- what it takes to make a crossover -

Before anything else: You cannot change the tweeter - or any other driver - in any of the constructions shown here without needing a new crossover. Read here on how "close" tweeters appear to be - and yet, requires very different crossovers: SS 9500, 9700 and 9900 tweeters

If you change front panel dimensions, placement of drivers and cabinet tilt in the constructions shown on these pages, you're on your own. The crossover will no longer work as intended, so please do not ask because I would need your speakers in my workshop to adjust the crossover.

Most common mails go like this: "I have these and these drivers; can you please make me with a crossover?" I can't, because to make a crossover I need this:

- Frequency response and phase from actual drivers mounted on actual front panel.
- Impedance/electrical phase of all drivers mounted in actual cabinet.
- Acoustic distance from drivers to suggested listening point.
- In short: I need your finished speakers on my workshop bench.

In very few cases I may have the actual drivers and a suitable test cabinet and I can set up the drivers and do what's described above. We're then into diy-service work and it usually takes 5-15 hours depending on overall complexity. This only provides a crossover that actually works on some basic parameters like frequency response, phase integration, power response, etc., but no guarantee it also sounds good.

The question is whether we can make a crossover at all without measurements
- and the answer is NO. It cannot be done, and crossovers cannot be calculated.

We cannot use manufacturers' data sheets to simulate crossovers either. The SPL data may be taken from an infinite baffle that does not in any way resemble your cabinet. These data do not contain proper phase data, etc.

Yes, you can set up a speaker with a crossover based on theoretical values that will ensure you won't burn your tweeter or midrange and you may sit back and think this is a great speaker - and maybe it is - but in most cases you wouldn't like it if you saw its measuring performance.
I have a friend who refuses to let me measure his speakers because he knows he won't like them if he sees how they perform. You may object that this person is perfectly happy with his speakers and so what? And this is perfectly true!

- nothing can compete with "good enough" -

I see a lot of people spending minor fortunes on expensive drivers and "high-end" components and never get the full potential of the drivers due to improper crossovers. The crossover is the heart of the speaker and can be made in numerous ways and to great extent determines the behaviour of the drivers. Only minor changes to the crossover can turn a great speaker into mediocrity and visa versa.

I receive mails from people having made a construction from some presumed "allround" series crossovers that floats around on the web. They're awful! They won't work, so don't waste your money - and please read 1st example below.

Example #1

1st example here is from a visitor I had, bringing in a beautiful Accuton 2-way from C220 and C23 drivers. The tweeter was hung on a hook on top of the cabinet (three O-rings), so no baffle support for treble. Usually a free "flying" tweeter measures pretty bad and this one was no exception:

Left: Tweeter response with (blue) and without (red) baffle support.
Right: SPL from drivers with no crossover attached.

Response from speaker with new crossover - tweeter still without baffle.

Below you can see the response from a series crossover my visitor had found on the web and which was recommended as a generally applicable crossover. Shame on whoever launched a load of crap like this! It's a disaster! Two C220 and two C23 Accuton drivers are serious money and throwing in some phantasy web-nonsense like this is a total waste of money.
Tweeter level is some 5dB too low and the gigantic peak at 3.5 kHz is hard to ignore. Right where the ear is most sensitive. I can tell this speaker had a weird sound!

Example #2:
Making the 2.5 clone from calculated crossover components:

Let's take a look at the 2.5 clone: This is basically an 18/18 dB filter at 3 kHz. Try to calculate component values for a 18/18 dB filter at 3 kHz for two 8 ohm drivers and compare to the actual crossover:
Try this site: http://ccs.exl.info/calc_cr.html.

The calculation sheet cannot take into account the impact of bypassing the 1.75 mH coil with the 47 ohm resistor. Not to mention baffle-step compensation. The calculation sheet is not taking into account the needed tweeter attenuation - nor tweeter polarity for that matter.

So, the real 2.5 clone filter does not look very much like what we have calculated. If we made a 2.5 clone crossover based on theoretical component values we would get a speaker that would yell off our heads from having a midrange some 5-10 dB higher than intended and probably some serious problems in the crossover region.

This is how it looks if we make a 2.5 clone from calculated crossover components. Upper midrange and lower treble is some 10-11 dB above average target SPL level.

OK, so we have heard that tweeters need attenuation and we add a 5R6 to the tweeter:

Well, we got the tweeter level down, but still a 10 dB peak at 3 kHz!

- And we have heard that we may need to invert polarity of the tweeter:

Hmm... Now we have a 18 dB slope from 3 kHz to 4 kHz... I wonder how this sounds...

Above when the 2.5 clone is right - for better or for worse.....

My best advice is to choose some well documented commercial kits or well documented diy-constructions from the internet if you want to enter the fascinating world of DIY loudspeaker construction. Take great care if you change cabinet construction.

Happy hours in the workshop!