August 10 -
I have crossovers to audition this week for
the S9800, MT with response to 13kHz, and the MTM 3 way. I have attached
an image of the MTM and you can see the problem with the supertweeter
mounting. Yeah, I can hear you laughing, but I don't have a CNC router.
The problem with current MTM is that I am getting comb filtering effects
above 6kHz. Without the supertweeter I do not get the comb filtering
problems. I'm not sure if the problem is caused by the placement of the
supertweeter or if it is some sort of cancellation problem caused by an
interaction with the midhorn. I will post an image of the response
tomorrow so that everyone can see it.
I would suggest that the industrial desgin team begin work on the
cabinet design. I would suggest a modular design with 8 separate pieces.
4 identical woofer cabinets, 2 horn/supertweeter cabinets, and 2 stands.
The woofer cabinets should have close to 4 cu ft internal volume and be
able to accept twin 4" flared ports up to 18" long with room for the
driver as well. I would suggest a baffle of 17" high x 19-20" wide. For
the horn subcabinet, the box needs to be 12.5" tall to house both the
drivers. With this configuration of cabinets, the system could be used
as an MTM, with a 17" stand added at the bottom (which could also house
the passive network) or it could be used as a vertical 2.5 or 3 way
depending on Don's preferences and needs
August 11 -
The comb filtering occurs in the predicted
response. There is no comb filtering in the raw 045 response. The comb
filtering only occurs when the 045 and 435 are used together. I will
post some pictures to illustrate.
I will see what happens if I raise the crossover frequency to 13kHz.
Here is the response of the 045Be with and without the crossover at
Here is the response of the 435Be on
the Sonoglass horn with and without the 4th order bandpass filter at 800
Here is the response of the 2 1500AL
woofers in the MTM format. The crossover is 800Hz 4th order low pass.
Ignore the response below 200Hz since it captures part of the room
Here is the response of the woofers and
midrange together. The slopes at the crossover point still need more
work to eliminate the dip in response.
Here is the woofer and midrange response
again and the supertweeter response is overlaid on the image. Notice the
comb filtering the occurs well below the crossover frequency. Can
someone send me a link to the S5800 crossover so that I can try that
This image show the individual
filtered driver response curves along with the system response. It would
be easier to see if I could do the curves in different colors.
Here is the response with the
crossover changed to 13kHz.
August 11 -
All right, the comb filtering is caused by
the placement of the supertweeter in the front-back or z dimension. I'm
not sure how JBL deals with this with their networks for the S5800, but
it appears that the midhorn in the S5800 is shallower than the Sonoglass
S9800 horn. This presents a problem for the MTM design. The supertweeter
may need to be moved backward into the cabinet or delayed electronically
to solve the comb filtering problem. I tried adding 180 degrees more
phase rotation by using a 6th order slope at 10kHz but that didn't help.
Here is the response of the 3 way MTM with the S5800 filter with the 045
moved back 8cm in the z dimension.
And here is the MTM crossover based on the
S9800 network with the 045 moved back 10cm in the z axis.
August 11 -
One of the assumptions of a Linkwitz Riley
4th order network is that all of the driver acoustic centers are aligned
in the z-axis. This assumption was met in the 4430 from what I remember,
and the woofer and midrange appear close to being aligned in the S9800
cutaway drawing. Now it is more clear why the 045 is offset in the
Is this comb filtering problem something people are willing to accept? I
don't think the project would have much appeal if it requried digital
delay for the supertweeter. Recessing the 045 into the cabinet would
likely cause more problems than it would solve unless someone came up
with a really creative solution. Also remember that it is best to keep
the woofers as close together as possible in an MTM and that the
vertical height of the horn/supertweeter assembly is currently 12.5
I already attempted to add another 180 degrees of phase delay to the 045
by crossing it with a 6th order acoustic network at 10kHz but it didn't
solve the comb filtering problem because of the high frequencies
I listened to the MTM with the newest crossovers today, and the high
frequencies sound too bright. However, it wouldn't be a problem to add
some downward sloping response when I do the system optimization. Based
on conversations with others who have built MTM's, they tend to sound
bright if they have ruler flat response, so most people add some
downward slope or a dip at 3kHz. Then again it may just be my room.
August 12 -
According to JBL's Greg Timbers: "The
comb filter effect is normal and cannot be eliminated, even with the use
of electronic delay. The wavelength at 10k is an inch or less and even
if the 045be were delayed to the z plane of the 435Be, even the
slightest movement up, down, left or right of the measurement position
would change the relative path length between the 435Be and 045Be to
where the interference pattern would return.
What I have done is use a 4th order on the 045Be high pass and adjusted
the values using substantial smoothing on the frequency response curve.
The polarity does seem to matter in getting the smoothest "smoothed"
I originally had the 045Be coming in where the main 435Be cuts off (15k)
which cost less parts, but we could hear the 045Be as a separate
location and it was distracting. I found that 8k - 10k minimized this
effect. In spite of the curve, we have had no negative comments about
the 045Be and the choppy response through the crossover range."
August 28 -
If Tim will ship the pair of H9800s and
drivers along with the bat slayers (045Be) I will whip up a pair of
combined horns that will be suitable for listening tests and basic
aesthetic study. They will be machined and carved plastic in an MDF
frame. (Trust me they will look pretty close to factory, a bit more time
and they would look factory.) If I got them by mid week, I can turn them
around by the end of the weekend. I will also incorporate a sub
enclosure that will be structural so that the 1500AL enclosures can
stack on the mid and high horns.
There is no way I can fabricate a pair of wood combination horns in less
than several weeks. I am not willing to make more than one pair of them
so we will need the industrial design to be locked in before I begin
As for the industrial design, if we could get the dimensions of the
front layout, i.e. woofer spacing and height of the centerline of the
H9800 from the floor that would be very useful.
Once armed with that info I can go back over all of the previously
posted drawings and sketches that have been posted by myself and others
and put them together in a more coherent form to aide the design process
and get the ball rolling again.