Ken's Layman's Guide
to Series Cross-Overs
There have been a lot
of questions lately regarding how exactly to design a series crossover
by ear from the curious on the MAD and PE message boards. More people
are hearing about these an want to know what the deal is and how
you come up with them. The typical response is to point them the
the known web sites or publications for perusal. In response to
this I've decided to share with the group on both boards, how I
come up with mine. This is only my experience and opinion and is
by no means intended to be the rule or to contradict anyone else
or their experience with these networks.
The biggest confusion
has seemed to come from the term "zeta". As John Kresovsky describes
it in his excellent papers, zeta is simply a description of the
damping of the circuit. Butterworth values in the circuit give zeta
= 1. Zeta less than 1 is an underdamped circuit, with a slight peaking
overlap at or around the crossover frequency. The circuit will typically
be more capacitive as it uses a smaller choke and larger capacitor
than straight Butterworth. Zeta greater than 1 is an overdamped
circuit, with a slightly sagging underlap at or around the crossover
frequency. It is more inductive than Butterworth, using a larger
choke and smaller capacitor. As would be expected, the lower zeta
values will give a slightly more than 6 dB/octave roll off due to
the peaking. A zeta of .5 gives a nearly 12 dB/octave roll off.
A zeta .7 is nearly 9 dB/octave roll off and as zeta goes up, the
roll off becomes even more gradual. Thus the need for a really beefy
tweeter and a woofer that can play cleanly at the top limit of its
range for high zeta. In my opinion, zeta should not necessarily
be the target for the design. It's the result. The crossover frequency
you choose is the most important consideration. When the right blend
of drivers and crossover have come together, you can then determine
what the zeta is.
First, don't do what
I did...... Research your driver choices extensively. I already
had a woofer and tweeter purchased, and as it turned out they were
not compatible with a first order series crossover at the frequency
I had targeted. This resulted in a nightmare of trial and error
that got me nowhere and I almost gave up. Since I liked the woofer
I had and the tweeter seemed to be the bottleneck, I began to search
for another tweeter. That did the trick. What I recommend is, unless
you have a lot of different woofers and tweeters laying around to
try, ask questions and get the impressions of others regarding what's
out there. Study the measurements for the drivers, whether they're
your measurements, someone else's or as a last resort the manufacturer's
measurements. Ask for advice or help if you have to in order to
find drivers with well behaved, extended frequency response.
Once you've determined
the drivers you want to try, figure out the padding to level match
the tweeter. This doesn't have to be exact as this point. You'll
tweak this later. Also, design a zobel for the woofer to get it's
inductance neutralized and rising impedance flattened. Now determine
where to cross the drivers. My experience is that the typical, full
range two way will need to be crossed between 1800 and 2500 Hz for
and 8"/1" combo and 2000 to 3500 Hz for a 6"/1" pair.
Next, build a pair
of test circuits at each end of the damping extremes you want to
try. Andy's adjustable spreadsheet is very helpful here, if not
absolutely necessary. I build a .7 and 1.2 zeta to start with. With
these two circuits, you have a base to work with that lets you modify
values without throwing off the crossover frequency a lot. Remember
that large changes in the cap value will mean a complementary change
in the inductor value to maintain a given crossover frequency. If
you increase the cap you'll need to decrease the choke and visa-versa.
Large changes will affect the damping. Don't get too crazy with
this if your amp is sensitive to highly capacitive or inductive
loads. Mine obviously isn't either way as I got very "creative"
at times and didn't blow anything up. This is the trial-and-error
stage that can get pretty repetitive and tedious at times and takes,
seemingly, forever. If you can model these with software, you'll
be ahead of the game. In a given tweaking session, don't make a
lot of large changes. This will through you off, big time. Make
small changes, no more than a couple a night and listen. Sometimes
for days if necessary before changing something. I've hit combinations
of values that blew me away at first, but on extended listening
turned out bad. Sometimes the changes will be very bad or good and
very obvious, sometimes hardly noticeable. Now fully tweak the padding
resistor and when you hit on the final crossover you like, get Andy's
spreadsheet again an figure the zeta from that. That's really all
there is to it. You have to listen and listen very closely to what
you're doing. Software cannot determine the final crossover, only
your ears can.
Here are some general
rules to follow. Some are known already, some are my own insight
from experience. I've played with, all together, four woofers and
about five tweeters of varying quality and have determined it's
the drivers and patience that make them work, not magic.
Use drivers with as
flat and wide bandwidth response as you can afford or acquire.
Equalize the impedance
of the woofer with a zobel.
Use a choke with as
low a DCR value as you can afford. This is what protects the tweeter.
A resistor alone in
series with the tweeter works better than a real L-pad. The parallel
resistor in the L-pad seems to dull the dynamics of the tweeter.
If the tweeter needs more than half it's resistance in series to
pad, you need a less efficient tweeter.
Try the LO baffle
step couple that Andy discusses on his web site. (AG: this
no longer exists because I didn't like the effect on the sound )
Put it it the negative (-) return leg of the crossover, after the
woofer, not before it on its positive (+) input. I'm still trying
to figure this one out. It just seems to sound better there. And
it does work.
When figuring the cap/choke
values, use the total impedance of the treble section. If you use
an 8 ohm tweeter with a 3 ohm padding resistor, figure for an 11
ohm tweeter.
Mount the drivers in
a vertical line, as close as possible. Trust me on this one. I'd
bet aligning the voice coils of the drivers would be a plus too,
if you can. Otherwise, try reversing the polarity of the tweeter.
Lower damping, low
zeta circuits have a forward sounding presentation. Really up-front
and in the room sounding. Probably the more "accurate" sounding
in the midrange. Hence the favouring of zeta = .7 by many who listen
to a lot of live music. Higher damping, higher zeta has a more relaxed,
laid back "euphonic" sound to them and seem to work better with
typical commercial and multi-miked recordings. I apparently prefer
a zeta of 1.1 to 1.3, typically. Works well for vinyl reproduction,
which is my primary musical source.
Don't get stuck on
zeta. It closes your mind to the combination that works best for
you drivers, taste and room.
You'll know you're
close to the right frequency when the image "locks in". I don't
want to sound vague and I don't know how, exactly to describe it
but you'll hear what I mean. It will literally float between and
around the speakers when right.
I'm done. Have
fun and happy listening! That's what this is all about...
