On 02/20/2013 09:20 PM, Eric Carmichel wrote:
Greetings to All:
The system is comprised of 16 channels. The advantage of 16 (or less)
channels is that I don’t have to use a dedicated Word Clock to sync
my MOTU FireWire interfaces. I’m fond of these interfaces, and two of
them can sync-up via the FireWire link without complication.
i don't know which MOTUs you're using, but don't most of them include
ADAT outs? so you could easily get 8 extra outs per MOTU without any
sync hassles, and just the minor investment into an 8ch DA converter.
I intend
to use two hexagonal arrays of small-sized loudspeakers (a bit larger
than the 3-inch coned Genelecs, but not much more so). One array will
be near floor level, whilst the second array is proximal to the
ceiling.
a few years ago, i was more or less forced to implement such a setup
because of a frontal video screen. it works, but it's not optimal for
horizontal sources. if you don't have any other constraints, i guess
it's advisable to start with a horizontal hexagon and add two rings of
three speakers each at +/-60° elevation.
your usecase seems to be a single listener in a fixed position, so my
advice may not apply, but at least for moving listeners, i can say that
rings near ceiling and floor are not as stable and easygoing as a main
horizontal ring plus smaller numbers of high and low speakers.
According to the literature (Malham, Rumsey, and others come
to mind, but I’m shooting from the hip), diametrically opposed pairs
may be preferred when the listener is centered in the array. Further,
it is purported that six speakers provide immunity against drawing
signals toward a single speaker.
BLaH2, iirc. to add some anecdotal evidence: for me, a hexagon is also
far superiour to a square for first order. plus it offers the
possibility of going for second order - not an option for your recorded
soundscapes, but maybe interesting for synthetic cues, which can be
rendered more reliably than in POA.
Eight speakers is probably overkill
and doesn’t leave me the four channels needed for a square array of
subs.
for first order and a single listener, certainly. for artificially
panned cues, maybe not.
Any thoughts as to whether the two hexagonal arrays providing
horizontal and height information should be offset or vertically
aligned?
no. i've only ever used aligned rings.
Regarding the need for subs: With ‘normal’ music content, twelve
speakers working in concert would provide more than adequate
low-frequency energy. But I’m going to be using live recordings where
a particular low-frequency sound could be coming from an extreme R,
L, front or back direction. In this scenario, I’d rather have the
subs handle the load but I still need to preserve ‘direction’ as
stated above. Because four speakers can provide adequate surround
sound, my intent is to frequency-divide the B-formatted signal and
send the highs and lows to their respective feeds via ‘conventional’
Ambisonic decoding. To be clearer, I will digitally filter the
B-format signal so that each of its four components (W, X, Y and Z)
are divided into a high and low-frequency signal component. The
low-frequency components will be decoded and sent to the square (and
likely horizontal) array of four subs. The highs will be decoded
based on the position of the 12 ‘full-range’ speakers. I use
full-range loosely here because the added bass channels aren’t for
enhancement, but to alleviate the 12 speakers from their low-end
duty.
that approach works very well, i've used it several times to good effect.
I haven’t determined the best crossover frequency, and this may be
determined in part by a combination of the speakers used and the
stimuli to be presented. I wish to use the lowest possible frequency,
but not to the point of driving the small speakers to distortion. I’m
guessing a digital (crossover) filter that is both maximally flat and
phase coherent is best, though slight dips caused by frequency
response anomalies are easy to EQ out. I use EQ judiciously because
it is generally just a marginal cure for a loudspeaker's
deficiencies. Upping the response at some frequency extreme merely
adds to distortion that is ‘measured’ (in SPL) as a boost at the
deficient frequency (or third-octave band or whatever). Only a
spectrum analyzer or critical listening reveals where the real boost
is occurring.
and then there's the room...
for normal p.a. use with small stacks, i like shallow filters. but if
you can't get the physical alignment correct everywhere (as is the case
when the listening area is large and the subs are at a non-negligible
distance from the tops), you might want to reduce overlap as far as
possible, because it will be wrong almost everywhere, regardless of your
time alignment. hence, i'd argue for 24db/oct linkwitz-riley. in theory,
8th order (48dB/oct) should be even better, but there may be other
problems in using those, and i haven't had the chance to do an a/b
comparison.
Although my proposed strategy focuses on subwoofers and low
frequencies, it may find purpose at higher frequencies. For example,
I read about Oticon’s carefully placed 39-speaker array in the
Hearing Journal (circa 2010). I recall that the speakers were
positioned within 1 cm of their ideal position. But what about the
‘acoustical centers’ of loudspeakers? How would this be determined?
Justification of such extreme placement would require knowledge of
phase characteristics and an exacting acoustic center. However, by
applying frequency division to the B-formatted signal, each speaker
within a bi- or tri-amplified enclosure could receive its own, unique
decoded signal based on its absolute position within an array. Of
course, this means a s- load of channels and amps, but perfectionists
and the technically inclined may find this appealing.
it is. particularly because you could use lower orders for low frequency
bands, and increasingly higher orders for the treble. if band-limited
active single drivers were available at the same economics of scale as
entry-level studio monitors, a three-band multi-order ambisonic system
could become interesting. for now, it's only of scientific interest,
because you'd be much cheaper off buying more fullrange speakers than
trying to build individually amped limited-range drivers.
Anyway, I’m curious as to what others may have attempted with regards
to bi-amping the B-format signal (or whether it’s a remotely good
idea), use of multiple subwoofers, and whether the hexagonal arrays
providing height information should be offset or vertically aligned.
As always, I appreciate the help and insights of the experts and
experienced Ambisonics aficionados.
an application of dual-band decoding in the context of large p.a.
systems is described in
https://stackingdwarves.net/public_stuff/linux_audio/ambisonic_symposium_2011/
the system i mentioned before (with the two rings at floor and ceiling
level) is briefly described in this poster:
https://stackingdwarves.net/public_stuff/linux_audio/aes_uk_2012/Poster-AES-UK-web.pdf
another dual-band combo of neumann kh120 and kh810 subs:
https://stackingdwarves.net/download/neumann-fotos/4-floor_speakers.jpg
https://stackingdwarves.net/download/neumann-fotos/5-rigging.jpg
these had a rather dirty overlap (the tops were driven down to 55hz with
their natural roll-off, and the subs were rolled off at 12db/oct around
80hz, no justification, it just sounded nice, with enough "oomph" :)
best,
jörn
--
Jörn Nettingsmeier
Lortzingstr. 11, 45128 Essen, Tel. +49 177 7937487
Meister für Veranstaltungstechnik (Bühne/Studio)
Tonmeister VDT
http://stackingdwarves.net
_______________________________________________
Sursound mailing list
[email protected]
https://mail.music.vt.edu/mailman/listinfo/sursound
