used with new technology.
This article is about digital multi-dubbing (hereafter referred to as DMD) as a recording technique copying the Les Paul system but naturally omitting the deterioration in quality that is an unavoidable feature of analogue tape overdubbing. Apart from the usual studio set-up of mixer, instruments, synths, drum machine, effects, decent level metering, monitoring and headphone foldback, all one needs is a single digital processor (such as one of the Sony F1/701/601/501 range) and two video tape recorders.
Just in case any of you are already starting to raise suspicious eyebrows, I had better establish my credentials. I usually operate as a classical music location recording engineer, but at home, I like to employ the F1 gear in 'pop' productions. I wouldn't want a 16- or 24-track set-up as it would never earn its keep. Hopefully the F1 is paid for! In my case the mixer is DIY and in a sense has been developed with DMD in mind. There's no reason, however, why a commercial mixer will not do the job if there are sufficient aux sends and returns and a suitable line input. In fact, with my other hat on, I was associated with the use of a Soundcraft 200 on some classical digital multi-dub sessions with vocal group Electric Phoenix which has now come out on EMI! Here with three passes through the system, four singers became 12 and for another piece seven became 21.
Figure 1 needs to be studied for the basics to be appreciated. Imagine that VTR1 is playing some backing already recorded. Its Video Out lead is connected to the Video In lead of the PCM processor. (For our purposes the word digital and the abbreviation PCM mean the same thing.) This will be processed by the D to A part, will come out of the stereo audio sockets and is routed to line input channels on the mixer. The mixer channel design must be of the highest quality, preferably without tone controls, noise producing mismatches or poor signal path design. It must have a genuine flat frequency response between 20Hz to 20kHz; the digital recorder has, so we mustn't introduce any weak links into the chain.
The mixer output splits three ways. One must be the studio monitoring of which a lot could be said. Suffice to say that in my case, it is Quad ESL63 Electrostatics driven by Quad 405:2. These are backed up by headphones: usually Sennheiser, HD420 or HD540.
Secondly a decent level monitoring system is required. But surely you will think the PCM has its own meters (which it does) and that in the end it is they which must show any overload. There's a snag though. The PCM's bar graph shows its output signal level and thus would not show the new record level! To complicate matters a little further, the 'over' lights would show overloads caused by the input signal as they are part of the A to D! The additional metering must be peak reading. VUs would be Virtually Useless.
Still with Figure 1, the final mixer output is the really essential one! It's to the line input of the PCM processor feeding the A to D part. Coming from the processor's Copy Out socket we have the video lead to VTR2's input. With levels set as will be described next, playing VTR1 and recording on VTR2 will give a one to one transfer in analogue through the mixer. On the way new material can be overdubbed. Of course, should the take not be good enough one simply rewinds both machines and has another go. The monitoring, visually and audibly will relate to the new composite signal.
With the cleanest possible path through the mixer (and this cannot be overstressed) up to ten or more passes are possible! With the albums I'm recording we average five passes. Invariably the Roland TR707 drums and bass are laid down first and there is really no noticeable deterioration with more than five passes. This with all the A to D and D to A and their attendant brickwall filters which all the non-practising armchair critics of digital so readily condemn!
Considering that we separately mix all the TR707's outputs, bass, synths, guitars, vocals apart from reverb and DDL effects, we would regularly exceed the capabilities of a conventional 24-track studio.
To get sufficient level on playback from the PCM processor it's best to use the headphone output. This is some 10dB higher in level than the line output proper. The record level controls on the processor need to be set so that the processor has just below unity gain, from input to Headphone Output. A crude but effective way of setting up for this is to temporarily couple the Copy Out to VTR1 Video In and to couple the headphone Line Out to the Line In. Initially do this with the gain controls at zero. Then, one at a time, advance each slowly and carefully until the inevitable positive feedback occurs. The bar graph will fly into 'over' at the point sought. Back off a little to kill the feedback. Around 2.3 on an F1's rotary knobs is the point, if mine is any guide to go by. Mark both controls for future setting. As feedback only builds up when the gain exceeds unity, this method is pretty accurate.
Now to do something similar with the mixer's line in to out circuit. Reconnect the processor's audio leads as shown in Figure 1. Leave both video leads connected to VTR1. The mixer's Line Input to output fader should be marked at a point around 3dB below its maximum travel, and this will be the normal working point. The channel's gain controls should then be set so that the unity gain point is achieved with the main gain control fader at the mark. Back off slightly as before. Then reconnect VTR2 as in Figure 1.
All this means is that it should be possible now to transfer a signal at very near unity gain from VTR1 to VTR2, but with the ability to increase or decrease the level on a given pass or passes by a few dB or fade the link up or down. The level of the transfer can be checked by recording a tone at -20dB on the processor's scale and noting its level after transfer.
But what about the quality of the mixer's line input? In my DIY mixer's case there are only two 5532 Virtual Earth mixer ICs in the link, no tone controls, and the channel gains are passive presets before the channel fader. Having tone controls in the circuit but set to zero is no guarantee that they are not seriously affecting frequency performance as well as causing all sorts of phase corruption. Additionally there's no phase inversion through the channel, nor through the PCM processor if one uses the Headphone Output.
One of the most useful things one can construct to aid digital multi-dubbing, and for that matter any other uses of a PCM processor and two VTRs, is a 6-pushbutton 9-facility switch box. It avoids continual repatching to get another facility, with the chance of getting it wrong and it can all be installed in a die-cast box. In my case it's fitted into the custom case of VTR2 and connected to the other VTR and processor by a standard DIN to DIN cable.
The arrangement is shown in Figure 2 and should be self-explanatory. The six pushbuttons are locking with the latch being released by the next button pressed. Latching the middle two allows recording on both VTRs simultaneously. The Copy Out socket is used on the processor instead of the Video Out as this allows automatic selection of digital transfer between VTRs when the Copy switch is operated. With the Copy switch set to normal the Copy Out and video out sockets are simply paralleled.
With this facility in the system, it's easy to set up for the first recording to be made on VTR1 and then bounced onto VTR2 and then back again and so on. I use a couple of L125 30-minute tapes for this. When all overdubs are complete, I then digitally copy onto a 'master' storage tape. Whatever facility is needed, it's all instant with a switch box like the one described.
At the time of writing this piece, I had just completed an album with Prime Factor, (a synth and drum machine band comprising Andrew Hope and Lee Sauli). This is to be the fourth album to be produced by DMD. They appear on Nakamichi BX300 duplicated real time cassettes in Dolby B and C forms.
As can be seen from the photograph, I had my DIY custom mixer with associated Surrey Electronics PPM and The Box soundstage analyser. The mixer uses eight channels of outboard graphic EQ. These excellent and most cost-effective Tandy 7-band units are switchable into eight of the mixer's input channels. I much prefer a graphic to any number of conventional tone control knobs. Seven bands are quite satisfactory, but for around £20 more per 'stereo' unit, Tandy have 10-band models which offer more flexibility. One of my EQ channels can be switched into the reverb send to make this facility even more versatile.
For these sessions I had moved from a Great British Spring reverb, which had served me well for years, to the marvellous Alesis Midiverb. This is quite an exceptional development and no doubt I'll follow it with a Midifex after more plumbing changes within the mixer to accommodate it. I feed it with a mono send and have a dedicated stereo return with pan pots to vary width and position.
There are four channels of Yamaha GC2020 comp/limiting available and these are patched as required via the mixer's prefade insert points. No need real for limiting or compression with synths or drum machine sounds but the GC2020s are invaluable with guitars and vocals. It's only necessary to limit synths if the sounds use high resonance settings as large peak levels can occur unexpectedly. But the GC2020s have a bonus: one noise gate apiece! I use these if I find any standing noise from any of the sources. Some of the synths and some drum machine feeds were slightly noisy in 'digital' terms, but the GC2020s' gates took care of this.
As I mentioned earlier, with DMD one is essentially choreographing the sounds in stereo at each overdub. Therefore you need a plan showing what's likely to happen at each pass and how many in total there are likely to be. I do like to keep the number down. I try to avoid having to go all the way through a piece just to put a cymbal crash on the end! I ask the musicians to plan the take so that they use their resources to make the most of the dub, it won't be the first time that a vocal part is added along with the guitar from the bedroom 'studio' interrupted by a dash down to catch the right place to add something from a synth. Remember that you can't do drop-ins like you can with multitrack.
There are several ways of ensuring that there's some stereo spread or other stereo effect. A couple of the synths at these sessions had stereo outputs and I find these simulations very effective. The drum machine outputs are separately panned, EQd and reverbed although I usually leave the toms in the TR707's own stereo feed, the reason is simple reason really: shortage of channels! What I need now is a way of further treating some drum machine sounds and the mono synth outputs. Dave Simpson of Thatched Cottage Audio tipped me off that the Effectron II DDL would do what I wanted for delay, chorus, flanging and so on and, instinctively trusting the amiable, bearded fellow, I immediately reached for my wallet. Two were purchased and the mixer once again set upon with soldering iron until it sprouted more post-fade auxiliary sends and dedicated returns with pan pots and reverb sends. This way I can separately and differently treat two different sources at the same time or really have a go on one sound with both Effectrons.
There's another way of treating sounds which I often employ when the musicians have footpedal devices such as chorus or phaser units. We DI before and after the device via a couple of DIY Sowter transformer DI boxes and suitably pan, EQ and reverb the outputs to a simulated stereo. Yet another way to spread or add character is to couple synths together and pan appropriately; for example at these sessions a MIDI link sometimes joined a Yamaha DX21 to a Korg DW6000 and Roland's own trigger links were used between the TR707/TR606 and the CSQ600 and SH101.
It should be quite obvious at this stage that the balance has to be created as one goes along. There's no postponing the artistic decision to one big mixing session, as conventional multitracking would allow! Interestingly Andrew and Lee in common with many other home musician/recordists are very used to working in mono via their 4-tracks and doing a lot of bouncing and overdubbing yet keeping their mix in a valid balance.
If any readers want to start digital multi-dubbing I can offer the tip that the new layer should always be at a lower level than one tends to start off with when setting up and rehearsing. The big danger is, of course, that the earlier layers get covered up. On this project all seemed to work well and the drum/bass sequence which was put down first is still solid enough in the final mix.
If readers want to sample an album of DMD, the author can supply the Prime Factor real time cassette album DMD104 at a price of £4. He would welcome any comments by phone or letter. Please state Dolby B or C. Readers with a PCM F1 or the like can have a digital copy of the production master if they send a Beta tape and £6 to cover the cost of copying and P+P.
All this is obtainable from: Whitetower Records, (Contact Details).
Basic Multitrack (Part 1) |
First Take |
At Home in the Studio - Living Dangerously with 'Doors' |
Choosing and Using: Portastudios |
Sound FX with the Uher CR240 |
Bits 'n' Pieces - An Introduction to Digital Audio (Part 1) |
Before You Start Making That Demo Tape - Demo Tape Discipline |
Willow - The Recording of a Film Soundtrack Music |
Rhythm methods - Drum recording guide |
Oddballs & Out Takes - Studio tales |
Is Analogue Multitrack Recording Dead? |
Hands On: Casio DA7 DAT Recorder |
Browse by Topic:
Feature by Mike Skeet
Previous article in this issue:
Next article in this issue:
mu:zines is the result of thousands of hours of effort, and will require many thousands more going forward to reach our goals of getting all this content online.
If you value this resource, you can support this project - it really helps!
New issues that have been donated or scanned for us this month.
All donations and support are gratefully appreciated - thank you.
Do you have any of these magazine issues?
If so, and you can donate, lend or scan them to help complete our archive, please get in touch via the Contribute page - thanks!