As described on
former page, the Tannoy LSU/HF/15/8 and LSU/HF/12/8
crossover schematics clearly states 1.2 mH coil to the
bass driver.
Original Tannoy Schematics for MG15 or MG12.
Let's model my client's driver:
The driver as-is, no crossover. It's seen worse, in
particular for a 15" driver, but smooth is hardly
the right word.
The problem is this:
Above MG15 impedance plot (free air). This 8 ohms MG15
has a steadily rising impedance towards higher
frequences,
and it starts at ~150 Hz, reaching some 19 ohms already
at 1000 Hz.
Quite unusual - and it has implications as the crossover doesn't face a linear impedance.
Above the MG15 impedance plot compared to e.g. SEAS
26WFX002. Quite a difference.
For subwoofers, this doesn't matter, but when we have to
make a point of crossover above 1000 Hz it does have
implications.
Linearising MG15 impedance takes an RC circuit of 47 uF +
8R2. Tannoy didn't do this, so let's go back to the -
-
original crossover and see what happens:
This looks anything but ideal. A huge 10 dB bump at 1500
Hz.
- and the
impedance:
Looks bad. 3 ohms at 1500 Hz.
What if
we used an impedance correction circuit?
This looks much better!
Let's try
my client's crossover:
This really doesn't solve the problem. The peak has moved
down to 1 kHz and the result is
recessed midrange and still a bump some 5 dB above
average level.
And
here's the impedance:
- where we have a strange dip in impedance at 1 kHz. This
is just not a proper designed crossover.
What
would happen if we tweaked the "1.2 mH"
crossover, by adding a resistor to the 15 uF capacitor?
Still don't like the 1 kHz peak - and the dip in the
midrange.
If we were to
maintain the 1.2 mH coil, this topology seems more
likely. Is this an LR4 @ 1150 Hz or an LR2 at 900 Hz?
I wouldn't put too much emphasis on what kind of filter
topology we're using with these drivers.
Main target is to get the driver down some 6 dB around
1100-1200 Hz.
Measurement
- my client's drivers:
Conclusion to this study:
First
of all: Does the Tannoy crossover (the
one on the drawing) make a 10 dB peak at 1.5 kHz
as predicted from old 2005 files? Yes it does!
(I'd be really surprised if they didn't). Above
SPL of bass unit without crossover (red) and
driven from 1.2 mH with 16 uF to ground. If you
have an 8 ohms MG15 and a crossover that says 1.2
mH with 15-16 uF to ground, you most likely have
this peak.
Now,
listening to this really isn't nice to
put it mildly, so I wonder if most MG15
crossovers really have 1.2 mH? From what I've
seen on the web, most parallel caps are 15 uF, so
my guess is that we really do not find 1.2 mH in
most crossovers, rather 3.0-3.2 mH.
Would some of you MG15 owners please unsolder at
least one lead of your bass coil, measure
inductance and report back?
Comments from
Mark:
Hello Troels,
I
was looking at the Tannoy monitor gold information from http://www.troelsgravesen.dk/MG15-problem.htm
and your request for confirmation of the low pass inductor
value. I have attached a
document with several independent measurements referenced in
message #25577 by members of the yahoo tannoy group.
Another
set of measurements is also available at http://fn8142b.tripod.com/crossover.htm
All of the values for
the wc3120 low pass inductor are around 3mH as you have
measured and nothing like the 1.2mH handwritten value at
http://www.hilberink.nl/codehans/tannoy37.htm
It is unfortunate that this misinformation hasn't been
corrected.
Sincerely, Mark
Howell
Download reported data here.
Thanks to Mark for reporting :-)
Looking
at the frequency response from the
driver alone, this driver appears to need no
crossover at all. It's flat. We only need to make
it roll-off somewhere around 1.2 kHz. The driver
does well up to 2.2 kHz, but we need a series
coil to generate a low-pass filter. So why not
stay with the 1.2 mH coil? Below can be seen what
happens from A) no filter, B) 1.2 mH alone and C)
1.2 mH and 6R8+30 uF to ground. This is indeed
very close to a first order filter with some
impedance flattening from the RC circuit. All in
all it makes something close to LR4 on the
acoustic side - or LR2 depending on target point
of crossover.
The
bass' low-pass section
can be made in
numerous ways and there may be differences
between drivers that may require minor
adjustments.
Based on modelling seven different drivers, the
solution presented here will with a great deal of
certainty provide an overall balanced
presentation.
Did
I forget to tell how much I've enjoyed
hearing the MG15s again? Even from these
non-optimised 100 L test cabs:
MG15s
can do things very few speakers can do.
Music all of a sudden becomes not only an
intellectual/emotional experience, but physical.
What the MG15s do better than most, is
reproducing the energy in upper bass/lower mid.
The absence of points of crossover in this region
is a key parameter - not to forget the very size
of the membrane. If anything should prove that
size matters, this driver does. In fact, it does
it so overwhelmingly that it may take some time
to adjust to the sound. We're simply not used to
feel the physical impact of music in our homes.
We may very well know what it's like in live
concert, but getting this within our four walls
calls for mental adjustment - and not least
finding the things that rattle around the house.
Treble from the compression driver driven through
the silver caps is never heard better, it's
clear, crisp and transient like conventional
domes can never match.
This
speaker does
Siri's Killer
Note without problems. I was quite
surprised to hear this. I didn't have this CD
back in 2005 and considering the lack of proper
phase integration between drivers, this came a
bit of a surprise.
In fact, this speaker seriously questions the
relevance of "proper phase
integration". The question is how good is
has to be to be audible? Reversing tweeter
polarity here does sound bad, no doubt
about it, and the impact on overall response can
be seen from the measurements shown in the
MG 15 Up-Grade file. The crossover here
is basically low order on the electrical side,
producing a LR4 on the acoustic side helped by
the inherent roll-off of the drivers. One thing
is phase shift, another thing physical size and
low inherent distortion. Use big drivers that
don't have to move much, and life starts getting
better and if I had a bigger house the MG15 would
be one of my stock speakers.
By
the way:I couldn't help thinking of the
Chario Sonnet
speaker when doing this crossover. Same point of crossover,
basically same filter topology. Does Tannoy's chosen point of
crossover in reality help mating the two very different drivers,
the huge 15" pulp cone and the weightless compression driver
dome? Seems like the ear is less sensitive to less than perfect
phase integration in this area.
End
of MG15 story. I think....
|
MG15 from three
different scenarios.
Modelled response of
C.
|