I’ve now finished the cross-over simulation and I’m moving on to the next stage of building the first cross-over prototype and starting the listening tests.
The cross-over:
(click on picture to zoom)
The cross-over is very simple and uses a LR4 filter topology (acoustical) between the tweeter and mid and a LR2 topology (acoustical) between the mid and woofer. The cross-over consist of only 12 parts and all sections are second-order electrically, besides the tweeter section which is third-order electrically in order to achieve an acoustical LR4 frequency roll-off.
Note! The woofer’s are connected with reverse polarity.
The tweeter level can be adjusted by changing the value of the single resistor R1. R2 and R3 can be replaced with one resistor if it can handle >20W.
I don’t reveal the actual component values until I’ve evaluated the cross-over prototype. The cross-over schematic is only used here to inform how the cross-over looks like. Depending on the listening tests and further measurements it can be changed little or completely.
Frequency response simulation:
As in most of my designs the cross-over is optimized from the 15deg off-axis measurements and thus, the loudspeaker is intended to be used with none or little toe-in when setup.
Using the 15deg off-axis measurement generally means that you have a less pronounced baffle diffraction influence in the frequency response when optimizing the cross-over. Generally I’m optimizing the speaker cross-over to have equally good phase behavior between 15-30deg off-axis frequency response, rather than having a good phase response at the on-axis measurement and a deteriorating phase behavior as you go off-axis.
When optimizing around the 15deg off-axis measurements, the on-axis frequency response tends to be a little “hot” in the mid and upper treble. By using the appropriate amount of toe-in, I think this slightly added frequency response gives a more presence and airiness to the sound and are usually beneficial and desired.
However, each design and the drivers used in it are unique and sometimes it might not work out properly. The cross-over simulations and measurements alone doesn’t tell if it’s good or not and I consider this and the tweeter level adjustments to be the voicing and fine tuning part of the loudspeaker design.
(click on picture to zoom)
Left: 15deg off-xis frequency response (no smoothing applied).
Right: 15deg off-xis frequency response with mid-range reverse polarity.
The mid-range driver has a built in “BBC” response dip between 2-3.5kHz. The tweeter has a very extended and smooth frequency response, especially considering that this is a co-axial driver unit. The loudspeaker has a system sensitivity of about 88dB @ 2.83v/1m.
When reversing the mid-range polarity we can see very deep notches around 590Hz and 2.8kHz at the cross-over frequencies. The deep notches indicate a very good phase response tracking and summation between the driver units around the cross-over frequency.
(click on picture to zoom)
Left: Frequency response 0-60deg (no smoothing applied).
Right: The frequency response curve is an average of four measurements (blue): on-axis, 15, 22.5 & 30deg and six measurements (red): on-axis, 15, 22.5, 30, 45 & 60deg.
The loudspeaker has a smooth power response with a slightly downwards tilting response at higher frequencies. The tweeter has an extended frequency response with very wide off-axis dispersion.
(click on picture to zoom)
Left: Four different tweeter levels.
Right: System impedance simulation
The resistor R1 can be changed to fine tune the tweeter level. The level 2 is the “reference” level used in the above frequency and cross-over simulations. The minimum impedance occurs @ 120Hz and is slightly below 3 Ohms. The system sensitivity of 88dB @ 2.83v/1m is mainly achieved by the low impedance driver units used in the loudspeaker. A 4 Ohm stable amplifier is recommended.
Summation:
As I said before, I don’t intend to make a clone of any of KEF’s loudspeakers and I don’t have a clue how the cross-over for KEF’s loudspeakers looks like, even though my chosen cross-over frequencies seems rather similar to those used in the R500 loudspeaker. It just happens to be very suitable frequencies to cross-over at. I would say that a suitable cross-over window is about +/- 100Hz from that I have chosen, at least the drivers seems to be engineered in that way.
With that said, the KEF driver is really great and I’m eagerly waiting for the cross-over components I’ve ordered for the first prototype and I can’t wait to have a listen to the Concentric Three – M5 loudspeaker! 🙂
For further details of the driver unit measurements see:
The KEF R500 loudspeaker using the same mid/tweeter and similar sized woofer’s:
Independent KEF R500 loudspeaker measurements: