Oops, It has been a while since I updated this page.
I still build the crossovers and a new series is in progress at the moment.
I can not maintain the price level from previous series since component prices have raised a little and the dollar rate is not even close to what it was then. The price level in USD is now approximately 10% higher than it used to be. I would have been happier if this price adjustment was not necessary but my margin is slim and it is a lot of work involved. This project is a hobby project and the work has to be done aside from my regular work and it is sometimes difficult to get the time necessary. Price information is found further down on this page. The rest of the text on this page remains the same as before.
Several months ago I decided to design an active crossover unit for use with subwoofer(s) and the design is now ready for the market (January 2001). I have made this design universal so that it can be used with almost any bass unit even if the main purpose is to improve the sound from my Quad ESL-63/Gradient SW-63 system in which the supplied filter unit is not entirely to my liking. You can read more about my previous solutions on the 'Quad Tweaks' page and these solutions are very good indeed, but the result is improved even further with active filtering of high quality.
The most important part of the crossover design is to make the high-pass filter good enough to disappear in true 'high end' systems, and also to make the high-pass filter steeper in order to improve the sound from the main speakers. Equally important is to have a steep filter for the low-pass section to reduce the coloration from the bass unit, this coloration influences the sound way up in the midrange if the filter slope is not steep enough.
The Quad ESL-63 sound improves quite a lot with a first order filter at 115Hz, and as I imagined the reproduction quality is raised considerably with my fourth order filter at 100Hz. It is my experience that ALL speakers and especially full range speakers and two way systems improve their midrange and treble sound quality very much when the low frequencies are no longer present. This is the problem three way loudspeaker systems try to address, but with passive filtering it is not an easy task because of the impedance variations at low frequencies and the large values on coils and capacitors that must be used. With an active unit the filter response will follow the calculated response almost exactly.
I have followed the discussions on the news groups and it seems that most subwoofers are used only at low frequencies with the main speakers running as full-range and this is not the best approach for audio nirvana. The reason for running them like this is of course the low quality filters supplied with the subwoofers and also the fact that the filters are usually located in the subwoofers and require long cables. The supplied filters are also almost always first or second order.
With active filtering some important factors are introduced for improved sound quality:
The main speakers are relieved from low frequency signals and the sound will be a lot cleaner and the box coloration (with ordinary speakers) will be reduced.
The main amplifier is relieved from low frequency signals and the sound will be cleaner.
The maximum SPL is almost 6dB higher with the same main amplifier if the subwoofer is up to it.
Integration between subwoofer and main speakers is simplified.
With the steep filters used in this unit the subwoofer will not disturb the midrange.
crossover is available as a kit with assembled
and tested PCBs and a box with a special designed front panel. I have tried
to make it as cheap as
possible without compromises in component quality.
Three versions are available:
DIY version with assembled and tested PCB.
Complete DIY crossover unit with box (except for the 12VAC wall transformer).
Complete crossover unit (except for the 12VAC wall transformer).
The front panel is available in two versions, one black anodised with brushed aluminium knobs, and one brass front with black text and matt black anodised aluminium knobs. The box is a black anodised aluminium profile construction, the size is 145/75.5/230 mm (W/H/D).
The PCB is professional made, double sided tin coated with component print. The size is 92 x 177 mm.
I have spent much time to get the signal paths as good as possible (especially in the high-pass section), the routing method is pre-route/auto-route/manual re-route.
The PCB component quality is high:
All resistors are 1% metal film
All capacitors in the low-pass section are 2.5% Wima polypropylene.
All capacitors in the high-pass section are 1% ICEL polypropylene.
All active stages use Burr Brown OPA2604AP opamps.
All outputs use 5% 1uF ICEL polypropylene capacitors for DC blocking.
The power supply section use 2x4700uF/12V regulators/2x4700uF/2x39mH/2x4700uF.
The capacitors in the power supply are Elna RJH, low ESR.
Switches are from Nikkai.
Potentiometers are ALPS carbon film.
Bypass selectors on the PCB use gold plated pins and jumpers.
The box component quality is also high:
All RCA inputs/outputs are gold plated and insulated.
Silver plated low capacitance coaxial input/output cables, except for the ground and bass section where a tin coated thicker cable is used (sounds better this way).
The box is solid metal for good screening.
Internal AC wires with screen for low hum.
The knobs are solid aluminium, giving that nice 'no-plastic feel'
The front panel is solid engraved metal.
The choice of Burr Brown OPA2604AP opamps was made because these are designed for good sound and the construction is also approved by me, with differential JFETs in the signal path, constant current sources and quite high bias current. I use high impedance in all stages to minimize the current swing in the opamps and the result is surprisingly good. Read more in the 'How Does it Sound' section.
Below are some extracts from the Burr Brown OPA2604 documentation.
Fourth order Linkwitz/Riley crossover
at 100Hz, which means that the crossover is phase correct and at -6dB.
Switch selectable (on/off) fourth order low cut Butterworth filter, -3dB at 16Hz (internal PCB switch).
Switch selectable first order low cut at 12Hz, 25Hz, 40Hz. This is meant to be used for correcting recordings with exaggerated low frequencies (quite common) or to increase maximum SPL. It will cause a slight phase shift but I don't think it is necessary to compensate for this, the 12Hz or 25Hz settings are the ones to use for serious listening. With the 12Hz setting the -3dB point is at 18Hz (around 20Hz with dipole correction).
Switch selectable mono for use with single subwoofer or warped records.
Switch selectable 6dB/octave slope for use with dipole subwoofers (internal PCB switch).
Delay filter for adjustment of correct phase at the crossover frequency. The phase delay is adjusted in the range 0-150° (from the actual phase difference, see simulations further down) with a front panel potentiometer.
The low-pass section is phase inverting and the polarity of the subwoofer shall normally be inverted.
The switches on the PCB for filter alterations and output choice are gold plated jumpers, I have searched for reasonably priced switches of good quality and these does not exist. To keep price down and quality up, I have chosen to use jumpers instead. These are cheap and probably affects the sound less than a good quality switch, and the setting is a one time procedure or at least a rare occurrence. With a good manual this is as easy as switch selection. The front mounted selectors for mono and extra low-cut filter use good quality switches.
I have only made a single end version since I do not believe that balanced action is any good with filter units. The reason for this is that in balanced operation you add a negative and positive part in anti-phase and if the two parts are not EXACTLY the same you will add the difference to the output. This is likely to happen in a filter design where the filter characteristics of the two parts will vary slightly due to component tolerances (not critical in single end) and this will add strange sounds that are in no way harmonic.
If there is a demand, I will make a PCB for balanced/single end and single end/balanced conversion. The filter will still be single ended, but balanced inputs and outputs can be used.
part of the construction was SPICE simulations, this is
much more effective than changing part values on a prototype board. The simulations
are pretty exact and in the end you get the result you want.
Above is the 100Hz crossover response (2nd order shown). Note the similar phase responses (completely overlapping), this is a feature of the Linkwitz/Riley crossover and guarantees a correct phase at all frequencies. The crossover point is at zero-phase and -6dB.
The result is bass response almost flat down to 20Hz (-3dB).
The phase delay with the potentiometer in its maximum setting is shown in the graph above. The low pass filter is inverting and if the subwoofers are connected inverted the phase delay is 120°. With speakers connected in phase with the subwoofer the delay is -60°.
The phase delay with the potentiometer in its minimum setting is shown in the graph above. The low pass filter is inverting and if the speakers are connected inverted the phase delay is -30°. With speakers connected in phase with the subwoofer the delay is -210° (can also be written +30°).
Resulting phase adjustment:
inverted speaker connection
-30° to 120°
normal speaker connection
150° to 300°
At 100Hz a 1° phase delay is equivalent to moving the subwoofer 1cm.
With the phase delay adjusted to correct phase at the crossover point the response is as in the graph above.
Gradient SW-63 Simulations (for comparison)
Above you can see the Gradient filter simulated (left graphs), note the bad phase correlation that I have described on the 'Quad Tweaks' page. My tweaked Gradient filter is simulated in the graphs beside (right) and the phase difference is now approximately 180 degrees. You just have to switch the polarity on either the subwoofers or the ESL-63s to have correct phase. The damping of lower frequencies in my tweaked version is not too impressing, still this version sound much better and the small amount of damping also makes the ESL-63s increase their performance a lot.
Summed Response for Dipole Filter with Different Low-cut Settings
The graph above shows the sum of the dipole subwoofer filter and the high-pass filter, the straight line indicates the theoretical 6dB/octave slope to be used with dipoles. This is with the extra low-cut filter set to 12Hz. The result shows a subwoofer response that is only 1.5dB down at 20Hz and -3dB at 18Hz. The Gradient SW-63 manages this at reasonable levels, but for even higher levels the 25Hz or 40Hz low-cut settings can be used.
The graph above shows the sum of the dipole subwoofer filter and the high-pass filter with the extra low-cut filter set to 25Hz. The result shows a subwoofer response that is less than 6dB down at 20Hz and -3dB at 25Hz.
The graph above shows the sum of the dipole subwoofer filter and the high-pass filter with the extra low-cut filter set to 40Hz. The result shows a subwoofer response that is less than 4dB down at 30Hz, but at 10Hz the response is -12dB compared to the 12Hz setting and you can play very loud.
Above are the simulated graphs for the plain subwoofer crossover, the -3dB point is at 18Hz and the output is -20dB at 10Hz.
The phase delay with the potentiometer in its maximum setting is shown in the left graph above. The low pass filter is inverting and if the speakers are connected inverted the phase delay is 40°. With speakers connected in phase with the subwoofer the delay is -140°.
The phase delay with the potentiometer in its minimum setting is shown in the right graph above. The low pass filter is inverting and if the speakers are connected inverted the phase delay is -110°. With speakers connected in phase with the subwoofer the delay is -290° (can also be written 110°).
Resulting phase adjustment:
inverted main speaker connection
-110° to 40°
normal main speaker connection
70° to 220°
At 100Hz a 1° phase delay is equivalent to moving the subwoofer 1cm.
So far I have only listened to my prototype and this sounds very
good indeed. With the extra low-cut set to its lowest frequency the extreme
low bass is sometimes a bit exaggerated, mainly because of the recordings. I
have altered the low frequency response a little in the production series, with
a 1.5 dB damping at 20Hz I think the extreme low frequencies will be perfectly
reproduced by the Gradient SW-63. Still, there will be lots of recordings with
exaggerated extreme low bass and then the extra low-cut filter is very handy.
This is a feature I have missed in subwoofers, studio recordings are often mixed
with speakers that are not capable of extreme low bass reproduction and the
sound engineers don't know what they are doing.
The phase delay potentiometer works very well and it is easy to find correct phase, and when set correctly the integration between SW-63 and ESL-63 is perfect. The bass reproduction is quite a lot better than with my tweaked and bypassed Gradient filter. Wonderful, low coloration, heavy bass without boom and lots of details.
The high-pass section is better than I expected to get from opamps, the sound is a lot cleaner and more detailed than with my previously used single capacitor filter. In the high-pass section I use the same kind of capacitors as I did in my bypass tweak and even if there are more capacitors in series and also a couple of opamps, the sound is improved considerably. Burr Brown has done a great job with their OPA2604AP, and I have no complaints at all.
To sum it up, if you have a SW-63/ESL-63 combination with standard filter you have NO IDEA how good this combination is. With the steep filters used and good quality components I think this filter will make a considerable upgrade with almost all subwoofers and speakers.
Of course I have now listened to the production units and these are actually better than my prototype. I will update the information above later but if you are interested in how the production units sound, you can read the customer feedback which is not as biased as my own opinion.
Feedback from customers
Simple Subwoofer Level Setting
are preliminary and may change but I think my calculations are pretty accurate,
and if the US dollar stays strong, the prices shall be correct. My margin is
slim on the first units, and if you tried finding the components by yourself
for one unit, I think you would end up in the same price range.
Assembled and tested PCB.
Aluminium box with special made front panel.
Six gold plated RCA plugs.
Connector (2.1 mm pin) for the wall transformer.
2.1 mm plug for use with external transformer if you can't find one with correct plug mounted.
Silver plated, low capacitance coaxial cable for internal connection.
Shielded, twinned cable for internal AC wiring.
CD with test tones, 100Hz sinus, sweep 50-200Hz and sweep 200-50Hz.
All this will cost 425 Euro.
This site uses frames, but if you have found this page via a search engine and do not see any frames, you can use the button below to get to the main frame of the MT Audio Design site.