We dive deeper into the theory and practice of hybrid arrays, and consider the benefits of using more than one array.
In the previous part of this series, I introduced the concept of hybrid stereo mic arrays, and explored some of the initial practical processes you can undertake to choose either an established named array — all of which are defined precisely — or to adapt the idea to tailor your own to the recording task in hand. This time, I want to dive a little deeper into the subject, and we’ll start with some useful tools that can inform your choice.
Visualisation
In Part 4, I mentioned the advantages of visualisation tools for optimising the SRA and capsule configurations, and the increased parameter interaction when tweaking hybrid arrays means good visualisation tools become even more helpful. However, there’s another way of looking at these things that I’ve found very educational and informative: The Stereophonic Zoom. This was the title of an academic paper presented by Michael Williams to the AES Convention in London in 1984, and it has been enhanced by its author ever since, culminating in the excellent Multichannel Microphone Array Design book that I reviewed a couple of years ago. That book covers a far wider range of mic configurations and applications than we’ve been exploring here, though, so perhaps an easier introduction to the concepts can be found in an article on Grace Design’s website: www.gracedesign.com/support/StereoZoom10.pdf.
Michael has, very generously, also published the associated graphs for a range of stereo arrays using different polar patterns on his website (www.williamsmmad.com/MAD/2ch/2ch.htm). I’ve redrawn one of his graphs for cardioid capsules configured in a stereo array. The capsule spacing is shown along the bottom axis and the mutual angle along the vertical axis. The blue lines indicate the SRAs provided by different combinations of capsule spacing and mutual angle, while the orange blobs identify the parameters employed by the named hybrid arrays discussed previously.
The capsule spacing and mutual angle of various ‘name’ hybrid arrays. The shaded areas (top and bottom) indicate regions where the direct/reverb ratio varies across the stereo image.
As you can see, the X‑Y coincident cardioid array, with zero capsule spacing and a mutual angle of 90 degrees, lies close to the blue 180‑degree SRA curve. The DIN array, with its 20cm capsule spacing and 90‑degree mutual angle, is offset to the right of the X‑Y array, sitting close to the blue 100‑degree SRA curve. Hopefully, examining this graph will help to make a little clearer how reducing the capsule spacing moves horizontally leftwards across the blue lines of increasing SRA and, similarly, how reducing the mutual angle moves vertically downwards across the increasing SRA lines. The shaded areas at the top and bottom indicate the regions where the direct/reverb ratio varies across the stereo image. In the upper area there tends to be too much reverberation in the central zone of the stereo image, and the direct sound tends towards a ‘hole‑in‑the‑middle’ effect. In the lower shaded area, the reverberation tends to increase and ‘pool’ towards the outer extremities of the stereo image. So, choosing parameters in these areas is not recommended.
Angular Distortion
The red lines, roughly perpendicular to the blue SRA curves, represent the ‘angular distortion’ created by the array, indicated here as a percentage error from the true position. Angular distortion, in this context, is the variance between the physical angles between sources in front of the microphones, and their perceived relative angles when replayed over loudspeakers. It’s analogous to the optical distortion created by a wide‑angle camera lens when it’s placed close to a subject.
As a general rule, most stereo mic arrays tend to expand the soundstage outwards — they push sources located away from the centre of the stereo image even further out towards the speakers, thus stretching the centre and ‘crushing’ the extremities. It’s usually a pretty subtle effect, but it’s certainly audible when comparing different array configurations in good monitoring conditions. It can also become noticeable when close‑miking a soloist who plays in an animated way, as any small physical movements can sound rather exaggerated when auditioned on the speakers.
The optimum conditions to minimise angular distortion vary with different polar patterns in the array, but for cardioid microphones the angular distortion is minimised by using moderate mutual angles (80‑110 degrees) and capsule spacings (20‑30 cm). In other words, hybrid arrays produce slightly less angular distortion than coincident arrays, and almost half as much angular distortion as spaced arrays. So it seems that capturing both time and intensity differences between channels is better than just time or intensity alone.
An illustration of angular distortion: sources further from the centre tend to be pushed outward towards the speakers.
As can be seen on the graph, the DIN array has the least angular distortion of all the hybrid arrays I’ve discussed so far, with NOS and ORTF both having a little more angular distortion. Not surprisingly, the coincident cardioids and Gerzon array have considerably more. In practice, this angular distortion is rarely even noticed, let alone problematic, but it’s still interesting to note how changing parameters affects this aspect of stereo‑imaging accuracy. This angular distortion is also illustrated in the Sengpiel Visualisation website mentioned in previous articles (https://sengpielaudio.com/HejiaE.htm). The five equally spaced coloured radial marks within the shaded SRA represent equidistant sound sources, and are replicated on a line between the two loudspeakers above, their spacing indicating the angular distortion created by the selected mic array configuration.
Combination Arrays
Spaciousness in a stereo recording is always a very attractive and desirable feature, but as I hope has become clear, achieving a pleasing level of spaciousness will often compromise the stability and precision of the stereo imaging. Wouldn’t it be nice if there was a way to achieve the best of both worlds? Well, there is... but it requires additional microphones and might make the stereo purists go weak at the knees!
The main risk, when adding extra microphones to an array, is the introduction of large phase differences between channels.
The main risk, when adding extra microphones to an array, is the introduction of large phase differences between channels, which can then cause unwanted colorations (comb‑filtering effects) when summed to mono. But if the level of the extra microphones is kept low (around 8dB below the main mics) those phase differences are unlikely to cause major problems.
So, why not rig a main pair using a near‑spaced hybrid stereo array known for reasonably accurate stereo imaging — ORTF, for example — and supplement the spaciousness by adding more widely‑spaced omnidirectional mics a little lower in the mix? The use of omnidirectional capsules also has the potential advantage of improving the low‑frequency extension, making up for the known weakness in that area of cardioid microphones. Once again, there are many variations on this core theme, but a particularly popular configuration has become known as the OCCO array. The letters indicate the layout of two pairs each of omni and cardioid capsules. I’ve occasionally seen it called the ‘Optimal Cardioid Capsule Configuration’, although that seems rather lacking given the array requires omni capsules too!
I’ve never seen a precise definition of the OCCO array, but most configurations employ the cardioids in a near‑spaced format, usually ORTF (but occasionally DIN or NOS) at the centre, with the outer omni capsules usually pointing outwards at around 45 degrees and spaced as wide as the shared mounting bar allows (I typically use 660‑690 mm). APE spheres on the omnidirectional mics can be useful, if available, to enhance their stereo imaging definition (as in the Decca Tree arrangement — see the separate box).
Top: a typical OCCO array, with an ORTF stereo array complemented by omni outriggers; and bottom: a Faulkner ‘Phased Array’, with two pairs of mics mounted on the same bar and pointing 45 degrees left/right.
Typically, the ORTF (or similar) cardioid pair would be considered the main stereo array, and its position and settings would be optimised and balanced accordingly — but ‘erring’ towards a slightly dry balance, as the omni outriggers will make the mix more reverberant. The outer omnis are then blended in to the mix, usually 8‑10 dB below the ORTF pair, to add the required spaciousness, ambience and LF extension. An alternative but equally valid approach is to consider the spaced omnis as the main stereo array, optimised for perspective and imaging in the usual way. The central cardioids can then be blended in just to add a little focus and positional definition to the mix. Again, to avoid the risk of coloration in a summed mono mix, keep the second pair 8‑10 dB lower in the mix than the main omnis.
Tony Faulkner, who I’ve mentioned previously in this series, has taken this idea further with an arrangement he calls a ‘Phased Array’ — a term borrowed from sophisticated techniques employed with multiple radio antennae, as well as in military radar and sonar systems. The idea is that by controlling the relative phase relationships between transducers, it’s possible to ‘steer’ the directionality or focus of the entire array, although I’m not convinced that’s really what is going on in this application.
The particular configuration Tony suggests, which he has used for several orchestral recordings, places all four mics on the same bar, angled 45 degrees left and right as shown in the diagram on the first page of this article, with the outer omnis spaced at 670mm and the inner cardioids at 410mm. With such a wide spacing, the SRA of the cardioids is only around 67 degrees, while the omnis have an SRA of around 100 degrees. Essentially the omnis act as the main stereo array, with the ability to blend the cardioids in to add focus and clarity to the centre of the orchestra as a kind of audio zoom control. Like the OCCO arrangement, this kind of multi‑mic configuration is neat and quick to rig — it requires only one mic stand — but it affords a useful degree of flexibility in the output sound character, in terms of perspective, spaciousness and imaging definition.
Separated Arrays
Of course, there’s no rule that insists the omnis and cardioids must be mounted on the same bar; that’s just a convenience. Often the near‑spaced array and the spaced omnis are deliberately separated, with one pair placed in front of the other (usually at the same height). This requires a more substantial and sophisticated mounting rig, but it’s a worthwhile investment because it opens up a whole new range of options.
There are multiple solutions to the mounting mechanics, but my preference is to use a pair of Manfrotto 154b mic bars. The first bar is mounted facing fore‑aft on a tall Manfrotto light stand (1004BAC), allowing a stable 3.6 metres of height (when weighted adequately with sandbags!). The second bar is suspended from one end of the first, facing left‑right with the omni mics near each end. At the other end of the fore‑aft bar, I hang a short stereo bar for the near‑spaced cardioids. The fore‑aft bar is then adjusted to optimise the balance, keeping the centre of gravity of the whole assembly directly over the mic‑stand centre. The Manfrotto mic bar is 650mm long, which is enough for most spaced‑omni setups, but the supplied tubes can be easily swapped out for longer 25mm aluminium or carbon‑fibre tubes should greater mic spacing be required. The same idea can be used to build cost‑effective and secure Decca Tree configurations (see the separate box).
An example of an array using separated spaced‑omni and cardioid arrays is with the omnis spaced 600‑700 mm apart and positioned around 600mm in front of the near‑spaced pair (which could be in ORTF or DIN configuration). The idea here is to match the perspectives captured by each mic array — the near‑spaced cardioid pair inherently captures less reverb than the spaced omnis, so needs to be mounted at a greater distance from the source to have a similar amount of reverberation. Like the ‘Phased Array’, the omnis would typically be employed as the primary pair for their spacious and natural sound character with extended low‑frequency reach, and the near‑spaced cardioids blended in to supplement focus and imaging stability.
Here, spaced omnis and near‑spaced cardioids are combined: the omnis offer a spacious, natural sound character and extended LF response, while the cardioids can be blended in for greater focus and image stability.
An equally valid yet reversed approach is to mount the near‑spaced cardioids at the front, closer to the source, with the spaced omnis around 600mm (or more) behind. In this case the near‑spaced cardioids provide the main stereo sound and the omnis are blended in to give more room ambience. This configuration has the advantage that the increased separation of the omnis behind the cardioids reduces coherence between the two stereo pairs, allowing a higher level of omnis in the mix before coloration occurs in a mono sum.
Next Month
In Part 8 of this series I’ll look at some of the post‑production processing options that can be used to further enhance or correct the stereo imaging obtained by the myriad coincident, spaced, hybrid and separated stereo arrays I’ve described throughout this series.
The Decca Tree
When discussing the concepts and applications of spaced omnidirectional microphones in this and the previous article in this series, I have deliberately avoided all mention of the Decca Tree array — even though it employs spaced omnidirectional microphones. The reason is that the true Decca Tree arrangement is actually a lot more complex than many realise, and requires specialist omnidirectional microphones fitted with Acoustic Pressure Equalisation spheres to work properly. The technique really deserves an entire article all of its own, but if you are interested in learning more about it I can highly recommend the book Classical Recording: A Practical Guide In The Decca Tradition by Haigh, Dunkerly and Rogers.
A full Decca Tree setup, with a central three‑mic tree joined by a pair of omni outriggers on either side, and a sixth mic placed above the double basses.
Most think of the Decca Tree as a simple three‑mic array but, when employed for large orchestral recordings it’s really a complicated six‑mic array. The familiar three‑mic tree is suspended above the conductor, but it also requires two additional omni outriggers positioned left and right (between the second and third desks of violins and cellos), plus a sixth mic positioned above the double basses. That combination, along with the specific microphones used, delivers the famous Decca Tree’s sound character.
The dimensions for the central ‘tree’ vary depending on the situation, but for a typical orchestral setup the left and right mics are generally spaced around 1300‑1400 mm apart (although I have seen wider spacing in some commercial recordings), with the centre mic placed forward by half of the left‑right mic spacing.
For a small ensemble, like a string quartet, a simpler three‑mic tree can be employed, with the left‑right mic spacing reduced to around 600mm, and the centre mic around 150‑200 mm forward. However, I’ve often found this mini‑tree approach less pleasing than other hybrid arrays for recording small ensembles.
Mounting the centre mic forward of the left and right pair is a very clever design feature of the Decca Tree. Spaced omni microphones can easily form a hole in the centre of the stereo image if spaced too widely, and the centre mic neatly fills that potential hole in the sound coverage. But more than that, by being slightly forward of the other two it captures sound fractionally earlier by a millisecond or two, greatly enhancing the stability of the stereo image because the reproduction over loudspeakers builds the stereo image from the centre outwards.
Traditionally, the omni mics employed in a Decca Tree are Neumann M50s, which house the omni capsule within a 40mm sphere, and this is critical to the performance of the array. The sphere not only boosts the microphone’s high‑frequency response but also narrows the directivity at high frequencies. So, these omni’s are directional, and this is the reason why the omni mics are angled left/right at roughly 45 degrees from the centre. In the six‑mic array, the outriggers are also angled slightly outwards by about 15 degrees, and this capsule angling is critically important in generating a coherent stereo image.
Most small‑diaphragm omni microphones can be fitted with APEs — either as manufacturer accessories or as 3D‑printed options — and I strongly recommend this approach if you don’t have access to M50s or equivalents. At a push, subcardioids or even cardioids can be used instead, maintaining the capsule spacing and HF directionality, but at the cost of the LF extension and phase linearity associated with omnis.