The tonewheel organ is back in vogue of late, aided by the various recent 'virtual' renditions of the real thing, and Hammond's release of the New B3. But how faithful does the software sound? We look at two contenders...
Back in the mists of time, in the days before MIDI, a Hammond tonewheel organ did keyboard duties for many a serious rock band. Notable users included Tom Coster of Santana, Keith Emerson of Emerson Lake and Palmer, and Jon Lord of Deep Purple. In jazz, the Hammond was (and still is) cooler than cool; a versatile instrument capable of producing walking bass lines, riffing chords and searing solos.
Over an admittedly much shorter time period, Native Instruments' B4 Hammond Organ VST instrument has also become something of a classic amongst software instruments, a staple plug-in which provides a sound still essential to many styles of music, both contemporary and retro. A more recent addition to Emagic's stable of virtual instruments is their answer to B4, EVB3. In order to assess the merits of these clones compared to the 'real thing', I took a trip to see a Hammond enthusiast, and made sure that I packed a Power Mac G4 loaded with software.
The purpose of my quest was not to review the simulations (all have been previously reviewed in SOS — EVB3 in SOS February 2003, available at www.soundonsound.com/sos/feb03/articles/emagicev.asp,and B4 in SOS November 2000, or at www.soundonsound.com/sos/nov00/articles/native.htm), but to compare the implementation of some of the important characteristics of a Hammond's sound and their use in performance. The ability to control the various parameters of a Hammond's 'control surface' while playing is of paramount importance if an authentic performance is to be achieved — more on this point later in this article.
Having just completed a PhD in Jazz, part of which involved in-depth analysis of the work of Jimmy Smith (possibly jazz's most famous Hammond devotee), Tim Dean-Smith was just the man to help dissect the minutiae of these software wannabes. It's fair to say that Tim is a Hammond nut (he would regard the description as a compliment). As a founder member of jazz group d'Organ Trio, he is the owner of a Hammond XK2 (Hammond's own most recent digital organ) and a T200, which is a smaller sibling (or 'spinet', as Hammond called them) of the mighty B3, the model that B4 and EVB3 seek to emulate. In the studio he plays the XK2 through a Leslie 122XB rotary cabinet, while using a Peavey keyboard amp with a Motion Sound Pro 3T rotating horn cab on top for live use. He has also played many B3s live, as well as A100s (known inexplicably as the poor man's B3, even though they include a built-in speaker system that the B3 lacks).
"For years I tried to get the best sound I could from a Yamaha DX7, TX81Z and an Alesis Quadraverb, but nothing was quite like playing the real thing. Reliable B3s are rare and very heavy for a non-roadie act to manage, however, so although I use a tonewheel spinet at home for some recording — a T200 — I take my XK2 out live with the Motion Sound rotating horn to simulate the real Leslie cab I have in my studio. Nowadays Hammond's digital B3 clones are very good indeed."
Unlike the multitude of analogue synths which have been modelled in plug-in form over recent years, the character of a Hammond organ demands that the designer of a software emulation spend some serious time on reproducing the inherent design deficiencies of the original instrument. If you've followed recent instalments of Gordon Reid's Synth Secrets series, you may be familiar with some of these, but if not, here's a quick recap.
The principle (similar to that of a pipe organ) is that timbres are constructed by adding harmonics to a fundamental in varying amounts. The tones are produced by rotating irregularly shaped metal wheels in an electromagnetic field and the output from these 'tonewheels' is mixed together using drawbars (sliders) which are marked in 'feet', like the stops on pipe organs: 16', 5 1/3', 8', 4', 2 2/3', 2', 1 3/5', 1 1/3', 1'. The chief difference between pipe organs and Hammonds is that the air-driven instrument only has fixed-stop relationships, that is, the pipes are either 'on' or 'off'.
If you've already read this month's review of Camel Audio's Cameleon, this might be sounding strangely familiar, and it's not surprising, as this method of sound creation is basically additive synthesis. On the Hammond, had it not been for some of the considerations listed below, the tonewheel sound-generation system might have proved to be a fairly limited and unmusical method of timbre creation. Indeed, the basic, unadorned sound of the tonewheel generator, which Gordon Reid spent two months at the end of last year analysing in Synth Secrets (see www.soundonsound.com/sos/nov03/articles/synthsecrets.htm and www.soundonsound.com/sos/dec03/articles/synthsecrets.htm) seems to our ears like a rather static, uninteresting tone. And yet it seems that this is what Laurens Hammond, the creator of the tonewheel organ, wanted from his invention — he apparently regarded it as a 'pure' sound. He hated Leslie rotary-speaker cabinets (made by a separate company) because they altered this tone — he wanted you to buy a Hammond 'Tone Cabinet' which was an amp and speaker in a box, like a Leslie, but without the rotation. Leslies were in fact banned from Hammond dealers' shops until Laurens Hammond's death in 1953.
In trying to reproduce the sound of a Hammond, one of the first considerations is that the basic electromechanical tone-generation system has a sound all of its own; the 'pure' sound that Laurens Hammond was striving for was 'let down' by limitations of the design and manufacturing process, and what he regarded as many 'imperfections' crept into the sound. Secondly, Hammond himself added modifiers to the sound, such as the famous Chorus/Vibrato (of which more later). And thirdly, most Hammond players use a Leslie cabinet as an amp (B3s need some kind of amplification... they have no built-in speakers). As Gordon has also shown, this makes a huge difference to the sound, and most players consider it part of the Hammond experience — even a model like the A100, which has its own built-in amplification, is usually also connected to a Leslie by choice. In short, the 'imperfections' in the sound of a real Hammond are responsible for much of what gives it its charm, and a decent software emulation has to make a good job of capturing these quirks.
The initial impression when opening EVB3 for the first time is that no parameter has been left unturned when designing the interface. There are no keys on the plug-in window as there are with B4 and with the virtual cover 'open' or 'closed', there is no attempt at an authentic look. This may be a facile point, but B4 does retain some of the visual feel of the B3 it is modelling, down to the rounded and bevelled drawbar control knobs, and the radio-style knobs in control view, where EVB3 has sliders with a numerical readout. Looks aside, the choice of colours for the legends on the EVB3 does make it difficult to read at certain screen angles, especially with non-resizable windows.
We looked first at how these software instruments fare at reproducing changes to the B3's timbre as a result of imperfections in design and construction (and the effects of ageing on components). Key-click, for example, is a percussive noise at the transient point of the tone, and is sometimes confused with Percussion (of which more in a moment). In fact, it stems from the design of the keyboard mechanism, and often gets worse as a Hammond ages and its key contacts corrode over time. As the Hammond organ developed, the designers got better at eradicating what they saw as an unwelcome intrusion into their emulation of the pipe organ, so the older a Hammond, the more apparent its key-click, and the more random its nature. EVB3 reflects this with sliders for Click Min and Click Max (between which extremes the click length varies randomly), Click Colour, and rotary controls for click volume — one for Click on, one for Click Off and one for Pedal Click. B4 only allows for the control of the overall key-click amount via a rotary control. The XK2 meanwhile, has 'soft', 'normal' and 'loud' options (which correspond to key-click samples) for the whole keyboard, but Tim was not bothered by this restriction. "I tend to use loud all the time! The XK2 works by rotating a list of sampled key-click noises, which creates a very good impression of the real thing, and I never find this limited list of samples a distraction when I'm playing."
Due to the electromechanical (as opposed to electronic) nature of the original Hammond's construction, many of the components interact with each other in a way that would not be possible if the design were based on PCBs. 'Leakage' is an example of this, and is associated with the fact that all tonewheels are active simultaneously, whether or not you are playing the notes associated with those tonewheels. Furthermore, although the drawbars act as attenuators, they are unable to completely silence the crosstalk between the continually active tonewheels and the output circuitry. All this crosstalk adds a mixture of noise and tonewheel output to the sound of the organ, giving it a slightly 'breathy' quality. This is present in the sampled sounds of the XK2 and B4 's modelled tones but in neither is it controllable, whereas dedicated sliders in the EVB3 allow you to determine just how 'poorly constructed' or 'worn out' your virtual Hammond is. But that's not all: EVB3 also includes sliders labelled Random FM (which aims to emulate modulation caused by irregularities in the running gear for each shaft), Filter Age (which recreates the change in centre frequency with age of the band-pass filters associated with each tonewheel generator), and Warmth, the euphemistic parameter that attempts to model the pitch fluctuation in a Hammond's sound caused by the actual variation in tonewheel rotation speed.
None of these is controllable in either B4 or the XK2, so the Hammond fan who simply must control every aspect of the sound of the instrument could make a clear choice here.
Modelling Built-in B3 Features
As already mentioned, aside from these quirks in the sound of the Hammond resulting from design limitations, there are also those aspects of the sound that Laurens Hammond intended should be there! One of these is the sound arising from use of the B3's built-in 'Scanner' vibrato system. Creating an effect which should not be confused with that created by the rotating horns of a Leslie cabinet, scanner vibrato was produced by scanning an analogue delay line consisting of several low-pass filters with a multi-pole capacitor. V1, V2 and V3 on the switch represent the three stages of vibrato, where only the delay line signal is heard. Settings C1, C2 and C3 are three stages of chorus, which was achieved by mixing the delay line signal with the original signal. From the B3 service manual: "The scanner vibrato varies the frequencies of all tones by continuously shifting their phase. It includes a phase-shift network or electrical time-delay line, composed of a number of low-pass filter sections, and a capacity-type pickup, which is motor-driven so that it scans back and forth along the line".
On EVB3 these settings are combined with a Rate parameter, which controls the speed of the Vibrato (where the presets control the Intensity), and a Chorus parameter, which controls the mix of dry and vibrato signal.
As you might imagine, on those original Hammonds with this feature, the chorus settings give rise to a boost in volume, which, although it is technically an imperfection, should be simulated if true authenticity is required in performance and recording, and is very accurately simulated on Hammond's recent hardware-based XK2 digital tonewheel organ. While B4 accurately represents the switching on a B3, where both the lower and upper manual vibrato each have an on/off rocker switch to accompany the six position rotary Vibrato/Chorus select, EVB3 has a seven-position dial, with 'CO' being the extra setting. When 'CO' is selected, Chorus and Vibrato are disabled but the slight volume boost is maintained. To accompany this, there are two On/Off switches for Vibrato/Chorus on EVB3 — marked 'Upper' and 'Lower'.
On a real B3, the Vibrato settings V1, V2 and V3 sound fairly 'cheesy' to modern ears, although they were advanced technological developments for their day. Tim's Hammond XK2 does a fairly good imitation of this if that is your bent, while the impression left by EVB3 is that, although there is a lot more control available, the depth of signal modulation in the original is just not quite available. There were two further comments on EVB3. Firstly, Tim felt that the original Hammond settings should be available as presets to which you can return at the click of a switch after fiddling with the Rate and Chorus sliders. Secondly, Tim noticed that the CO setting on the 'Preset' dial was only accessible anti-clockwise. Presumably this is to reflect the fact that it would be controlled by a fader, not a rotary control when connected to a control surface such as Logic Control.
NI's B4 also allows variation on the switch settings via the mix and depth controls in its Control view, but the presets are marked on each dial's continuum. Sound-wise, B4 was able to ladle on more molten Gruyère than even the XK2. I favour this approach since I have always thought that the parameters of modelled instruments should be capable of extending beyond the ranges of the original instrument — it's the Spinal Tap-like need in me to 'go up to 11', although I'm aware that not everyone agrees on grounds of authenticity.
Percussion & Loudness Robbing
Percussion does not refer to a drum machine-like capability, nor to the drum-like noises available as presets on some Hammond organs of the smaller 'spinet' type, but to the addition of a harmonic of the fundamental tone controlled by a short envelope, which has the effect of adding a woody 'plink' type attack to the note. On a real B3, there is a rocker switch for selecting either the octave (second) harmonic or the 'octave and fifth' harmonic (the third harmonic); there are also rocker switches for selecting 'soft' or 'loud' (maximum envelope amplitude) and 'fast' or 'slow' (envelope decay time).
Hammond's XK2 copies these facilities with the strange complication that, by default, it routes the percussion through any selected chorus, which blurs its effect. This routing can be changed, but only if you also disable the internal Leslie simulation, which means that you need to be using an external Leslie such as the 122XB to achieve that effect.
B4 's top panel only offers a single on/off switch for percussion, but there are rotary controls on the Control view for volume, decay and harmonic. As before, the preset values are marked on the continuum, with the option to go above or below the minimum/maximum settings of the B3. The equivalent to the B3's rocker switches on Emagic's EVB3 are the 'On/Off' and '3rd' switches. Like the XK2, EVB3 (but not B4) allows the selection of both the second and third harmonic, a setting which is not available on the B3. Time (envelope decay) and Volume are available on the top right of the control panel as sliders alongside Velocity and Up Level, which enables the user to define the relative volume of the upper manual, simulating the B3's characteristic of reducing the volume of non-percussive registers slightly when percussion is engaged. EVB3 allows the user to make the percussion velocity sensitive which, although it gives some creativity to the user, can lead to a less authentic sound, when this parameter is overlooked. None of these sliders have preset positions marked on their travels.
In use, once the inconsistencies of Chorus and velocity had been ironed out of the percussion settings for the XK2 and EVB3 respectively, all three simulations added a decent, authentic-sounding percussion sound to the attack phase of the timbre. In terms of authenticity, Tim felt that the B4 just had the edge (literally) over the other two in that the percussion integrated with and augmented the sound to a more convincing degree. Despite the number of parameters available for tweaking, again the EVB3 was disappointing for the sonic restraint it showed at the top end of these parameter values.
A bizarre feature of the Percussion switch on a real B3 is that it disables the ninth drawbar (1') so switching it on has a very noticeable effect on the basic timbre if that drawbar is in use. Performers use this feature not just to add percussion but also to change the sound on the fly. Neither EVB3 nor B4 recreate this effect, perhaps because it is a design quirk of the original, although in some ways the omission from the software instruments is odd, as their attention to authenticity is very close in most cases. Interestingly, Hammond's XK2 does offer this feature, although you can also choose to switch it off (ie. have both Percussion and the ninth drawbar if you are so inclined).
If there were no circuit to prevent it, opening more drawbars would result in an increase in volume. On the original B3, Hammond added a technology called 'Loudness Robbing' to prevent this, which seems perfectly simulated (but not controllable) in both the XK2 and B4. EVB3 also offers a fader for this although even when set to maximum, this effect is not as noticeable in the modelled sound as perhaps it should be. It could be that the designers of this instrument modelled their version on a real B3 whose loudness robbing circuitry was not as effective as on others, and it is possible that the subjective differences already seen in a number of areas might be explained by this.
The Sound Of The Leslie
As befits this most important part of the sound of the original Hammond, the Leslie speaker is simulated pretty convincingly in both EVB3 and B4, albeit with some important distinctions. As in most other areas, EVB3 offers some fairly arcane options, allowing you to specify motor control type and cabinet design, but it also sports one important Leslie feature absent from the main control panel of the B4, the brake (although it is accessible via controllers — more on this point later). I'm not sure why Native Instruments have left this off the main plug-in window — many players such as Joey DeFrancesco use the brake for performance effects on several numbers in their set — and lots of 1960s rock organists did the same.
B4 does have a 'rocker' switch for rotor On/Off, but this has a slightly different effect since the braking of the rotating speaker takes a discernible amount of time, and has its own unique sound. This is also true of the time it takes the speaker to reach full speed from the braked position.
Where B4 scores is in the presentation of the sound of both speakers. In a Leslie cabinet (such as the model 122 or 147) both the upper horn and low-frequency unit rotate, and they do so at different speeds and rates of change due to their different masses. B4 gives you speed control over both, where the EVB3 only lets you control the faster speed of what seems to be a single virtual rotating unit The latter does have a slider to allow the user to control the rate of acceleration and deceleration, but again, it is only for a single unit.
Also, B4 allows control over two virtual microphones to the extent that the sound of each speaker can be directed to each side of a stereo field. EVB3 appears to offer but a single 'microphone' which can be angled. The effect resembles that of similar features in the amp simulators found in many guitar effects.
Tim, however, felt that the mic-placement simulations in both B4 and EVB3 sounded too clean. There is no room ambience or rotor noise, both of which are the usual result of placing a mic in front of a real Leslie cabinet.
As mentioned earlier, Hammond's XK2 has its own internal Leslie simulation, but there have been gripes about its authenticity. As Tim explains, "The XK2 was designed to be used with an external Leslie, and it sounds great through my 122XB. Mind you, a Yamaha DX7 also sounds pretty good through one of these!".
There is a major difference in the layout and operation of the drawbars in each of the software simulations. B4 has the upper manual drawbars to the right of the lower manual ones, which I would have guessed was the right way round, but, in fact, the opposite orientation, as on EVB3, is more authentic. The drawbar banks, from left to right, are: Upper, Pedal and Lower, as on the B3 (and indeed most Hammond organs). Both have nine drawbars per manual, where the B3 has two sets of nine per manual.
I hooked up Emagic's Logic Control (since Logic has to be the host software for EVB3) so that Tim would have real-time access to the plug-ins' playback parameters. If you've ever seen an experienced Hammond player at work, you will realise that performing on the instrument is made up of much more than just playing the keys — frequent movement of individual (and sets of) drawbars, as well as 'toggling' of Leslie, percussion, vibrato and chorus settings contributes to a constantly evolving sound.
Obviously, since EVB3 is a Logic Instrument, every single parameter is represented on the Logic Control surface, giving an unwieldy 16 virtual 'pages' of fader data, whereas B4 required only eight. This is not a criticism in itself, but leads to some coincidental operational differences between the two instruments.
Emagic's technical support informed me that there has been some discussion about writing a custom layout for EVB3 with Logic Control. As of this writing, however, the parameters are assigned to faders on a first-come first-served basis, which results in the upper manual drawbars being split across two pages of Logic Control. This makes hands-on drawbar abuse very tricky. Apparently the intended custom layout would have put all nine drawbars on one page utilising the master fader for the ninth drawbar — a far more usable proposition.
B4, by contrast and a happy coincidence, has the first (lowest) drawbar at the 'end' of one page and the other eight all on the next page. Since it is these higher eight (of the nine) which are in many cases most responsible for tonal variation this makes using the Logic Control to waggle B4 's drawbars a more realistic proposition. The first drawbar is still missing, however, so even this layout makes one timbral adjustment option difficult. Tim commented: "Removing the fundamental is really good for giving a transistor radio EQ to the organ, and being able to radically change the sound on the fly like this is a key part of my performance technique".
It's fair to say that this test revealed a limitation in using a single Logic Control for such a complex procedure for which it was never designed. I also tried out both B4 and EVB3 using my old Yamaha Promix 01 as a control surface (it has 19 assignable faders, enough to control the drawbars of both manuals simultaneously), and, although setting up the MIDI assignments is no small chore, the more rugged faders did feel a little more like the real thing compared to the silky smooth lightweight action of the Penny & Giles faders on the Logic Control.
Aside from this important consideration, EVB3 's parameters all responded smoothly and predictably to operation via Logic Control, whereas B4 often responded tonally to the movement of a hardware controller before the on-screen graphic reflected the change, or vice versa.
To a recording musician, the operation of drawbars on a Hammond is perhaps counter-intuitive. An increase in volume occurs when you move the drawbars towards you, the opposite of the action of a fader on a mixing desk. Logic Control is, of course, primarily designed as a mixing control surface, so confusion could (and did) arise here.
EVB3 's drawbars move 'away' with the Logic Control's faders, reducing the drawbar volume. B4 's drawbars did the opposite, ie. the volume increased as the Logic Control fader moved away (but of course the drawbar graphic moved in the opposite direction, reflecting what would happen on a real Hammond!). Changing from one to the other caused us much head-scratching, but those of you who have done more mixing than Hammond playing would adapt very quickly to either — unless, of course, you opt to have both Instruments in your system...
Automating EVB3 under Logic is, of course, very easy. Using Logic v5's track automation, every parameter is available, and the discrete state of the switches is clearly labelled and graphically defined. Not surprisingly, perhaps, B4 is less easy to control in Logic. For example, the two discrete states of some switches are not represented graphically by discrete levels on the track automation but by a continuous curve of controller values (see screenshot above). Also, as mentioned earlier, the Leslie brake facility missing from B4 's main control window is available via track automation. The value of 60 shown on B4 's track automation in the screenshot results in the Leslie being 'braked'. The screenshot also shows how Logic clearly displays and labels the states of the rotor speed switch for EVB3. Running B4 under Logic, it's up to the user to learn which value represents which state.
All the parameters of the XK2 (including all drawbar and switch commands) transmit and receive via MIDI. You can even control the Leslie from the keyboard in this way. The drawbars are also servo-assisted, so sequencer playback results in a flying-fader display, just like a modern control surface.
Other features mark the XK2 as the product of the company who originally built the B3. For example, the operation of the percussion (correctly) drops the drawbar volume by 3dB.The XK2 simulates the B3 by default, but can also provide good impressions of other, earlier Hammonds with three 'voicings' entitled B3, Mellow and Brite, and will also retune the keyboard to B3 tuning or equal if required.
Despite our intention to go right into the detail of each simulation, the overall feeling was that we had still only scratched the surface of the complexity of these instruments, particularly EVB3. As a 'studio' Hammond for the non-Hammond player, either EVB3 or B4 fills the role very competently, the former with its almost manic attention to detail, the latter with its immediacy of sound (albeit with slightly less control). The intention when using one of these instruments would have to be that the performance detail is added after the basic note-entry run-through, even if you own a control surface. However, from the perspective of experienced Hammond players who would want to create these aspects of their performance in one live pass, there is still no substitute for a 'real' control surface, that is, a keyboard with the full complement of drawbars and switches, all accessible at the flick of a hand, not the click of a mouse.