Many of you are new to building your own speakers. The way many
of us started is to look in a catalog or, nowadays, on a website and
see how inexpensive very high quality speakers can be... but with the
catch that you build them yourself. Many of us think "fine! no
problem! I can do that!" and we set our hearts and minds on building a
set of audiophile-quality speakers. Does this rings a bell?
You see a great set of drivers in a catalog, buy them, and then
realize that you need a crossover. Maybe the drivers aren't even
all that great -- suppose you bought one of the major parts supplier's
"house brands." If you're like me and you buy from
Parts Express or
Madisound, you might try
something like the Dayton 295-305 6 1/2" woofer and the 275-070 1 1/8"
silk dome tweeter. At $16.50 for the woofer and $15.50 for the
tweeter, how can you go wrong?
Well.. the truth of the matter is you can't go truly wrong with these
drivers at these prices. The real question is how good can they
sound? Based on my recent project using these drivers in a
simple 2-way design, they can sound great. But without using
rather expensive measuring hardware (CLIO, $589) and sophisticated and
complicated modeling/simulation software (Calsod 3.10g, $269) and
putting in a lot of time learning to use it and design the system, how
can you get a good response? Moreover, what IS a good response?
Is a flat frequency response the ticket to heavenly sound?
In this brief discussion, I look at 1) is a flat response really
what's important? Or is it HOW you achieve a flat response
(hint, I wouldn't have mentioned that if it wasn't where I'm heading);
and 2) what happens to those of us who try to use fancy drivers with
stock crossovers? Is it pointless? I'll answer that one outright
-- it's not pointless, but it's more complicated than one might hope.
Lets start by looking at what happens when we take our Dayton drivers
and hook them up using a standard Linkwitz-Riley crossover at 2000hz.
Most people will tell us that with a 6 1/2" woofer, this is a pretty
safe crossover. What the graph below shows, however, is that
with just stock components (or a pre-made crossover), the results are
going to be less than satisfactory.
What
Figure 1 shows is our drivers connected with normal polarity to a
stock 2 khz LR crossover. I've also added 3db of tweeter padding
(an "L-Pad"). The published sensitivities of these drivers would lead
you to use a little less padding, but lets assume that you learned
enough to do this off the bat, or used a variable L-Pad. If you
look at the BIG 8 db hump between 2 and 3khz, you see where our
problem comes in. The Dayton 295-305 has a peak in its
rolloff... a very big one. If we look at the raw response for
this driver, as shown in figure 2, we see the actual peak occurs in
the 3-4khz range.
If you
listen to this speaker with the standard crossover, you're going to
think that this is a very harsh and nasally sounding woofer... and
then you're going to think you should have spent a lot more money and
bought a Scan Speak, Seas, or other exotic driver. While these
expensive drivers have the potential to sound fantastic, they're no
easier to work with... you would have just had a more expensive harsh
and somewhat unpleasant sounding speaker (although not always, the
Vifa P17 has a very smooth frequency response and can be quite easy to
work with without a lot of testing equiptment).
One
thing that Parts Express recommends that most DIY'ers should do
with their 6 1/2" woofer is to install a zobel filter. With just
a little impedance compensation, our stock system will will respond as
shown in figure 3.
This,
of course, looks pretty good. It's a nice flat response.
The 4db of baffle step (rise from 200hz up to around 1000hz) won't be
very bothersome, especially if the speaker is close to a wall or on a
bookshelf. When we reverse the phase of the speakers to "test"
to see how good we did, we'll get the response shown in figure 4,
which at first glance, looks a-ok.
Here's
the part where we pat ourselves on the back and think that we have a
nice speaker... sure it's a little grainy with some less the
fully-satisfying upper midrange qualities, but we did it! And
you should be pleased with yourself, as for the money, it's not a bad
sounding little speaker.... and since we didn't have testing
equipment, we didn't notice that the null is well below the crossover
frequency.... but we just figured that the sound of the speaker was a
little nasally and next time we'll try some expensive woofers and
it'll get better... or will it?
If we
look a little closer at the responses of the individual drivers, we'll
notice that maybe slight harshness in the midrange isn't native to the
woofer after all. In fact, we we look at the response for the
woofer, we see that same hump we discussed earlier... just attenuated
somewhat due to the impedance compensation. Like I said before,
we could stop here and have a nice little speaker... but what happens
when we buy those Focal drivers and they don't sound like $150 woofers
even with a nice, relatively flat frequency response? It's in
the individual response curves that we find the answer, as illustrated
in figure 5.
The
hump we see above at around 2800hz is giving us a little 'peaky' sound
in the woofer. Does it ruin the sound of the speaker?
Probably not... but removing the hump can sure make a difference.
If you don't have testing equipment and modeling software should you
keep reading? Sure... but this is where the story ends for
improving the sound of these drivers, except for tweaking by ear and
experimenting, without such equipment. Consider the rest of this
discussion a caution against purchasing expensive drivers without
planning to get a crossover network designed for them or building from
a proven kit. Getting the drivers to produce a flat overall
response, by ear, is certainly possible... but unlocking any driver's
potential, even an inexpensive one like the 295-305, takes a little
more work... but it is certainly worth it!
Figure
6 shows the system response and individual driver response for these
same drivers after I've used a 3rd order filter for the tweeter and a
2nd order filter (with impedance compensation) for the woofer... and
added a parallel resonance filter, or peak filter, to the lowpass
section. Some will argue the various crossover designs and their
advantages, I chose this topology (i.e., the parts used in the
crossover and the way we put them together) because it seemed to be
the least invasive method of taming the woofer's sound. I tried
a 3rd order crossover for the lowpass, but it seemed to "choke" the
speaker, substituting one problem for another.
Here
we not only see that each driver is adhering very closely to the
target 4th-order Linkwitz-Riley slopes and producing an acoustical
crossover at 2khz, but there is very nice phase alignment of the
drivers throughout the crossover region. Most importantly, the
peak is gone. While not particularly flatter than what we saw
above using the stock crossover and impedance compensation, this
design has a substantially cleaner and more open sound. As well,
the "nasal" quality of the woofer is gone (mostly -- the Dayton 6.5"
woofer is simply not capable of producing the same quality of clean,
smooth midrange as drivers cost more money, this driver's strength is
its' ability to produce deep bass response at a level that other
similarly priced, and even more expensive drivers, can't), making this
driver combination sound as good, or better, than drivers costing
twice as much. In a side-by-side comparison, this design was
overwhelmingly preferred over a 2-way design using the Audax AP170Z0
and TM025F7 tweeter. Am I just trying to sell Dayton speakers or
something? NO! The truth is that the AP170 has a nasty peak around
3600hz that was not compensated in the speaker we compared with the
Dayton 2-way. The AP170 sounded a bit harsh, in comparison, and quite
forward.
Looking at the reverse-phase null, illustrated in figure 7, we see
it's nice and deep. While there's disagreement to some about how
important this is, it does show that our drivers are meeting the
target slopes with good phase tracking. To me, that means our
speaker will sound very coherent (i.e., you won't hear individual
drivers... you'll hear one smooth source of sound.) This quality
is sure to be reflected in the sonic image and soundstage the system
produces.
Ultimately, many of you are going to want to build "world-class"
speakers. Someday, I hope to as well. These are just some
of the lessons I'm learning on the way to being able to consider such
a lofty goal. Until then, I'm realizing there's a lot more to a
great speaker than great drivers... now I've learned that there's more
to a great speaker than a flat frequency response... as with so many
other things, it's how you get there that matters.
