This is the VE9SRB version of the MI3 fractal loop antenna.  The design concept was presented by Dr. Nathan Cohen (N1IR) in several Communications Quarterly articles.  It is also discussed, and clearly illustrated in one of his patents.

This particular version of the MI3 fractal loop was independently created and proportioned from my existing MI2 file.  I tried to keep the antenna within the overall size of the MI2 fractal loop.  The MI3 is about 2.77 m by 2.6 m while the MI2 is 2.8 m by 2.66 m.  The MI2 has a total wire length of about 27.6 m while the MI3 has about 45.7 m of wire.   I cannot compare these proportions with the N1IR MI3 as I do not have access to his EZNEC file.  I do not expect the proportions to be the same, although performance should be somewhat similar.

This loop has two principal resonances, one at 11.48 Mhz and one at 21.83 Mhz.  These are lower than the MI2 resonances, primarily a result of the increase in total wire length in the structure.  The resistance at these resonances is 17.6 and 9.6 ohms, respectively, which is somewhat consistent with the N1IR results.  It is my opinion that my MI3 has slightly more wire than the N1IR version, causing the resonant frequency to be somewhat lower.

One interesting point to note is the calculated gain of the MI3 versus the MI2.  In his CQ articles, N1IR claims that these fractal loops have high efficiency and gain at low frequencies.  I find that this is not the case.  As an example, N1IR quotes the gain for both the MI2 and the MI3 to be exactly the same at their resonant frequencies, 2.0 and 4.0 dBi, respectively.  Given that the MI3 has significantly more wire than the MI2, this should not be true.  My gain calculations are consistent with the increase in wire length, where the MI3 has 1.0 dBi and 2.1 dBi gain, at its respective resonances.   Additionally, a graph of gain vs frequency presented by N1IR indicates that as frequency is decreased the gain of these loops does not fall below 1.5 dBi.  Examining both the MI2 and the MI3 with decreasing frequency, it is evident that both efficiency and gain drop off dramatically.

For additional validation of the resonant gain and resistance performance of the MI3, a modified MI3 was designed having less overall total wire.  As a result, it is expected that the resonant frequency, gain and resistance will increase.  These results are presented in the MI3Mod.ez file.