I already had a 20m 1/4WL ground plane model, but with only three 1/4WL sloping radials and it gives the following figures at 5m agl:
F(MHz) 14.15, R48.42 j2.366 with 1.06:1 VSWR
With three 1/8WL sloping radials, again at 5m agl:
F(MHz) 14.15, R30.14 j-171.3 with 21.7:1 VSWR
What was interesting to note was very little change, if any, in gain or take off angle.
I'm not convinced adding three more radials to this would rectify the situation, maybe improve it slightly, however I'll be back when I've altered the model to include them.
edit: Ok I've just added three more sloping 1/8WL radials, and as I suspected, a fix they aren't, here's the data:
F(MHz) 14.15, R31.13 j-100.7 with 8.63:1 VSWR, again the gain and take off angle remaining almost unchanged
It is an improvement, but I think we'd soon get to the point of diminishing returns, I haven't added a coax to this antenna, and that may improve things somewhat, but this would vary from installation to installation and certainly can't be relied upon to this the antenna. A small matching transformer at the feed may resolve the VSWR issue, but that wouldn't change the fact the radials are too short, but with similar gain and take off angle, would that matter?