First I simulated the antenna in EZNEC+ 6.0. The antenna is about 6m with 9uH loading coils halfway on each dipole segment.
Above, the parameters in EZNEC. I created two coils at about 9uH, which I calculated as 0R+791j ohm.
This is how the antenna looks like in EZNEC.
The SWR minimum is 1.34 at 14.05 MHz, meaning that the antenna should be great for WSJT work at 14.0760 MHz.
Then I created the two coils using 32mm PVC and 23 turns of 1mm2 multicore copper wire. I calculated the inductance to be 8.8uH, and a prototype coil proved to be in the ballpark of the calculated value.
Picture of the coil (in the rain).
I used a cheap 1:1 China-Balun as the center isolator.
Then, I mounted the antenna on a test location on my balcony and tuned the antenna using a NWT150 scalar network analyser and a directional coupler.
The above picture show the test setup using the SNA, a directional coupler, and a 50 Ohm dummy load for calibration of the SNA.
The first sweep gave a minimum SWR of 2.0 at about 13.5 MHz. By shortening the antenna, the minimum moved to 14.07 MHz. Minimum SWR is still 2.0. I suspect that the antenna impedance somewhat below 50 ohm due to the loading coils. In addition, there was about 50cm of wet snow on the roof just 1m below the antenna when I did the measurements. This might affect the result, which deprives me from experiencing the holy grail of 1.0 SWR.
The verdict: Honestly, I am not sure how good the antenna is, since I do not have anything to compare with. In addition, the band conditions on 20m has been really poor and my softrock only outputs about 1W. Hence the chances to obtain great DX is limited. Nevertheless, I have made a few contacts up to 2500 km.