Short answer: you can't pick a crossover from the datasheets, because the datasheet isn't your speaker. The graphs are measured on a big test baffle, and the drivers behave differently once they're in your box. A crossover is also much more than a frequency: it's part values matched to how each driver really behaves, plus getting the level and timing right where they hand over. Guess it and you usually get dips and bumps right where voices sit. The good news is there are three easy ways around it, and none of them stop you building.
One of the questions I get asked the most
It goes something like this: "I've got this woofer and this tweeter, I'm putting them in a box about this big. Where should I cross them, and can you send me a crossover?" Often there's a guess attached, like "I was thinking around 2 kHz?"
It's a completely fair question. Most of us start out assuming a crossover is like a box volume: look up a couple of numbers, do a sum, done. I thought the same once. The honest answer, though, is that I can't tell you, and nobody else can either, without measuring those drivers in that box. Anyone who gives you a number off the top of their head is guessing, and I'd be doing you a disservice if I did the same.
Here's why.
The datasheet isn't your speaker
The frequency response graph on a driver's datasheet is real, but it's measured in a particular way: the driver is mounted in a large flat test baffle, much wider than any bookshelf speaker. That's done on purpose, so drivers can be compared fairly. It just isn't what happens in your living room.
Put the same driver in a 200 mm wide box and a few things change:
- The edges of the box start to matter. Sound spreads out across the front of the cabinet and bounces off the edges, which adds small peaks and dips. Where they land depends on the box width and exactly where each driver sits on the front.
- The bass gets thinner. On a narrow front panel, lower frequencies wrap around the box instead of going forward, so the speaker loses some level below a frequency set by the baffle width. This is called baffle step, and it needs correcting.
- The drivers sit at different depths. The tweeter's sound usually starts from a point further forward than the woofer's, so the two arrive at your ears at slightly different times.
Most of these changes land right in the band where you'd cross a two-way over. So the very part of the graph you'd use to choose a crossover point is the part that changes most once the drivers are in a box.
A crossover is more than a frequency
"Crossed at 2 kHz" sounds like a complete description. It isn't. Two speakers both crossed at 2 kHz can sound nothing alike, because the number leaves out most of the design:
- The part values. The capacitor, inductor and resistor values depend on each driver's real impedance and response in the box, not on a nominal 4 or 8 ohms. The same circuit on two different drivers gives two different results.
- The level. Tweeters are usually louder than woofers, so the tweeter gets turned down to match. By how much depends on the measured levels, in the box.
- The timing. Around the crossover both drivers play at once. If they're in step they add up smoothly; if they're out of step they partly cancel. The slopes, the polarity and the drivers' positions all decide which.
- The fixes. Baffle step correction, taming a cone breakup peak, flattening an impedance rise. Whether a design needs them, and how much, only shows up in measurements.
Passive crossovers explained goes through what each part does and why the online calculators miss most of this.
What a guessed crossover sounds like
A guessed crossover rarely sounds broken. It plays music, it might even sound fine at first. The problems tend to be in the upper midrange, roughly 1 to 5 kHz, which is exactly where voices, guitars and most of the detail in music sit:
- A dip where the two drivers partly cancel, so vocals sound a bit distant or hollow.
- A bump where they overlap too much, so things sound shouty or tiring after half an hour.
- A tweeter that's too loud or too quiet, so the speaker sounds harsh or dull.
The frustrating bit is that no driver upgrade fixes any of that. You can put a tweeter worth three times as much on a guessed crossover and still have the same dip.
That's why, in my experience, a properly designed crossover on budget drivers will usually beat premium drivers on a guessed one. The crossover shapes how a speaker sounds more than the drivers do. Good drivers deserve a crossover that lets them do their job.
A real example: SB Acoustics' Ara
SB Acoustics' Ara is a two-way bookshelf with a 6.5" Satori mid-woofer and a Satori 29 mm tweeter, in a cabinet 391 x 190 x 388 mm (H x W x D). SB designed it, measured it and published it. A few things in it show the point nicely:
- The standard Ara crosses at 3,000 Hz. That's SB's number after measuring the drivers in that cabinet, and it's higher than most people would guess for those drivers.
- The Ara Be uses the same mid-woofer and the same cabinet, with SB's beryllium tweeter, and crosses at 2,300 Hz. Change one driver and SB moved the crossover by 700 Hz and redesigned the circuit.
- The Ara Tx crosses at 2,500 Hz, again in the same cabinet.
- The front panel leans back about 7° to line up the timing of the two drivers, which SB says allows a simpler crossover. The box itself is part of the crossover design.
Same family of drivers, same box, three different crossovers. SB has a measurement lab and this is their whole job, and they still measured each version and designed each one from scratch. That's the best case I know for not guessing.
It also means the 3 kHz is right for the Ara, not for any Satori pair. Put the 7.5" Satori next to the same tweeter, or the same drivers in a different box, and the answer changes again. Which is kind of the point.
Three good ways forward
None of this means you can't build the speaker you had in mind. It just means picking one of these routes for the crossover.
1. Build a proven design (easiest)
Pick a published design where someone has already done the measuring. Build the box to the drawings, use the drivers it calls for, and the crossover matches. SB's kits are made exactly for this: the Ara speaker kit can be built from SB's own cabinet drawings, with SB's assembled crossover pair, or bought with finished cabinets. The rest of SB's range is under speaker kits.
The one rule: build it as drawn. Box size, port and driver positions are part of the crossover design, so changing them puts you back to guessing. Kit or your own design? covers the trade-off.
2. Go active with a DSP plate amp
Fit a two-way DSP plate amp, like a Hypex FusionAmp, in the back of each box. The crossover is then done in software, so instead of buying and swapping parts, every change is a few clicks. You still need to measure to get it right, but you can start simple, listen, measure and refine as you learn, without a pile of spare capacitors.
Two things to know: the amp needs room in the back of the cabinet, so plan for it before you cut, and Hypex amps are ordered in from Hypex in the Netherlands, so there's a lead time (longer over Christmas, when Hypex closes).
3. Measure and design it yourself
This is the most work and the most satisfying. You need a calibrated measurement mic, free software and some patience:
- A calibrated mic. A USB mic like the Dayton Audio UMM-6 is the easiest start, and the IMM-6C plugs into a USB-C phone, tablet or laptop. For crossover work an XLR mic with a two-input interface makes the timing easier to measure; Measuring speakers on a budget explains why.
- REW (free) to measure each driver in the finished box, plus its impedance.
- XSim or VituixCAD (both free) to load those measurements and design the crossover around them.
The order matters: drivers first, then build the real box, then measure, then design, then buy the crossover parts. Measuring speakers on a budget covers the gear and the technique, Beginner Guide Part 6 walks through a first measurement, and Beginner Guide Part 7 covers building the board and what an original design adds.
It takes some learning, but it's very doable, and once you've measured one speaker properly you'll never look at a datasheet the same way.
Build it anyway
I'd much rather someone built a speaker that isn't perfect than didn't build one at all. If you've got your drivers and you're keen, get the boxes built. A box is a box: the crossover can be sorted properly once it's finished, and that's the right order anyway. Your first speaker teaches you more than any article will, this one included.
If you're stuck choosing parts or working out which way to go, send me an email. I'm happy to point you in the right direction.
Quick questions
Can't I just use the crossover point the tweeter's datasheet recommends? That figure is usually the lowest the tweeter can safely be crossed with a certain slope. It's a limit, not a design. Where the crossover actually ends up depends on both drivers in the box.
What about a ready-made 2-way crossover board crossed at a fixed frequency? It's built for an imaginary pair of drivers with flat responses and a fixed impedance. It'll make sound, but it's the same guess with a soldering iron. Why you can't just swap out drivers explains why a crossover only suits the drivers it was designed for.
Can I use the Ara crossover with different drivers or a different box? No. It was designed for the Ara's drivers in the Ara's cabinet. Change either and you're back to measuring.
Do I need an anechoic chamber to measure? No. A normal room works fine with the right technique (gating and near-field measuring), covered in Measuring speakers on a budget.