When I introduced the original Eros
design, I attempted to balance many factors -- frequency response,
impedance, sensitivity, driver rolloff rates, etc. In the past year, I've
listened to the original design almost every day. I'm still impressed by
the exceptional performance of the drivers, especially when considering
their modest price tags, but I've always had a nagging feeling that I
could improve on the original.
In the original, I was never completely comfortable with the
impedance dip of 2.85 ohms or a very subtle forwardness in the midrange.
Late last winter, I began designing new crossovers for the Eros. Since
then, I have built and listened to at least 20 different variations. Some
crossovers had as many as 37 components and used 6th order acoustic
rolloff rates on each driver, while some had as few as 6 components. I
experimented with conjugate filters for impedance compensation, lowering
the sensitivity of the system to smooth the PL18's natural response hump
in the 800hz to 1khz region, and various other concepts. Each time, I
would listen to the new crossover for about two weeks, then switch back to
the original. Each time, the original won out due to the good sensitivity
and dynamic, open sound it permitted the drivers to present.
Then I started looking down the last road -- series crossovers. I
began an in-depth study of various methods for controlling driver rolloff
rates (necessary to achieve good horizontal and vertical dispersion
characteristics). I tried various first and second order series
crossovers, and ultimately tried hybrid designs (for example, a first
order series network on the tweeter with a second order topology on the
woofers). These crossover showed promise in their ability to improve the
driver dampening outside their passband, reducing out-of-band resonances
and increasing the resolution of fine detail.
The new crossover is a carefully optimized series
network that results in sensational on-axis and off-axis performance, as
well as improved vertical polar response over the original. This is
clearly evident in the pinpoint imaging and wide soundstage. Further, the
dynamics of the speaker seem to be considerably improved. The midrange
response has been tweaked to produce a slightly less forward soundstage
and to slightly enhance the high-end response. Overall, the Eros MKII is
one of the most musical, least fatiguing speakers I've ever heard (and I
recently spent a few weeks with a set of $10k+ Avalon Eidolon's in my
house for comparison!).
What's more interesting is that the minimum
impedance of the system is now a very healthy 3.7 ohms (occurring at about
200hz) and the impedance phase angle (a measure of how hard your amp has
to work to keep up with impedance swings) is an extremely pleasant < +/-
12.5 degrees. Compare this with many commercial speakers that have a
swing of +/- 45 degrees. Lastly, the impedance swings only from 3.7 ohms
to 6.6 ohms in the 200hz-20khz range. Click on the picture to
the left, for the
big impedance picture.
The frequency response shows -3db points at 37 hz
and 20khz, with an average sensitivity of 90.5 db (+/-<2db) @ 2.83V/1m.
This is easily high enough for my 8 wpc 300B SET monoblocks. Those of you
with low-power, single ended solid state amps should be equally pleased
with the volume you are able to achieve.
Since series crossovers do not easily permit the
measurement of the individual driver rolloff rates, I've included the
Calsod predicted response for the system. Although there are slight
variations between the Calsod prediction and the actual measured response,
please note that there are small variations in component values between
the Calsod-optimized component values and the actual values used (mostly
minor tweaking of the soundstage placement through changing C9 and C5).
Since the enclosure design has not changed, I am not
posting any new information in that regard.
Here is a link, however, to the original schematics. Also, I am not going to prepare a
detailed parts list for the revised crossover -- all of the parts are
easily obtained values which you may obtain from your
supplier of choice.
My only strong recommendation is to use the lowest gauge foil
inductors you can afford (preferably 12 awg copper or better). I tried
this design with regular air-core inductors and they seem to veil the
upper midrange slightly.
Good luck!
