NEC-LIST: NEC-4 problem with salt water ground

I recently ran into an example of an apparent problem with NEC-4 using
salt water ground characteristics (conductivity = 5 S/m, rel.
permittivity = 81), where it gives a negative input resistance for a
single-source model. NEC-2 gives a believable positive resistance for
the same model. Here's the model:

CE
GW 1,21,0.,-15.5448,15.24,0.,0.,15.24,.0008138
GW 2,21,0.,0.,15.24,0.,15.5448,15.24,.0008138
GW 3,21,0.,0.,15.24,0.,0.,.1524,.0008138
GW 4,21,0.,0.,.1524,0.,50.292,.1524,.0008138
GW 5,21,0.,0.,.1524,0.,-50.292,.1524,.0008138
GE 1
FR 0,1,0,0,1.83
GN 2,0,0,0,81.,5.
EX 0,3,21,0,1.,0.
XQ
EN

The result using NEC-4 is an impedance at the source of -61.4 + j621
ohms. NEC-2 reports a believable 14.8 + j105 ohms.

The problem with NEC-4 is evidently the combination of highly conductive
ground and the low (0.0009 wavelength) radial wires, since raising the
radials or decreasing the conductivity restores apparently valid operation.

I recall from some past experience and correspondence that NEC-4 isn't
nearly as tolerant as NEC-2 of extreme ground conductivities. But it was
some time ago and I haven't been able to locate the information. Can
anyone shed some light on what guidelines should be followed to avoid
this problem?

Thanks,
Roy Lewallen

-- 
The NEC-List mailing list
NEC-List_at_robomod.net
http://www.robomod.net/mailman/listinfo/nec-list