I wanted a speakers that could be used in small poorly damped concrete room and so that they are easy to be used with subwoofer. Structure had to be simple because there were few other people that wanted new speakers for their hometheaters as well. Good approach was assumed to be small two way speaker in closed box. Some two way structures with various elements were tested with more or less encouraging results. 4 inch element in closed box was considered too small to be used without subwoofer and subwoofer would need rather high cross-over frequency. 6 inch element was considered suitable. It goes down to about 70 Hz in closed box and thus it should be easily combined with sub.

Dynamic loudspeakers

Element or element pair that was chosen is G17 RECOAX/TVA-GR coaxial element from Seas. Several successfull commercial speakers has used that or similar element.

Box size should be about 6 to 8 litres. At first glance element frequency responses seem to be far from ideal. Significant resonances at woofer element and tweeter response is what it is because of woofer cone.



Box was kept simple. Materials used are 20mm birch plywood, 22mm MDF, 25mm solid birchwood. Some of them have bracing inside and some have sand-epoxy layer on their inner walls. All boxes have volume about 8 litres and dimensions nearly equal so they can be easily compared.

Dynamic loudspeakers

Box speakers have their usual box sound especially at upper bass region. So do these. I am not saying that these speakers sound boomy but they are boomy compared to ESL panels. That made me to study how much box produces sound and can this boominess be due to wall resonances.

Measurements and crossover

AB listening test revealed some differences between speakers. Measurements confirmed those differences but their origin was in element manufacturing tolerances, not in enclosure material. Measurements also revealed different box resonance frequencies on different materials. Even if we could feel those differences we weren't able to hear them.


Speaker has very neat power and free field frequency responses. They are easy load to amplifier. Minimum impedance is about 7 ohms. All Seas coaxial elements have small dips at highest frequencies when measured directly from front of the speaker. Maximum sound pressures aren't so noteworthy and in closed box element won't produce any low frequencies but that was intentional choice in this particular project. In small room and with subwoofer SPL-levels are adequate.

Speakers seem to be suitable for most music. Some experiences with rising frequency response made them sound better with music material like jazz but with most pop-music sounded too bright. Stereoimage is accurate but very different compared to large panels. Sounds mixed to center keep firmly in there and sounds mixed to sides come from there but there seems to be gap between sides and center. Panel has more uniform sound field when well positioned but centerimage is not so accurate.

Midrange quality was improved when amplifier was changed and second order filter doesn't sound so cold and metallic it used to sound. Mid and high frequencies aren't so detailed than they are in ESLs but different frequency balance and directivity might explain at least some of the differences. Bass doesn't go very low but I get used to it very fast. It is not boomy or irritating in any way.

Althought free field response of those two presented filters is nearly equal is there significant difference in tonal balance. First order filter might be better for those who hasn't been used to bright ESL sound.

Coaxial Speaker Measurements

Arto's Electrostatic loudspeakers

All frequency responses are measured in anechoic room at TUT.

Element response is shown below. Peak at 55Hz is due to measurement room. Near field response show no sign of it.

Arto's Electrostatic loudspeakers


Several crossovers were tested. Chorus AW-8 crossover was tested first. It is rather complex and sound was considered rather flat. Different second order filters were tested with more or less rising free field response. One of the second order filters that was satisfactory in listening tests is presented below. Filters are designed using only one inductor. I wanted to get rid of them because they have few undesirable properties like resistance and magnetic coupling. They are also large and expensive. Tweeter level can be easily adjusted by adjusting series resistance if it sounds too bright.

Arto's Electrostatic loudspeakers

Arto's Electrostatic loudspeakers

Responses are measured in angles of 0, 20, 45 and 60 degrees. Level is 6dB attenuated between measurements.

Coaxial speaker polar pattern (PDF)

It is also possible to use first order crossover. One designed version is shown below. It is used in most speakers build. This has little less brigtness than previous version. Greatest difference between these two filters is crossover frequency. In second order filter acoustic crossover is little less than 2 kHz and in first order filter it is over 4 kHz. Woofer resonances are less attenuated and directivity at middle range is greater with first order filter.

Arto's Electrostatic loudspeakers

Wall resonances

Wall resonances were tested using with microphones and accelerometers. Initial tests showed that MDF box without bracing had sharp peaks at 500Hz and 800Hz. At those resonances, sound output was measured to be only 10 to 15 dB lower than sound levels from element. Normally output was 30 or more desibels lower. Plywood box had only one large peak at about 400Hz with few desibels lower level and lower Q-value than MDF box had. Box had no internal bracing either but had 40mm thick front wall. Yellow curve shows sound output from MDF enclosure.

Arto's Electrostatic loudspeakers

Sand epoxy mixture was added to two of the boxes. Layer thickness is about 1cm and it is only in two sides of the box. Notice different frequency scale. Resonance that was most easily felt from box surface was at 120Hz although sound level measurements didn't reveal any significant peak at that region. This resonance peak was significantly reduced by adding sand-epoxy to the walls but it had only little influence on 700Hz area that made most of the sound output. So at least in this case adding sand wasn't very successfull method to eliminate wall resonances.

Arto's Electrostatic loudspeakers