Var. 1: This design calls for an 8 Ohm woofer and a 4 Ohm tweeter. How is that possible?
Var. 2: I have an 8 Ohm woofer. What happens if I use a 4 Ohm tweeter with it?
Var. 3: This design has a 4 Ohm tweeter in it. My amp says it's only safe for 8 Ohm speakers.
I feel obligated to write an answer to this question because even I asked it back when I was just a neophyte. The experienced DIYer whom I asked responded to my inquiry by not writing back :-(
Really, the answer is pretty painless. We can answer it in varying levels of detail, depending on how much you want to know.
a 2-second answer, think of it this way: you need to remember that
tweeters are almost always louder than woofers. In order to bring the
level of a tweeter down to match a woofer, we need to attenuate it
somehow; the most common way is to add impedance in the crossover
(often using resistors in some sort of L-pad or series configuration).
Thus, we have just raised the impedance of the tweeter and it's no
longer 4 Ohms, but rather something higher. Your amp is safe, and the
drivers are now level-matched. Hooray!
Would you like me to explain further?
For example, the Vifa Logic AC25SG05 tweeter (used in the Orient Express and China Syndrome)
is a fairly loud tweeter. Its sensitivity is somewhere around 93 dB @
1 Watt. Let's compare that to the woofer used in the Orient Express,
the Silver Flute W14RC25, which has a sensitivity of about 87 dB @ 1
Watt. But that's not the end of it! Due to Baffle Step Losses,
we will lose another 4-6 dB, assuming the driver is mounted in some
sort of enclosure in a room (and not flush-mounted in a wall). So in
reality, the woofer is more like 82-83 dB @ 1 Watt -- that's a whole 10
dB less than the tweeter! (Incidentally, this final number I arrived
at (woofer level, after Baffle Step Losses), is usually the actual
sensitivity of a speaker quoted on its spec sheets... or at least it ought to be. Some companies "inflate" their sensitivity rating by not including baffle step losses into the calculation.)
you just plugged this pair of drivers into your amp without padding
that tweeter down somehow, the speaker is going to sound completely
irritating, what with that tweeter playing so much louder than the
There are a myriad of ways designers attenuate drivers in order to match their levels. Sometimes, the use of capacitors and inductors in the normal High Pass filter offers enough impedance to the signal to bring it down to a usable level. But much of the time, additional attenuation is needed, and that is where careful use of resistors comes in. A designer will usually place resistors in series with the filter, or sometimes use an L-pad configuration, and sometimes both. Which way he lays out the resistors depends on how well it accomplishes his goals of achieving a flat frequency response, proper acoustic phase integration with the woofer, and also keeping the system impedance of the speaker in-check.
But there's one more important thing you should know!
When loudspeaker manufacturers give you a "Nominal Impedance" of a speaker, the number is incredibly misleading and a wild guess at best! The actual impedance of the speaker is really all over the map (depending on the frequency), and most likely dips below the quoted "nominal impedance" at certain frequencies.
Here is the measured impedance graph of the Polk Audio RTi A1, a speaker whose spec sheets claim that it is an 8 Ohm nominal speaker.
What's important to notice here is that the actual System Impedance of a speaker changes, depending on the frequency. This is completely normal, and all speakers do it. (For instance, the fact that it has two spikes in the bass response tells you the woofer is in a vented enclosure. The hump around 1500 Hz is showing where the crossover in the speaker is making the transition between the woofer and the tweeter)
What is more significant than the "nominal impedance" of a speaker are the impedance minima, which are pretty low for a retail speaker. It dips down to just above 4 Ohms at 50 Hz, and below 4 Ohms (warning!) at 200 Hz. Nonetheless, Polk Audio sees it fit to call it an "8 Ohm" speaker. If it were my design, I'd say I still had some work to do, getting that impedance back up into a "safe zone." But Polk is one of the most successful speaker manufacturers out there, so I'll assume that this speaker is not overheating people's amps and sending them into protect mode around the world, and everyone is happy.
Now, let's bring this discussion back on home
Just like the System Impedance of a speaker as a whole, the actual impedance of a driver is almost always complex.
Let's get back to our examples. Here is the imedace graph for the Vifa AC25SG30 we talked about above.
As you can see, the impedance of the driver is not simply just 4 Ohms all the way across the spectrum. There is a notable bump at 1200 Hz; this indicates the tweeter's resonant frequency. Above that, the impedance of the driver gradually rises to about 6 Ohms. This rise is completely normal, and all conventional drivers do it, because their voice coil acts like an inductor at higher frequencies.
While we're at it, let's look at the measured impedance of the Silver Flute W14RC25.
I had to scale this one down a bit so you could see more of the curve, but as you can see, the actual impedance load of this driver is far from the simple 8 Ohms as quoted on the spec sheet. The massive spike around 45 Hz (goes way off the chart) shows the resonant frequency of the driver, and just like the tweeter, there is a gradual rise in impedance as the pitch goes up. Again, this is caused by the voice coil of the driver acting like an inductor at higher frequencies. Why does Silver Flute call it an 8 Ohm woofer? I dunno; because they felt like it I guess... gotta pick some number. Logically, if you look at the impedance response between 150 Hz and 1 KHz, the impedance is basically 8 Ohms, and that is a large portion of the useful range ("passband") of this woofer.
Yes, it is true that some drivers actually do exhibit a nearly flat Impedance, but these are unusual. For instance, here's the measured impeance of the Bohlender Graebner Neo3 PDR.
Planars, electrostatics, and ribbons don't use voice coils like traditional drivers, so there is no "inductance rise" in the impedance. However, while this looks nice and flat now, once we add in the actual High Pass filter to use this driver in a speaker, the impedance will be all over the place, just like all the rest of the speakers in the world!