Copyright 2010 © Troels Gravesen
18W/8545-00 and D2905/9500, series filter
Check out other parallel filter combinations here:
Some three years ago I made a two-way, transmissionline floorstander from these speakers and the sound wasn't bad at all. Eventually I got caught up in other projects and the drivers went back on the shelf. The 8545 was part of the drivers tested for the Acapella midrange - where it did a good job again - but was beaten by a small margin from the SEAS W18EX001 for this particular application. When I launched my "from the archives" section on my wensite, I thought this two-way, including a parallel filter, might be part of this collection. I hadn't realised that this particular construction would create the largest amount of responses to this section so I better get started.
The 18W/8545-00 is indeed an old
fellow. I don't remember the exact time of
release, but it must be more than ten years ago we saw
the first constructions utilising this driver, usually in
smaller 15 litre vented enclosures. I know that a number
of people have chosen this driver first to the 8535 due
to large magnet - and the magnet is huge. Large magnets
are known to increase transient response but the
trade-off is reduced bass extension. Not quality, but
extension. This driver cannot play the deep bass as heard
from the 8531 driver. The Qt is too low. We might add a
series resistor to the driver to increase Qt and get a
deeper bass but then we would sacrifice efficiency.
Playing around with the simulation programme a 15 litre
vented cabinet would get close to the theoretical optimum
volume. However, a 24 litre vented cabinet would take the
F3 down to 38 Hz compared to 47 Hz from a 15 litre
cabinet with the same 36-38 Hz port tuning. This possibly
with a slightly less precise bass but it might be worth a
shot. What speaks for a 24 litre vented cabinet is that I
have a pair!
Initially I thought I'd try a completely new approach to this midbass driver and mate it with the Vifa DX25TG tweeter as I have had good results from this tweeter lately. And instead of a parallel filter I'd try a series filter with the DX tweeter. And I'm possibly going to try this also. The 8545 + 9500 is a classic, seen in numerous speaker constructions and why should I try to copy something that so many probably have done so well before? But as a starting point, this will be the basic layout. Then we'll see how things progress.
Things don't always turn out the way you thought it would. After struggling for many hours with the parallel filter I gave it up. First of all I had placed the tweeter right in the middle of the somewhat narrow front panel (190 mm) giving it 80-95 mm to all edges. Disaster! Huge dip at 3.5 kHz and huge bump at 2 kHz. Awful! Impossible to get the crossover right to match these conditions. So, off went the upper part of the front panels and new MDF inserts were made to off-centre the 9500 tweeters. Much better response. Coming to think of the SP95 I didn't remember having this much trouble getting the 9500 working properly. So, in went the SP95 series crossover and now things started to work in the right direction. The 9500 has a lot of response capability at 2 kHz and somehow it appeared easier to control from this series filter. Well, you don't just throw in a filter from another construction and get it all. No way. A few more evenings of fine-tuning component values and eventually ending up with a 4th order HP-section to the 9500 - just to get the damned thing down at 2 kHz. The point of crossover is around 3 kHz and some will argue that this is too high for the 8545 driver. I'm not sure I agree. Try running the 8545 from a 3rd order filter to 3 kHz and disconnect the 9500 tweeter and play loud. Does the 8545 - alone - sound aggressive or smeared? I don't think so. These carbon fiber cones or often accused of having a rough upper midrange and nobody asks how the tweeter is performing - because in most cases the tweeter to some degree shares the 1.5-3 kHz range with the midbass depending on the slope of the crossover used. At least from this heavily coated cone the upper midrange/lower treble appear smooth and clean. The un-coated 8535 from the 2.5 clone is another story!
At this stage I'll take the risk of
releasing the series crossover and measuring performance
because this speaker is slowly beginning to sound really
good. From my experience much of the possible
aggressiveness in the upper midrange/lower treble region,
800-2500 Hz, originates from three factors:
I'm positively surprised by the three-dimensionality of the 8545 driver. Much better than I remember - or ever have had. The impact on horizontal dispersion from having a point of crossover at 3 kHz has been measured and is displayed in fig. 5. From 0 - 30o we very much observe the same response, only the tweeter starts declining above 15 kHz. No surprise.
19-09-2004: My initial plan was to try implementing the Vifa DX25TG tweeter also, but the current set-up is doing so well that I'll leave it here. This is a speaker that handles all musical genres equally well. It has a good bass performance and with the modest sized cabinet it should also have a high WAF.