Last Updated: 27 August 2000
I didn't want to buy another pair, so there were three choices... the Monitor Audio MA5's (which sound a bit lacking to say the least) or the KEF Q35's. Since the KEFs sound already excellent, I might have choosen to buy a bargain box pair (such as the decent JPW ML310s, or more upmarket, the Tannoy Mercury M1s) to experiment with, to test the software with...
However, what I actually did was to buy a new set of drivers to build my own - from Hi-Fi world, I got the drivers used in their KLS9 floorstanders for 160 quid, which consist of their 8" Aerogel bass/mid (HM210Z0), and a 25mm fabric dome tweeter (TW025M0). After setting them up with a cheap 1st order x-over, temporarily installed in the MA5 cabinets, I decided that I didn't really like the tweeters much, so I took the plunge, and spent a considerable amount of money on a matched pair of Morel MDT-33s.
These fabric-dome tweeters are among the best in their class, utilising a basic design licensed from Dynaudio. Resonance is around 750Hz, with the drivers typically being flat to within -/+ 1dB from 1 to 17KHz, as well being sold in matched pairs. This should make them very suitable for a 2-way design which uses a larger than typical mid/bass driver, such as is the case here.
The cones of the Audax drivers are made of a special material called HDA (High Definition Aerogel) which is created by mixing carbon and kevlar fibres with an acrylic gel. This produces an almost-optimal material for a speaker cone with good internal damping and light mass. This particular driver was chosen since it was used in Dick Olsher's Poly Natalia speaker with apparently good success, as well as Noel Keywood's KLS9 design for Hi-Fi World mentioned above.
The Old Design
The cabinets I ended up with were a little strange... based around 12mm thick compressed paper tube (treated with wood hardener), of approximately 220mm in diameter. The front 1/3rd has been chopped off, and a piece of 12mm MDF slapped on the front. The bases are thick acrylic discs, while the tweeter is suspended on the very top of the cabinet (more MDF sheet) on a bed of blu-tack.
Inside the cabinet, there's some cheap wool stuffing at the base, and a bit of thin foam just behind the mid/bass driver - pretty weak damping, but it's the best I can manage for now. The foam behind the mid/bass should hopefully be replaced with Deflex pads, to improve dispersion. The main problem at the moment is the underside of the speaker tops - no damping or dispersion going on at all! The diagram shows bracing panels, but none are actually present - never got the chance to get any made... :(
Cabling is QED Qudos Silver for the bass/mid drivers and Maplin silver speaker cable on the tweeters. Silver solder used to connect both. An obsession with silver? No, I just hate oxidisation. The Morels had 3.3uF ICW audio grade polyprop. capacitors across the terminals in order to roll off the top end, since my ears used to be in pain after 10 mins or so of listening to them.
The reason for the caps is not only to roll off the upper end (probably necessary due to the quality of the DACs used, rather than through any fault of the tweeters), but also to protect the tweeters from high-frequency noise transmitted from the PC's insides. It's important to remember that the drive units are connected to the power amplifiers.
The New Design
After extensive listening, I still wasn't happy with the sound which the design was producing - the reasons for which only recently became clear on the Madisound forum (apparently, the HDA drivers don't like narrow baffle designs, and need a lot of room around them inside to function well, with attention being paid to the back wave path). The general consensus on the basslist seemed to be that an 8" mid/bass was usually too large for a quality 2-way design (excepting a few rare cases), so I decided to move to a more compact design using a smaller mid/bass.
I decided this time to go for a driver
which was more likely to produce good results in the hand of an
amateur speaker builder such as myself. 4" drivers offer a
very wide frequency range with good transient response, but sadly
no real bass capability to speak of.
Next were considered drivers in the 6.5" range, but none seemed to have the frequency response that I was looking for. It looked like it would have to be a 5.25" mid/bass.
One driver which immediately came to my attention was the Audax HM130C0, which was used to good effect in Von Schweikert VR6 ($12,500/pr) and VR8 ($18,950/pr) in modified form.
After looking at the datasheet,
observing the excellent response, my mind was made up, and I
placed the order to buy a pair of the drivers.
When it came to designing the cabinets, I thought
I'd make use of a box design program this time, since the shape
would be a standard cuboid rather than anything out of the
ordinary.
The program which I eventually settled on is called WinISD - still in beta, but very usable, it features a large database of driver data (including the Audax unit I wished to use) and a decent user interface.
A few minutes use, and I already had a sensitivity graph on the screen. After some tuning, I settled on a sealed box of approximately 4 litres capacity, with an f3 (-3dB point) of around 100Hz or so.
Why a sealed box? Well, I didn't like the idea of a rear-firing reflex port, and a transmission line is beyond my design ability... so I'll suffer the decreased bass response for improved control of what is there. Besides, it should be easier to add a subwoofer to a sealed design with a smooth rolloff.
Since the driver is only a 5.25", I
may attempt to apply some equalisation in order to bring the f3
point down a little (perhaps to 80Hz or so) but it remains to be
seen how well the Audax will handle being equalised - it is
likely to make the midrange suffer, but hopefully I'll be able to
equalise a little without adversely affecting things too much.
I now have a pair of HM130C0-VR8s ordered from Madisound, as well as the standard versions. They look identical to the normal ones apart from the sticker on the back. The way to tell them apart is to press down on the cone - the standard version will hit the endstop quite easily, while the VR version seems to have a progressively resistive suspension to prevent the driver from bottoming out.
Differences are also present in the frequency response, as the graphs will show below. The VR8 seems to enter breakup sooner, but avoids the big peak at around 4.5K on the standard version, instead replacing it with a dip... don't have any TS specs on the modified drivers, but they definitely seem to have deeper bass than the standard versions as well.
The new design is now looking in somewhat better shape since the
tweeters are properly mounted, but overall is still not quite
there yet. I've taken a newer picture, which is visible below...
Here are a couple of old shots inside the speaker... before and after lightly stuffing with polyester wool... the back panel now has bubble wrap on it, and the wiring is now all silver plated OFC rather than Cat5 cable. Resistor visible has gone too...
 |
 |
There are still a number of things to be done regarding the speakers, as well as the tweeter mounting... for a start, the corners on the fronts need to be planed off to reduce the baffle width a bit. On the inside, the woofer cutout needs to be made a little wider to allow the woofer to breathe a little better. Also, the sealing of the back panel is currently very lacking, and will need some attention before the design can be considered near completion.
Drivers For Sale
Since I don't strictly need the HM210Z0 drivers anymore, I'm willing to sell them for 100 quid - that price includes Interlink delivery and ~4m worth of QED Qudos Silver. Contact me if you're interested... someone could certainly do something very impressive with these drivers if they were willing to take the time.
The Backup Design
Since the Monitor Audio cabinets were sitting around doing nothing, I decided to buy some new 8" drivers for them. Yet more Audax, yes... this time a woofer from their new polymer chassis range, the AP210G6 for 18.50GBP each. They are coated paper, and have a sensitivity of around 90dB. Plenty of bass output, although the top end is a little ragged.
The tweeters are currently the old Audax TW025M0s, but these are likely to be replaced with the original Monitor Audio fibreglass ones, due to my sonic preference for them. With nothing but a very basic crossover, these speakers actually don't sound all that bad - hopefully the digital crossover should elevate the performance to an entirely new level.
Resources
Old Design
Audax HM210Z0 F.R / Impedance (thanks to Alex Megann for scanning the datasheet in!)
Fullrange Response w/brickwall @ 1500Hz (unsmoothed) (~40deg off-axis, DIY cabinets)
Factory Measurements
Own Measurements
All measurements are made with an M31 measuring mic from Linear-X, which has not been corrected (included graph shows 100-17000Hz +/- 0.6dB), on-axis unless otherwise stated. Measurements were taken using a ten second snippet of white noise, which were then averaged.
Audax HM130C0 (left), 10cm
dist, MM cabs, uncorrected
Audax HM130C0 (right), 10cm dist, MM cabs, uncorrected
Audax HM130C0-VR8 (right), 10 dist, MM cabs, uncorrected
Audax HM210Z0 (left), 7cm dist, MA5 cabs, uncorrected
(note that the dip ~400Hz is due to the damping foam in the MA5 cabinets)
Morel MDT-33 (left), 10cm
dist., MM cabs, uncorrected, corrected*
Morel MDT-33 (right), 10cm dist., MM cabs, uncorrected, corrected*
(note that 73-tap correction filter was created from left tweeter measurements - no averaged filter created... yet! also - the plummet at ~20KHz is not only tweeter related, but due to the Vortex 2's Sigmatel 9708 codec, which starts to roll off at ~19KHz...)
MA5 fibreglass tweeter (left),
5cm dist., MM cabs, uncorrected, 1800Hz
1st order HP
MA5 fibreglass tweeter (right), 5cm dist., MM cabs, uncorrected
