It is difficult to integrate dipole panel with omnidirectional woofer. Change in directional behaviour can be most easily heard in small rooms where room modes tend to mess up low frequency region anyway. By using dipole woofer directional characteristics are closer to panel. Besides room modes affect sound little less because of higher directivity.
Many people have said that dipole bass is more natural less resonant and colored than enclosed bass. Dipole woofer has usually less group delay and less box resonances. In some cases lower distortion levels have been measured as well. This doesn't really mean that dipole would be superior to enclosed woofers. At least nobody has really proven this to be true in general. Enclosed woofer can be made in non resonant box, with low Q-value so that group delay is minimized, and distortion at same SPL is typically lower if same element is used. It might just be that there just happends to be more low performance examples of enclosed woofers.
Anyway dipole was tested and found to be much easier to get working with panel than enclosed woofer. Quality of bass is really good as well, even if element is not so expensive.
Structure and Electronics
Electronics is similar that have been used in earlier versions. Polarization is done with voltage doubler ladder and signal voltage is raised with tube amplifier output transformer during tests and are replaced with similar toroidal transformers that were used earlier.
Panel structure is similar that is used in first hybrid, only little smaller. Dipole is in H-baffle that is designed so that crossover can be set to 300 Hz region. Panel has clothcover to reduce bass resonance. Without extra aqoustic resistance woofer could cause excessive excursion to diaphragm at resonance frequency.
Woofer element is 12" Peerless from SLS-series.
Dipole woofer correction must be done with active filter and extra amplifier channel. Passive realizations aren't well suited for this application. Active filter is also much more versatile.
My way of working produces usually more than one prototype. I like to test new ideas and in this case different filters. It is annoying and time consuming to build whole filter board for every prototype. One option is to use some digital adjustable filter in development phase and replace it with analog version in the end if necessary. I solved the problem by building mothercard with power supply, soft start, output relays and input/output buffers. Mothercard has 6 connectors for filtercards. Filters itself are very simple and new versions can be made with tiny modifications to earlier versions.
Device is simple but proven to be very useful in active speaker development.
Element responses were measured at distance of 1.5m. Resonance peak at 50 Hz is standing wave problem of the anechoic room.
Filtered response. Panel equalization is typically rather easy task. In hybrid panel low frequency resonance is well below crossover frequency so it can be ignored. In this panel stator contstruction causes resonance at 15kHz, that can be filtered out with notch or with simple low pass. Dipole woofer frame causes peak at the region where dipole operation begins. Dipole correction can be made with 1st order lowpass for both elements. Peaks may require additional filtering.
Equalized element responses are crossed over with 2nd order filter. Two 2nd order all pass filters are used to match delays at cross-over region. Cross-over frequency is as high as possible to increase overall sensitivity. Further increase in cross-over would increase woofer distortion to intolerable levels.
Large panel element is highly directional so responses to different angles are less significant that they are with normal dynamic elements. Above image has responses at 0 20 and 45 degrees. They are all measured with an earlier filter prototype.
Integration between panel and woofer has succeeded very well. SLS woofer seems to be well suited for dipole operation as well as operation with ESL panel.
Sound is something that can be expected from ESL panel. Little hard and cold but sharp, open and accurate. While original version of Portal had irritating midrange it is not the case with this panel so apparently IM distortion is in control. Speaker is well suited for all kinds of music.
Large panel element and dynamic woofer element can never be perfectly matched, but by using dipole woofer problems are minimized. Horisontal directional patterns are more or less similar, but high panel element has much greater vertical directivity. Difference between Portal and dipole hybrid is clear. While full range becomes darker when listened off axis, hybrid panel sound more like subwoofer. This may cause problems in very small rooms, but doesn't seem to be a problem in larger listening room.