22D7: four MDD systems for a 3D effect with omnidirectional speakers

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
The 22D7 project uses four Faital-Pro 3FE25 speakers in MDD configuration. Two are optimized for omnidirectional reproduction of high frequencies, project 22C7. Two are optimized for low frequency reproduction, project 22L7. In total there are 32 coherent sound emission points, distributed in the space in the three directions of the XYZ Cartesian axes.

I don't know if audiophiles can consider the reproduction of the 22D7 project as 3D, I don't have HiEnd devices to compare. A physical basis for a 3D effect exists. This is the first project with four speakers and it is statistically unlikely that it has already guessed the optimal configuration. Already so the sounds are easier to recognize, on single notes the attack and the decay can be sensed, sentences and words are more easily deciphered, the number of details increases and more times I have had the impression that a song already heard lasted longer. long.

In the field of DIY it is very difficult to obtain a sound field with these characteristics and an expense in material reduced to around 100 euros. Another important detail: the construction is simple for an expert DIY.

22C71L8 I wrote a post on an alternative 3D project to this
other documentation is published at 22C71L8.
The 22C71L8 prototype uses only one driver per channel with waveguides applied to the two sides of the cone.

My site is: Claudio Gandolfi - MDD
My current working hypotheses are described on the page acustica.

The last projects reported in the full range and planars-and-exotics section:
54m42 54m42 an MDD project (42 alveolar polypropylene wave guides on one driver)
other documentation is published at 54m42

34c9 Compact aluminum project
34c9 a MDD full range speakers.
34c9 - MDD project with aluminum pipes. Version: Compact.

66c9 66c9 a low-cost MDD project (paper waveguide for educational use)
66c9 - MDD project with paper pipes. Version: Compact.

22M7 Simple and inexpensive wide-band rigid PVC project.
other documentation is published at 22M7.

MDDHX135 an MDD project with the HX135 driver
other documentation is published at MDDHX135.

MDD3FE25 an MDD project with the 3FE25 driver
other documentation is published at MDD3FE25.


  • 22d7 32 colla.jpg
    22d7 32 colla.jpg
    113 KB · Views: 227
  • 22d7 18.jpg
    22d7 18.jpg
    112.9 KB · Views: 231
  • 22D7 FR THD Damped Left.jpg
    22D7 FR THD Damped Left.jpg
    86.8 KB · Views: 229
Last edited:
Have you tried using a bend for the longer pipes? Say you want a 3m pipe and do a nice 180 bend at 1m and then at 2m’s, so a 3m pipe would only be about 1m high
I thought of various configurations with curved waveguides, I preferred to try first with only straight paths because I want to be sure of the consistency of delayed wave fronts. Now that I have clarified the ideas about the propagation of high and low frequencies in the room I could try curved guides in the optimized unit for bass.
If you are going to try bending pipe. Let’s say round pipes, you need to fill them with sand or water and freeze it before you start the bending. That way the inside bend is nicely done
I did not know the sand system, it will be useful to me.

Soon I want to try to change the lengths: from 1 to 2 meters in the 22L7 project, up to 1 meter in the 22C7 project. If it works as I think I could use a flexible spiral sheath for the downhill section.
new version 21D7

I published on the site the measures of the new version 21D7. The dimensions are similar to the previous ones, the maximum length is limited to 2 meters. Now the unit optimized for high frequencies (22C7.v2) reaches a length of one meter and has the resonances distributed over two octaves, Lmax = 4 * Lmin. The logarithmic series is 305, 371, 453, 552, 673, 820, 1000 mm. The unit optimized for low frequencies (21L7) reaches a maximum of two meters but I have increased the length of the shorter waveguide and the resonances are distributed over an octave, Lmax = 2 * Lmin. The logarithmic series is 1104, 1219, 1346, 1486, 1641, 1811, 2000 mm.


  • 21D7 FR THD Damped Left.jpg
    21D7 FR THD Damped Left.jpg
    86.8 KB · Views: 155
  • 21D7 FR Damped Left - 21L7 green - 22C7v2 blue.jpg
    21D7 FR Damped Left - 21L7 green - 22C7v2 blue.jpg
    77.2 KB · Views: 154


  • 21D7 22C9 62 R.jpg
    21D7 22C9 62 R.jpg
    22.6 KB · Views: 57
  • 21D7 22C7 65 L.jpg
    21D7 22C7 65 L.jpg
    20.6 KB · Views: 55
Parallel or series?

In projects 21D7 and 22D7 the two speakers can be connected in series or in parallel. All the measurements shown on the web page are made with the parallel configuration. I tried the configuration in series and as expected the impedance has increased, the efficiency has decreased. I did not expect to perceive an increase in the quality of reproduction. From the distortion and frequency response measurements, almost nothing changes, the graphs are superimposable (my measurement errors are possible).

I have two hypotheses to think about: the amplifier and the phase of the two drivers.

I think that if 3 Ohm impedances were too low for the amplifier (TEAC A-H01 source USB) I would see effects like increased distortion in parallel with respect to the series connection, from my measurements I don't detect anything like that.

The force applied on the moving coils of the drivers depends on the current, in the parallel connection the currents can be out of phase at the same frequency, the relative delays are eliminated with the connection in series. At the moment I have no tools to measure this mismatch or its effects, while waiting for any suggestions I can only advise you to try the configuration in series.
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.