Topband: Inverted L observations and question.

[RFD]

I have been using a 1/4 wave inverted L for a number of years self resonant on about 1.9Mhz.  I tuned it to resonance at 1.825Mhz with a few coil turns.  I tapped down the coil to achieve resonance and up from the grounded end of the coil with the xmsn line to achieve minimum SWR.  The ground system was comprised of a six "elevated "radials and also about 10 radials on the ground.  The "elevated" radials were not very high maybe 6-8 ft above ground.   All the radials were about 100ft long.

I recently decided to try and improve the efficiency of the L by adding to its length and raising the elevated radials a bit more.  I ended up with a length of about 150ft......50ft. vertical + 70ft + 30ft vertical(dropped down at 90degrees from the far end of the horizontal section).  The 70ft section is more or less horizontal to the top of a nearby tree.   I erected 3 new non resonant elevated radials (about 100ft each) at about 12-15ft above ground hung in trees!  On the ground are 8 radials but they are not connected to the elevated radials.  The 150ft L wire is tuned to minimum SWR at 1.825Mhz. using at 500pf wide spaced variable capacitor.   

(Worked CN2WW last night for the first Q on this antenna and a new one to boot.  Also Heard ZP6CW calling me on 1.826Mhz but not good enough to copy for a valid QSO.)

The bandwidth between the 2:1 SWR points is about 80khz with just the elevated radials connected.  When I connect the on-the-ground -radials the bandwidth gets worse i.e it changes to about 100khz.   This is opposite to what I expected.  I thought the additional radials would lessen the ground losses but they seem to have increased the losses.  Also I noticed during my initial setup that with only the on-the -ground-radials connected I could not tune for low SWR inside the band at all.  The lowest SWR appeared to be less that 1.7mhz ....I ran out of adjustment room on the MFJ259B.   Conversely with only one elevated radial connected and the on-the-ground-radials disconnected I was easily able to tune for low SWR (1.2:1) at 1.825Mhz.   

Anyone out there with insights as to what's going on?

73 Frank VO1HP