ScanSpeak 2 x 18W/8543-00 + 13M/8636-00 + D2904/9800
18W/8543-00 may be replaced by 18W/8545-K00 or -00 without changing crossover.
A friend of mine came up with this project and had decided on the well-known Scan-Speak 18W/8543-00 drivers for bass. These drivers have been seen in ProAc speakers and deliver a solid bass and due to the polypropylene membranes also know to deliver a smooth sound without being edgy at all (i.e. no serious cone break-ups). The midrange driver was new to me and one of the reasons for choosing this one was because it looks good! Hmm...this is probably not a good reason for picking a driver but after looking at the data from the dst website I had no arguments for not choosing this driver. I should later learn that it also sounds very good. The 9800 is the best metal dome tweeter I know of. So, I made the cabinet drawings and driver layout, my friend routed the front panels and the cabinetmaker did the raw cabinets and I had to assemble the whole thing and - not least - produce the crossovers.
After having finished the speakers - and being very happy about the result - I asked my friend if I could publish the construction. After all, he paid for all the hours going into the project. He said yes....if I would promise that all "cloners" would make their cabinets from thin balsa wood. So, there you go.....
This project did fit in very well with a lot of things I have been targeting lately: High sensitivity and the use of smaller upper midrange drivers. The 18W/8543-00 midbass has a claimed 88.5 dB sensitivity (large magnet and low membrane weight) and take a look at the blue curve below and you will see the FR from the 8543 driver at 2.8 V applied at a measuring distance of 1 metre. A quite sensitive driver. Two of these in parallel should produce a sensitivity of 94 dB/2.8V (not to be confused with efficiency), but the lower bass is not going to handle the same frequency range as the upper bass, hence an overall sensitivity around 91-92 dB/2.8V. This relatively high sensitivity is an immediate positive quality you realise when you connect to the amplifier. With a 100 wpc (8 ohms) amplifier this speaker can rip off your ears without serious distortion and looking at the impedance curves, it's an easy load on the amplifier too. Due to the midrange it can play low too without loosing grip of program material and it can do well for a Saturday party set-up. Quite versatile I should say.
My 50 wpc valve amp likes it too, but due to the pp cone midbass drivers and the kevlar midrange, this speaker is quite tolerant on programme material and I like the solid state amp the best here. With the valve amp it became a little too smooth and nice. Transient response is excellent and the overall level of transparency is far better than anticipated despite the soft cones. Maybe the valve amp wasn't too happy about a 4 ohms impedance. It did well, but the solid state was slightly better.
Making a 3½-way crossover does - as always - takes quite some time and without the LspCAD it would have taken ages. The good thing about these drivers is that they can handle frequency ranges far below and above of their target area:
Blue = 18W/8543-00. Red = 13M/8636-00. Green = D2904/9800.
As can be seen from the frequency response graphs the bass driver handles well op to 4-5 kHz without any serious break-ups. The 13M is indeed a very broad-banded driver, doing fine all the way up to 10 kHz and the 9800 tweeter is flat down to below 1000 Hz. So, major surplus on all drivers as the points of crossover is going to be around 700-800 Hz between bass and mid and around 4 kHz between mid and treble.
The mid can really go down to 300 Hz but initial testing with an electronic crossover suggested a point of crossover around 800 Hz giving the best vocal performance. The middle midrange from 320-640 Hz is very well handled by the upper 8543 bass driver. Mind you, this driver is well know for 2-way constructions handling up to above 2 kHz.
Details on cabinet construction, use of damping materials, etc. can be found in SP38, TJL and other files and only basic dimensions and in particular driver placement will be given here. It is important that drivers are placed exactly as shown on the drawing, otherwise the crossover will not work as intended.
The cabinets are made from 19 mm MDF and the front panels from 25 mm solid birch wood. Internal panels are added 4 mm self-adhesive bitumen pads. 10 mm polyester foam is placed on panels around the two bass drivers (two middle sections). The midrange cabinet is 1.5 litre volume with bitumen pads on all sides and damped with MDM3 damping material.
Due to the thickness of the front panel it is very important to chamfer the driver holes, in particular for the midrange. If you don't, the midrange will sound congested due to poor rear ventilation. I dare say this does more to the sound than the quality of the crossover components. Damping of the mid enclosure has quite some impact on sound as well. Don't forget to experiement here. Too much damping makes a dry, forward sound.
I roll up ½ x MDM3 sheet to be placed at the bottom of the cabinet and similar at the top. ½ x MDM3 is placed at the rear panel behind the bass drivers. I all I have used 1½ x MDM3/cabinet.
Vent is 65 (ID) x 70 mm, giving a vent tuning of 33 Hz.
The finished Zahras
Birch veneer and solid birch front panel
Due to the smooth rolloff of the drivers, the crossover can be made quite simple. Basically 2nd order filters are used except for the tweeter that needed a 3rd order filter to match the mid-driver and to produce a good phase profile. As can be seen from the lower graph above, the phase tracking between drivers is unusually nice. As shall be shown later, this was followed by actual measurements, where inverting polarity of the mid-driver produces broad and deep suck-outs in the FR response. Due to the 2nd order topology, the mid is connected with negative polarity and the tweeter has same polarity (negative) as the middriver.
LspCAD predicted response.
The crossover is divided into two sections and the mid + tweeter is placed on the inside of the front panel.
The crossover for the two bass drivers are placed on the cabinet plinth as far away as possible from the mid-tweeter section.
All internal cabling is done with rigid, silver plated and teflon coated cable, Teflar 2, from LCAudio.
Crossover for bass drivers. Place the coils as seen above to minimise interaction.
Glue the coils to the board by dumping them in a load of silicon glue or Superfix, the
latter my preferred glue for making crossover. In this way you are sure that nothing will
rattle inside the box once in place. Sourcing a rattling noise inside a finished box can drive you nuts!
Crossover for midrange and tweeter. The placement of coils may look a little peculiar,
but again the layout was chosen to minimise coil interaction.
Complete crossover kit excl drivers available from Jantzen Audio: firstname.lastname@example.org
Contact Jantzen Audio for shipping details - please state which country.
All technical questions at email@example.com
You may add two ports of 68 x 150 mm (cut to 80 mm length), terminals, wire and damping material to your order.
Impedance of all drivers in cabinet. Purple/black = 8543 in parallel. Green/yellow = middriver.
Red/blue = tweeter. The impedance for the bass drivers are before the vent tuning was changed to 33 Hz.
Red = summed response of bass and mid. Nice suck-out at 850 Hz. Blue = bass and mid same polarity.
Red = mid and tweeter with opposite polarity. Blue = mid + tweeter with same polarity.
Significance of mid attenuation. 0R, 1R0, 1R5, 2R2. 1R5 is preferred. The sonic impact of this is hard to ignore
and don't miss the opportunity to hear what happens to the overall sound from such a small changes in driver attenuation.
Looking at the frequency response graph, you may wonder if this speakers isn't balanced to have an overall dark sound. It isn't. Despite the significant tilt downwards - a little more than the BBC dip - the speakers have a slightly bright and forward sound. In an absolutely non-aggressive way I rush to say. Due to the dispersion characteristics of the individual drivers, the on-axis response may look different from speaker to speaker despite my attemt to tonally balance the speakers in a similar way. The 13M driver has a much better sound dispersion in the 1-3 kHz region compared to a 6" going up to e.g. 3 kHz, thus needs a little further attenuation to produce an overall balanced sound. The tweeter, as always, has an attenuation until it only "dissapears", i.e. if you can still spot the tweeter in the overall sound, be sure it's playing too loud. The 9800 is very sensitive to this and changing the tweeter attenuation resistor from 2R2 to 1R8 is clearly audible.
Impedance of finished speaker. Vent tuning still 33 Hz here. 33-34 Hz in final construction.
This is a 4 ohms speaker but an overall very easy load on the amplifier.
Step response from finished speaker.
As can be seen: tweeter and mid withnegative polarity. Bass with positive polarity.