Ghost In The Machine
Soundcraft Spirit Automated Mixer
What do you give a desk which already has everything? Comprehensive MIDI mute automation supported by a Steinberg software package would be pretty good for starters. Dave Lockwood undergoes a spiritual awakening with the new Spirit Auto.
Soundcraft's Spirit Studio range was the company's first attempt to tap into the ever-growing market for quality equipment aimed specifically at the small private studio operator. It was, in many ways, a remarkable design effort, resulting in a compact desk that offered all the best features we traditionally associate with Soundcraft — a conspicuously clean signal path, excellent build quality, and clear, logical operating procedures — at a price that tackled even the budget mixer manufacturers head-on. Opinion was almost unanimous: 'A great desk; if only they had put MIDI muting on it!'. Indeed, the Spirit undoubtedly lost sales to inferior competitors because of the omission of MIDI muting. Soundcraft's argument on this was that they felt MIDI muting was only ever going to be a short-term, stop-gap feature, prior to the advent of affordable 'real' mixing automation systems — and they were right.
The Spirit Auto uses dbx VCAs for level automation, and you can automate the desk either via a Steinberg program (which you must buy separately — it does not come free with the Spirit Auto) for Atari ST, or by recording standard MIDI controller messages into a sequencer. (JL Cooper will soon be offering Softmix as a dedicated Spirit automation program for the Mac.) With the dedicated Steinberg software, data conforms to a proprietary Soundcraft/Steinberg 8-bit format (using SysEx messages), thus doubling resolution that would be available with ordinary controllers.
Retaining the option to utilise the standard protocol where necessary, however, was a smart move, allowing the automation features of the Spirit to be employed in a more limited system, with, say, only a 1MB Atari available. The dedicated automation program can be run concurrently with Cubase, under M-ROS, but you would need at least a 2MB system, with a key expander (the automation program is dongle-protected), and multi-port MIDI distribution, since SysEx format data or dense controller information is best kept separate from note data.
Little is visibly changed on the Spirit chassis, apart from the addition of a Snapshot switch; all communication beyond normal operation of the desk's controls takes place via the Atari's mouse and keyboard. The dedicated Steinberg program has a particularly graphic and informative display, making this a very comfortable system to work with. It's almost a musician's rather than an engineer's automation system, having much in common with MIDI sequencing in terms of its operating and editing procedures. Cubase users will particularly appreciate the consistency of the interface, but I suspect almost anyone could feel at home with the package in virtually no time at all.
The automated components in the system are the channel faders and both the channel and monitor mutes. The notable absentee in that line-up is, of course, the monitor level control, apparently an option rejected primarily on economic grounds.
Much as I want to dive straight into the interesting bits, it would perhaps make more sense to first recap briefly on the desk itself, especially for those entirely unfamiliar with the system — if you already know your way round the Spirit, then pass on to the 'Automation' section of this review.
The Spirit Auto is an 8-buss design with in-line monitoring, with direct outputs for each channel to expand the number of tracks that can be simultaneously addressed. With the extensive EQ and aux. facilities available to the monitors, they can function almost as full-featured additional channels. The system is not expandable; a straightforward choice of 16 or 24-channel frame sizes is offered. With four well-equipped dedicated stereo effects returns also incorporated, a nominally 24-input console can effectively handle 56 inputs, all with a worthwhile degree of control.
The Spirit's control surface is not sloped, but the desk is compact enough for everything to remain easily visible and within reach. Despite its unusually shallow steel chassis, enhanced by distinctive moulded end-caps, the Spirit is no lightweight. The whole thing is immensely solid, with no sign of flexing in the chassis or deflection of the panels under pressure, auguring well for long-term reliability. Finish and styling conform to the Soundcraft 'new look' of pale grey with colourful control knob pointers, extending up the side and across the top of the knob, giving the clearest possible indication of position. The switches are not quite so clear, being all grey, with tiny white legending on the panel.
All connectors are mounted in the same plane as the controls, at the top of the operating surface, making them both accessible and visible, but perhaps also meaning that a permanent installation will look a little more untidy than would otherwise be necessary. Unfortunately (unlike Allen & Heath), Soundcraft have not seen fit to respond to suggestions to move the insert point to the edge of the connector group, where it could be accessed most easily. If you're not planning to use a patchbay, the present location of the insert requires you to stand up whenever you need to patch anything, just so that you can see what you are doing.
The XLR mic inputs use Soundcraft's pad-less, electronically balanced circuit, with individually switchable 48V DC phantom power. The Line input is via balanced quarter-inch jack. Channel inserts utilise the standard tip-send, ring-return stereo jack configuration, and channel connections are completed by further quarter-inch jacks designated Track Send and Return. Track Sends pick up either the group outputs, repeating in blocks of eight across the desk, or their own channel's Direct output. Track Returns receive the signal returning from the tape machine, if there is one, but can also function as additional line inputs. All other connections are via balanced quarter-inch jacks — except for the MIDI sockets, rather oddly located on the front edge of the console, right up against the operator. True, the connectors don't project beyond the front overhang of the 'arm rest', but you still have to find a neat way of dressing the cables away to their destination. I suspect they're located there simply because there just wasn't anywhere else (without using the rear panel, which would obviously negate one of the fundamental design concepts). The MIDI connectors are accompanied by a DIP switch panel and 16-way rotary switch to set system defaults (MIDI channel, data protocol, and so on).
A compact external power supply (non racking), connects via a decent length of multicore, hard-wired at the PSU end, mating with the chassis via a screw-down multi-pin. The PSU incorporates the system's only power On/Off switch, and supply-monitoring LEDs for the +17/-17V rails, as well as the 48V phantom supply.
The Spirit follows the modern convention in providing two separate signal paths through each channel, with the option to split facilities between them. The primary path receives input from either the Mic or Line socket, governed by the Gain pot (10 to 60dB) with the Line switch dropping input sensitivity by 20dB. The secondary, monitor path features a centre-detented -10/+20dB Trim pot, feeding a rotary level pot and pan control.
The 4-band EQ has fixed-frequency, shelving HF and LF bands, offering 15dB of cut or boost at 10kHz and 100Hz, plus two sweep mid bands. The whole lot can be operated entirely within the channel path, or reconfigured into two independent 2-band systems, with the sweep section remaining with the channel, whilst the shelving filters are transferred to the monitor path by activating the 'EQ to Monitor' switch.
The 'split EQ' configuration requires the range of the sweep mids frequency controls to be more extensive than in many a 4-band configuration, for effectively the mids must function adequately as a 2-band EQ on their own — Lo Mid covers 50Hz-800Hz, whilst Hi Mid handles 500Hz-15kHz. The flexibility is there to allow you to use the full 4-band setup on input signals going to tape, leaving the monitors all flat, but then split the EQ on mixdown to maximise the number of EQ-able inputs. Although you can't switch the sweep section into the monitor path, you can reverse the input path on every channel individually, via Channel/Monitor Input Reverse. This also places the monitor signal on the long fader, as opposed to the small rotary level control of the secondary path.
EQ In/Out switches remain a glaring omission on the Spirit — EQ gain controls, however, are centre-detented, and the circuit topology is optimised to make this condition a virtual bypass. Soundcraft EQs usually sound nice, and the Spirit's is no exception. You can use plenty of it without making your sounds too coloured or unnatural, and it can be as selective as any EQ without a dedicated 'Q' (bandwidth) control can be expected to be.
The feed for the channel insert point sits between the two EQ stages (post fixed, pre sweep), allowing both the insert send and return to be treated separately when the EQ is fully active within the channel path, or giving a choice of pre or post-treatment EQ, according to which section you use. There are no insert points on the secondary signal path.
The auxiliaries are similarly partially transferrable between channels and monitors to maximise flexibility. Of the six sends provided, there are two post-fade effects sends and one pre-fade foldback send dedicated to each signal path. However, the effects sends from the monitor path (3 and 4) can be re-deployed to the channel path, leaving just foldback on the monitors, whilst improving the channel lineup to four separate effects sends plus foldback.
The Spirit remains limited in its aux-linking facilities — only the two foldback busses can be linked via on-board switching, so to allow both monitors and channels to access the same effects (in aux-split mode), you will have to resort to the usual external paralleling trick (most units have 'stereo' inputs which are summed to feed the processing stage, offering an easy way for two auxs to feed the same effect).
Two effects sends per input (in 'split' mode), is not really generous by present standards, but there has been no upgrade from the original Spirit Studio to incorporate pre/post-fade switching on the foldback sends, enabling them to be used for effects on mixdown. If you envisage running the system largely with all facilities operated within the channels, however, four postfade sends is a reasonable allocation.
The ability to feed the selected control room monitor source into the foldback mix, via dedicated level pot, is a welcome facility. This allows a decent foldback mix to be created almost instantly, rather than having to recreate the whole balance on the individual foldback pots. The two foldbacks and four effects auxs each have master controls, with AFL monitor switching, but no automated mutes.
The monitor level and pan controls separate the auxiliary section from the main channel Pan, which functions in conjunction with the channel routing matrix (four odd/even group pairs plus mix buss). Almost inevitably in a desk of this size, this is located alongside the top half of the fader slot. Both monitor and channel path have a basic PFL facility plus momentary-action mute switch, with associated green LED. PFL switches have red LEDs which function also as Peak warning displays, indicating a level within 4dB of clipping. Very smooth, long throw (100mm) faders, with the usual 10dB of gain above the unity position, complete the Spirit's channel facilities.
Sub-Groups/tape-sends and the stereo buss are governed by long-throw faders, but with the unity point right at the top of their travel, increasing resolution. Groups incorporate PFL switching, and may also be routed directly to the mix buss for normal (non VCA) audio sub-grouping. Adjacent odd/even pairs can feed the mix bus in stereo, or a Mono switch can be activated on any pair.
Four dedicated stereo effects returns are incorporated, with routing direct to the mix buss, or via groups — a full routing matrix is not provided, however, and returns can only access a specific pair of groups, restricting the ability to lay effects to tape without repatching. Ganged controls are used, with a centre-detented Trim facility (-10dB to +20dB) for gain matching, and a 2-band shelving EQ (+/-15dB @100Hz/10kHz). Level and Pan (stereo balance when both inputs are used) controls, plus feeds to both foldback busses, round off the facilities.
The Spirit incorporates an integral electret Talkback mic (non-latching) with level control, routable to foldbacks, tape-sends, or both. Control room monitor source, governed by a master Level control with Mono switch, may be the stereo bus or a single two-track return circuit.
PFL/AFL monitor level trim is provided, and metering for the eight groups and stereo buss is via 10 16-element, green-amber-red, LED arrays. Meter operation may be internally selected to either a peak-reading or averaging, VU-type response, and can be calibrated via a set of small trim-pots. An on-board oscillator (1 kHz or 10kHz) with level control completes a comprehensive set of master facilities for a desk of these dimensions.
Steinberg's software for the Spirit Auto is a visual delight. It seems to be an entirely new program, and in no way a revamped version of any of the systems they supply to other manufacturers. There are just two screens: the main, graphic fader screen, and a new, beautifully implemented, edit display. In the 16-channel version of the program, the main page simultaneously displays all 16 channel faders, plus both channel and monitor mutes, with eight sub-group faders also shown. In the 24-channel version all input channels are shown simultaneously, but they can be scrolled aside to access the subgroups. A typical array of Steinberg transport controls, locators, and positional displays occupies the strip across the bottom of the screen.
Recording mutes is simply a matter of setting the relevant channels to Write mode, putting the software into Record, and then activating Play from whatever is the master transport in the system — all SMPTE frame-rates and MIDI Timecode are recognised. Any fader and mute activity on channels in Write will recorded against timecode, with the graphical representation of the fader knob on-screen mirroring your movements as they occur. In Write mode, the position of the fader directly determines the level of the VCA. Entering Read mode on the written channels and replaying the same section of tape will reproduce the same mutes and fades that you recorded; however, the fader knob icons will remain stationary whilst an 'equals' sign graphic moves up and down the fader slot to indicate the actual VCA level being read.
"Soundcraft's Spirit Auto is a fine example of designing to a price without unduly compromising on the essentials of performance."
The three principal operating modes are Write, Read, and Update (equivalent to 'Trim' in some automation systems), the latter reading all the moves made to date, but applying an offset according to how the fader is moved. A third fader graphic is provided to help here, indicating the last recorded VCA level (ie. the one you are offsetting from). With this much going on, it certainly had the potential to look a bit of a mess, but the graphics are beautifully executed, and it all works a treat. The purpose of Update is, for example, to cover a situation where you have written a series of fader rides to level a vocal track, but then wish to make the whole part louder or quieter, whilst retaining all the previous moves. One tends to use Update more and more, rather than re-writing sections, towards the final version of a mix. An important 'Auto Write Off After Record' option was a late addition to the software (which was still being revised right up to the deadline for the review); this automatically selects all channels safely to Read after a Write pass.
The remaining modes are Local — in which data is simply passed through the program and echoed back to the console, operating the faders and switches, without affecting the recorded information — and Auto. Auto lets you switch faders into Write mode automatically, simply by moving them 'through' the present recorded value as it is being read back, enabling a seamless takeover from the existing data.
To further assist with smooth transitions, the usual 'ramping' options are offered, allowing the user to set preferred times over which the system will crossfade from previously recorded values to new values, when going from Read into Write, and vice-versa. Simultaneous switching of specific groups of faders and mutes is facilitated by the provision of a Record Mode Memory, implemented via the keyboard space-bar, which allows you to toggle between two setups. The programmable Punch In/Out facility works exactly as in Cubase, with the additional option of selecting it to automate the Record Mode Memory Toggle instead. Neat.
The eight Sub-group faders on-screen have no connection with the desk's eight 'group output' faders. The screen faders represent the internal VCA sub-group controls. Channels assigned to a sub-group will have their VCA levels amended by the operation of that fader, rather like the Update mode, but potentially operating on a number of channels at the same time. All the individual faders will retain their own 'internal' moves, but their actual VCA levels will be added to or subtracted from by the action of their assigned group fader. The on-screen group faders can only be moved by mouse, but operate under the same modes as channel faders; however, they also offer a memory-saving opportunity, since recording group data does not actually alter the data of all the contributing channels — removing a channel from its group will always restore it to its previous level. Switches follow the group assignment.
As an alternative to screen-grouping, it is possible to define any normal channel fader as a Master Channel. Other channels selected to that master group will be affected by operation of their master fader, just like the software groups. The purpose is much the same, to control a group of signals whilst maintaining their relative, internal, balance. In this mode, data is actually written for each contributing fader.
The edit screen is a model of economy and clarity, presenting the information in a manner that aids intuitive understanding, even for the novice operator. At last, we have 'fader movement' (actually VCA level) shown as a solid block in the vertical axis, depicted against time in the horizontal axis. Channels are stacked vertically (up to 11 faders and mutes on screen at the same time), and you can zoom in and out in either dimension, to maximise either time or level resolution on any individual channel. Clicking the left mouse-button and dragging over an area of fader data will define it by shading, and you can then cut and paste the section anywhere within the same or another track. Using the keyboard Delete key when an area is shaded activates 'Relative Delete', in which the level is maintained throughout the section at whatever it was at the start of the selected area.
Using the right mouse-button activates a Steinberg toolbox, offering pencil, rubber and divider tools. With these you can redraw or erase sections of the mix, whilst the dividers allow straight lines to be easily drawn between defined points. It is all so simple to use. If you do go wrong, the Undo key will get you back to where you started before your last alteration. You can even 'undo' the 'undo', to restore the edit. This procedure doesn't just apply to edits — if you don't like what you have just recorded in a mixing pass, you can restore the previous version via Undo. This is a vast improvement over having to remember to hit the 'F' key before you stop the tape, as in other Steinberg automation systems.
Finally, if the combined efforts of the Auto and Ramp facilities and close attention with the toolbox have all failed to give you what you want, there is always the Smooth facility. Activating Smooth will even out any remaining abrupt level changes within an area that has been shaded.
If you're wondering whether the presence of VCAs in the signal path has degraded audio performance to any noticeable extent, rest assured that the Spirit Auto displays the same high-quality sound on which the excellent reputation of the range was originally built. The desk is still arguably the most impressive performer in the budget sector, with low noise, tons of headroom, and well-controlled crosstalk in all critical circuits.
In use, the Steinberg automation impinges to the minimum degree upon the intuitive processes of mixing. It is not a complex system, and yet it is comprehensive; nothing of any significance is beyond its abilities. Above all, however, I must praise the system for its ease of use. I was working with beta-test software and without documentation for most of the review period, and yet managed to feel comfortable with it in no time at all. Anyone with previous experience of Cubase, and the benefit of a manual, would probably feel like a 'power user' in about 10 minutes!
It is the Edit display that sets this program apart; indeed I ended up using this screen for recording too, once I had learned the keyboard control routines. Fader data is displayed in real-time as it is recorded, and it is often possible to go straight into action with the toolbox to correct little discrepancies that you were aware of during the pass, without even wasting time bothering to listen to a playback. From a single edit display, the program manages to reconcile the differing requirements of both editing in fine detail, and offering a good overview, with powerful manipulation of mix structure. Musicians think of songs in terms of parts — verses, choruses, and so on — so it makes sense to allow them to handle mix data in the same way.
Remarkably, despite the software's continuing updates during the review period, I encountered no operational problems whatsoever; the program has a robust feel about it which inspires confidence. One slight niggle is that when functions such as global fader status selection can be accessed from single keys, it seems slightly perverse that Edit has to be selected via Control-E, with exit to faders again via Control-F. Why not simply use the 'e' key alone to toggle between the two screens?
File handling offers no unnecessary complications, with just one file type, the Mix file (*.SMX), encompassing both setup (ramp preferences, track names, and so on) and actual fader and mute mixing information.
Automating the Spirit solely via MIDI data works rather well, and offers a significantly more economical route — you don't have to buy the Steinberg software — to genuine dynamic mix automation for those already equipped with a suitable MIDI sequencer. Resolution is obviously limited to that allowed by the 7-bit Control Change data format, but I couldn't actually generate any audible glitching in any remotely realistic mixing situation. An Autochase facility is incorporated within the Spirit, which automatically sends a snapshot of all the automated switches and faders every two to three seconds. Thus, by recording the chase data, even sequencers that do not have their own 'chase controllers' facility will never be more than a couple of seconds from being up to date, wherever they are re-started.
Ease of use in this configuration obviously depends on the abilities of the particular sequencer employed. My tests were conducted using 'the big two' — Steinberg Cubase and C-Lab Notator, and I suspect there is nothing to significantly better either of these as an interface, on the Atari platform at least. Writing and reading basic data is obviously not a problem; it is in the editing of the data that you might expect limitations to surface. However, both sequencers provide all the processing tools necessary to enable you to do more or less everything that you can do in the dedicated automation program.
A MIDI Manager template for Cubase was supplied with the pre-production review model, offering very graphic representations of the Spirit's automated switches and faders (presumably this template will also accompany production models). Snapshots and fader grouping are also supported. For the C-Lab user, this can be matched by the facilities of the Realtime MIDI Generator screen. Although RMG controller information is normally channel based (each fader represents the same type of data, but on a different channel), you can get round this by using P-User definitions, which can be unique to each control element. Here too, snapshots and grouping are possible. The P-User RMG screen elements will only transmit, of course, unless a Realtime Transform is set up to convert incoming controller data to the the relevant P-User events.
Even better is C-Lab's Hyper-Edit screen. Substituting faders and switches for Hyper-Edit's 'Instruments' gives a display almost on a par with the dedicated program. In fact, it looks remarkably similar!
Handling the streams of controller information that can be be generated by a dynamic automation system, alongside note data, is always less than ideal, so a multi-port MIDI system is probably advisable for operation in this configuration. The Snapshot facility should be used as much as possible, in both MIDI and proprietary protocol modes, to conserve memory and bandwidth. Whichever system you are using, the best mixing procedure probably consists of setting up a fader balance and mute configuration for each section of your track, and then storing these as snapshots to be recalled at the appropriate points. Data-intensive fader movements can then be confined to important fine-tuning operations.
This system works beautifully, and proved a pleasure to use throughout the test period. Reservations will undoubtedly centre on the decision not to automate the monitor level controls and the master fader. It is understandable that cost considerations dictated the former, but the latter I find more difficult to justify. I agree that there would have been a very small performance compromise involved, and yet I still think that most users would have preferred the option of master fader automation. In typical usage of the Spirit Auto (if there is such a thing), probably involving mixing tape sources via the channels and MIDI sources via the monitors, it is easy to see how standard Controller 7 MIDI mixing could substitute for fully automated monitors. There is no real substitute, however, for an automated master fader — fading the source channels (which could easily be done via the VCA sub-grouping), whilst leaving the output wide open, is not really good engineering practice. The problem is compounded when you have a compressor across the mix bus; instead of fading the compressed mix, you merely fade the input to the compressor, which produces an audibly different result. The design team should think again about this one, and at least offer automated master faders as an option.
Soundcraft's Spirit Auto is a fine example of designing to a price without unduly compromising on the essentials of performance. The original Spirit concept seems to have been to make a Soundcraft-quality console available in the budget market. Subsequent competing designs, specifically pitched against it, may have had an apparent edge in facilities — more auxiliaries, dedicated monitor EQ, expandability — yet the Spirit stood on its own merits. It was, and is, all about quality, and you either bought that side of the argument, or you didn't. This model will not be without direct competitors for very long, but they will be faced with the same problem — the Spirit Auto is going to be a hard act to follow.
Spirit Auto 16:8:2 £3,172.50 inc VAT.
Spirit Auto 24:8:2 £4,641.25 inc VAT.
Spirit Studio 16:8:2 £2,120.08 inc VAT.
Spirit Studio 24:8:2 £3,119.63 inc VAT.
Spirit Live 4 12(+2):4:2 £1,173.83 inc VAT.
Spirit Live 4 16(+4):4:2 £1,219.65 inc VAT.
Spirit Live 4 24(+4):4:2 £1,768.38 inc VAT.
Spirit Live 4 32(+4):4:2 £2,642.58 inc VAT.
Steinberg Spirit Automation software £399 inc VAT.
Soundcraft Electronics, (Contact Details).
Review by Dave Lockwood
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!