Ambisonics - BBC Soundfield Experience
Author's Note: This is probably
my first major article on Ambisonics, written for Studio Sound when I
was Editor of its sister publication, Sound International. With current
interest on surround techniques and the pros and cons of single-point mic systems,
the BBC's early experience of the Soundfield mic in stereo and in ambisonic
surround still provides interesting reading. Also of interest is the philosophy
taken by BBC engineers when it came to capturing a performance with a completely
new technique, in a location that the radio audience knew well, often by having
been there themselves or having heard many broadcasts made using traditional
techniques. Do bear in mind that references to "now" mean 1979!
This year, the BBC took delivery
of two Calrec Soundfield Mark III (production model) microphones. Prior to this,
however, the BBC had already assessed the prototype Mark II mic. Richard
Elen describes the BBC's experiences with this novel mic system, described
by some as "the most important mic development since Blumlein".
It is perhaps unfortunate that the
prototype Calrec soundfield mic didn't make its appearance on the BBC scene
until after the first run of matrix-H and HJ experimental surround-sound broadcasts
had been completed. It is hoped the transmissions in the latter format will
start again soon, and will no doubt make use of the soundfield mic, but in the
meantime, BBC studio managers and researchers are amassing a good deal of information
on the performance of the mic in stereo applications, which will obviously be
a great deal of use when surround broadcasts are continued.
The IBA has also completed its first
set of broadcast evaluations of the ambisonic technique, deciding that a 3-channel
ambisonic transmission format is to be preferred to a 2 or 2 1/2-channel format.
The 3-channel system has certain technical problems associated with it, but
these will no doubt be overcome, and with favorable reaction to ambisonics from
the EBU and FCC it may well be that this innovative British-designed system
will form the basis of the record and broadcast output of the future.
Of course, the technical difficulty
of transmitting or cutting three or more channels of information disappears
when we consider digital multiplex encode/decode techniques; even before this,
however, it is likely that analog methods will be devised to cope satisfactorily
with the three channels required for "horizontal" surround-sound, or four channels
needed for full "with-height" reproduction.
In the meantime, however, it's well
worth looking at the BBC's experience with the soundfield mic, because apart
from the UHJ surround-sound possibilities of this mic, it promises a great deal
in terms of stereo flexibility and response.
To find out more about the BBC's
work in this field, I discussed the system with Bob Harrison, from the Assistant
Chief Engineer's Department, Radio Broadcasting, and Antony Askew, who is a
Senior Music Studio Manager. Bob Harrison's role enables him to experiment a
fair amount with new techniques without being so far removed from actual operational
situations as to be unable to relate his findings to the limitations of broadcasting
requirements. Tony Askew, on the other hand, is frequently responsible for "driving"
the entire mixing system at major "serious music" events like the Promenade
Concerts, where the soundfield mic has been employed over the last two seasons.
Without doubt, the BBC has been in an excellent position to evaluate and apply
the soundfield mic "in the field". This process of extended and meticulous investigation
is still continuing, of course, but already some firm results have been emerging.
In fact, the story of the BBC's involvement
with mic techniques of the soundfield type goes back to 1976, when the late
Ben Bauer brought a prototype Ghent mic over from the States for the
BBC to examine. The Ghent mic is specifically designed to offer exact
encoding to the Columbia SQ "quadraphonic" standard, and part of the specification
includes the requirement for the mic to eliminate phase anomalies, particularly
in the front quadrant. Phase--and frequency response--anomalies are big problem
in mic systems which make use of Blumlein and other coincident pair techniques,
as used extensively for BBC music productions. With two discrete microphones
it is theoretically impossible for the pair to exhibit full phase coherence
over the entire audio range, resulting in problems with image location as the
sound wavelength approaches the distance between the capsules of the array.
The Ghent mic was designed
to correct certain anomalies--at least in the front quadrant--and thus it was
expected it would perform well in stereo configuration. This indeed proved to
be the case at an Albert Hall test: stereo imaging was excellent and the general
"cleanness" of the sound was very encouraging. At the very least it showed that
the idea of developing a mic without phase- and frequency-response anomalies
had a very good chance of success. However, the Ghent mic had the disadvantage
of offering no variation of Polar response characteristics; its results were
thus largely dependent on finding exactly the right position and orientation
in a hall, and this was very hard to obtain. But the system showed promise.
The arrival of a mark II prototype
soundfield mic around the middle of 1978 enabled the operational research in
this direction to proceed a good deal further. It was first utilized in the
recording of the last concert for the Cheltenham Festival in July 1978 with
the BBC Symphony Orchestra. The mic was positioned about eight to ten feet from
stage level at a point which provided a good musical balance of the orchestra,
and the derived stereo output from the controller unit was monitored.
At first hearing the results were
not very rewarding. The overall sound balance so close to the audience, on a
hot sticky July evening was dry and lifeless. Even so, the clarity and precision
of the stereo image produced by the mic was exceptional and definitely supported
the claims made for the mic. For the first time, the BBC had a mic available
which, firstly, had a flat response from 40 Hz to 10 kHz; secondly, all its
parameters were variable from a remote position.
Equalization in the mic produces
the effect of "true coincidence" of the capsules, thus eliminating phase anomalies,
and manipulation of the B-format output of the mic's matrixing network enables
the operator to adjust the effective physical orientation of the mic, including
pan, tilt and so on at the control unit, or--by recording the B-format ambisonic
signal on four tracks of a multitrack--alter these parameters after the event.
Truly an amazing facility.
The theory behind the mic also offered
the possibility that a single mic could be used to make perfect 2-channel phase-matrix
recordings of the matrix-HJ type--previously, a good matrix-HJ program had been
dependent to large extent on the manipulation and use of existing stereo techniques:
the problem being that it was important to offer a good stereo balance as well
as surround sound, and the interaction of conventional BBC stereo mic techniques
in a surround-plus-stereo configuration was difficult to predict or set up with
any degree of the necessary accuracy without a great deal of effort. The soundfield
mic, however, offered the promise of truly compatible surround and stereo decoding:
a strong necessity if the BBC experimentation with the UHJ part of the Ambisonic
hierarchy is to continue.
The soundfield prototype was then
installed at the Albert Hall for evaluation on a series of Promenade Concerts.
The version in use at this time--the Mark II prototype--was fitted with a Reading
University equalizer and power supply and a prototype Calrec control unit with
Reading electronics. Peter Fellgett and his technical designer Jeff Barton took
time off to make sure everything was working correctly and reliably. The microphone
was placed a little in front of the usual stereo main-pair position, about 15
to 20 feet up and about three or four feet in front of the fountain in the Albert
Hall, and it was here that the first major advantage of the Calrec mic to the
BBC was demonstrated.
Standard practice for rigging a stereo
pair at the Albert Hall requires the dropping of a line over each end of the
balcony to haul up a sling arrangement. The mic is centered in the sling and
the audio cable is run off along one side of the sling line to the sockets at
balcony level. Due to the weight and asymmetry of such an arrangement, it is
often the case that the mic array twists as it is raised, necessitating at least
half an hour of raising, lowering and correction of position in addition to
the hour or more required to set up the sling in the first place. The soundfield,
however, took a mere 20 minutes to rig, dangling on its cable from a tightly-stretched
sling. Any change of mic attitude could be corrected with ease by adjusting
the B-format attitude knobs on the control unit. The BBC's current method of
slinging the mic also includes a draw string to the back of the arena, with
the main sling a little further forward than before: this enables front-to-back
movement of the mic's physical position to be carried out.
At this stage the mic was, of course,
used in stereo only: the derived stereo signal from the control unit was brought
up on the mixing consoles as an external feed, which could be compared in an
A-B fashion with the "conventional" mics and incorporated into the main output
if the Studio Manager performing the concert mix so desired. During last year's
season, the mic's output was available about 60 to 70 percent of the time and
was often used in the main mix.
Before discussing the results of
this particular series of tests, it is important that we take a brief look at
the standard BBC approach to miking up a concert of this type, and the basic
premises upon which the "sound" of a BBC "serious music " broadcast are based.
BBC stereo techniques are currently largely reliant on the Blumlein or coincident
pair arrangement, often assembled from discrete microphones. Over the years,
BBC sound balancers have built up what we might term and an "acoustic blueprint
" of the type of "sound" a music broadcast should have. This attitude, of course,
is one that has produced superlative results, and continues to do so. It does
mean, however, that impressions of an entirely new system will be based on what
such a broadcast usually sounds like: this will modify sound balancers' opinions,
which are, of course, highly subjective. A further consideration is the fact
that venues like the Albert Hall are quite well known to many radio listeners:
they will have attended concerts themselves and will have a fairly good impression
of the sort to sound they ought to be hearing. They're also used to BBC broadcasts,
and the broadcasts themselves must create the best possible illusion of the
sound of the Hall. Indeed, Proms are one of the few examples of a situation
in which the listener is actually aware of what the "original sound" would be,
as opposed to a studio recording or broadcast.
The " broadcast illusion" is to a
large extent enhanced by actual limitations and "faults" in conventional
mic design. With a standard set up which depends on one or two main stereo pairs
backed up with ambience and solo mics, methods of compensating for the massive
change of the Albert Hall acoustic between rehearsal and performance are well-known
to BBC sound balancers. This change must not be underestimated. Seven thousand
people in the Hall have a tendency to remove any ambience picked up naturally
by the main pair: ambience mics, with their attendant "muddying" of the overall
sound, must be introduced at relatively high levels into the mix to create the
right sound. In these conditions, such effects as "lobing" on the main pair--where
the response of the mics at higher frequencies is rather more directional and
peaky--tend to make the resulting output more satisfying to the listener. For
example, "lobing" tends to move the string section outwards from its actual
position in the sound-stage to produce a more impressive stereo spread. HF "presence"
provided by the complex phase cancellations that occur when several stereo pairs
are superimposed also tend to brighten up the sound and make it more lively.
These effects are almost entirely the result of manufacturers' attempts to compensate
for the anomalies in stereo mic design inherent in variable polar diagram capacitor
mic types--methods varying from one maker to another--and as a result of the
anomalies themselves. The difference with the soundfield mic is that it does
not suffer from these useful but anomalous features. These factors should therefore
be borne in mind.
In many ways, the appearance of the
soundfield mic came at a good time for the BBC in terms of mic technique. There
had previously been a general swing away from a single coincident pair in favor
of several superimposed coincident pairs supported by close and distance spaced
mics; this move unfortunately coincided with the surround sound experiments,
and almost certainly magnified the technical problems of compatibility earlier
discussed. But by the time the new mic arrived, there had been a move back to
primary dependence on a well placed main stereo pair: as a result it was not
too difficult to incorporate the soundfield's output into more traditional setups.
It was expected that the soundfield
mic, because of its true coincidence with attendant lack of phase anomalies,
and its previously unheard-of flatness of frequency response, would similarly
not suffer from the somewhat impressive audio effects such anomalies introduced.
This indeed prove to be the case: the mic produced exactly what would be expected
from a truly coincident array. Once again there was the excellent stereo imaging:
Tony describes it as " electrifying" and says that things actually sounded for
the first time as they did in real life. Acoustic pianos, for example, lacked
the usual "clankiness" which occurs with some microphones. The Studio Manager
actually performing the mix is given total autonomy of the type and positions
of mics used at these concerts: in the case of the soundfield mic, he also had
a free rein in the application of the mic in the mix, making full use of the
variable altitude and polar diagram controls.
One surprising feature of the mic
in this respect was the fact that the array could be switched to "omni" without
the loss of stereo imaging! This effect is made possible on the Mark III production
models by means of the addition of a control to vary the degree of effective
coincidence of the capsules.
The one problem encountered with
the soundfield mic during these experiments was the very fact of its technical
accuracy: the lack of "lobing" and other anomalies tended to produce a very
clear, but somewhat dull and unimpressive, sound. This can be compensated for,
however, with the use of a rising equalization curve, adding just a touch of
"sparkle" to produce a result that is psychologically more satisfying, and more
like the usual sound that listeners expect.
These experiences with the mark II
resulted in the BBC placing an order with Calrec for two production models,
which were subsequently delivered and used during this year's Proms season.
Tests have also been performed with the Mark III in control conditions at the
BBC's Maida Vale studios, where studio managers have been able to experiment
with different positions and settings of the mic, and compare it with other
types. Initially the power unit on the new modules gave a little trouble, but
Calrec rapidly redesigned and eliminated the teething troubles--prevalent in
any new item of production. Bob Harrison praises Calrec's avowed keenness to
"get it right ".
In parallel with the Albert Hall
evaluations in stereo broadcast configuration, comparison tests were performed
by recording the B-format output on four tracks of an 8-track Brenell recorder,
placing on the remaining tracks a selection of outputs from conventional stereo
pairs. Apart from producing useful ambisonic demonstration material, these tests
provided an opportunity to evaluate the new mic in a number of operating modes
on an A-B basis with conventional techniques under control conditions.
The tests also proved useful in terms
of discovering the general performance of the mic and the best way to interface
it with recording equipment. Experiments proved that a level of about two hundred
nWb/m on tape combined with dbx noise reduction offered a dynamic range in excess
of 84 dB: it was possible to record an entire concert, with large variations
of dynamics and playing levels, without adjusting the gain of the system. Some
experimental recordings were also made in B-format on a Teac four-track, and
it is interesting to note that these seem to indicate that the B-format is far
more tolerant of tape azimuth errors than a normal-format recording.
Bob and Tony feel the Soundfield
is about the best mic they've come across. The variable controls have proved
exceptionally useful, the rigging time in a concert situation is very short
compared with a conventional the race, and the sound, although little dull without
EQ, offers superlative stereo imaging and depth. It offers "the best sound from
any part of the orchestra" that they have found, and the derived stereo signal
can be listened to comfortably for long periods without strain. But an ideal
mic will not necessarily give an ideal sound in non-ideal conditions--like the
average concert broadcast--and certain adjustments, like the addition of a touch
of EQ, are necessary to produce a suitably satisfying result. The fact that
the controls can be, in general, continuously varied during a performance to
optimize the esthetic parameters of a broadcast or recording, suggests that
the mic will have a great deal to contribute to the normal BBC setup at a concert
where there may be as many as three separate groups--TV, Radio and the BBC Transcription
Service--taking feeds from the main mics. The soundfield's output can be fed
to a separate control unit for each group, so they can each obtain a fully variable
independent main stereo signal without duplicating mics and without affecting
the versatility of any of the users' signal feeds. It is the BBC's intention
that a second soundfield mic should be tried near the stage, in addition to
one in the original position, to provide a versatile output for more "intimate"
miking, where a digital time delay would be employed to bring the output of
this mic back into perspective with the main system.
The BBC has very few complaints about
the soundfield system: surprisingly few, considering the radical new techniques
which this mic introduces. It finds that the production model appears to be
a couple of dB noisier than the mark II prototype for some reason--hopefully
Calrec will look into this, as it can't be a fundamental problem. The BBC also
finds the azimuth control a little coarse: it would prefer a four-position quadrant
switch plus a pot to alter the azimuth within a quadrant, or perhaps a forward/backwards
switch pointing the mic in one of two positions at one 180 degrees to each other,
plus a pot to vary the azimuth over the 180 degree segment, instead of the present
continuously rotating 360 degree azimuth control, which doesn't offer sufficiently
accurate positioning or repeatability (although the 360 degree control could
be very useful in rock studios). The BBC is also interested in the soundfield's
possibilities for opera, where a switch to "position" the mic figuratively "sideways"
in addition to the present vertical inversion control would enable the mic to
"look" at the orchestra pit and stage with equal ease. At present this can be
done by cross-patching the X and Z outputs, but a switch would be more convenient.
Many operators feel a dominance control is one of the most useful features of
the control unit, but the present control is too coarse in operation, and ought
to be replaced by a continuously variable front/back dominance control and an
up/down dominance control.
One other slight misgiving with the
mic is the fact that it requires a 60V phantom power voltage. The BBC, in common
with many broadcasting organizations worldwide, uses a standard DIN voltage
level of 48V for phantom power. This is a standing voltage on all BBC mic lines
in such places as the Albert Hall, and means that the soundfield mic presently
requires its own separate cabling and cannot be plugged into existing lines.
This is a disadvantage, and is presumably quite straightforward to rectify.
There is also a slight feeling that
the present control unit is over complex for purely stereo use: abandoning the
modular concept to produce a simplified stereo-output-only controller would
perhaps make the mic more affordable, placing it in the same price range as
other condenser mic arrays, although such a version would of course be rather
more limiting for ambisonic productions.
All in all, however, the BBC's opinion
of the new Calrec soundfield mic is a very good: it is a very significant development
in microphone and audio design which offers great possibilities for current
broadcast and other professional requirements and offers even more for the future
of sound reproduction as moves continue to bring about the standardization of
a technically successful surround-sound medium: the UHJ system.
This photograph shows the current
(production model) control unit required to operate a soundfield mic, as used
by the BBC. The signal is fed from left to right via the following modules:
the mic input module which takes the feeds direct from the mic head ("A-format")
after EQ in the body of the mic, and offers individual mutes on the equalized
capsule outputs plus a -20 dB A-format pad; the A-B matrix module which converts
the A-format signals to standard B-format (X, W, Y, Z outputs corresponding
to a front-facing figure-of-eight, omni, left-facing figure-8 and up-facing
figure-8 respectively) and offers mic-inversion facilities; the B-gain module
which offers switched control of the B-format level (+6, +14 and +20 dB lift)
and a rotary gain-control attenuator; the B-format monitor module which offers
individual PPM monitoring of the B channels plus overall levels; two soundfield
modification modules which control azimuth and elevation, and dominance; a mono/stereo/quadruple
module which handles the coincidence angle and polar diagram settings; and two
output modules, one equipped with ambisonic decode/loudspeaker layout control
and main output level control, the other controlling the stereo-output width
and headphone-monitor level. To the far right of the unit is the power-supply
module with power-on pushbutton.
No comments:
Post a Comment