I needed a mobile antenna that could easily be removed from the car and did not require any holes to be drilled into the car.  The design that I chose was a center loaded, 6 ft. vertical with a capacity hat.  The loading coil was adjusted for 40 meters and 20 meters, switchable from outside of the car.  The ground plane for the antenna was made out of a 4 ft. by 5 ft. piece of plywood covered with galvanized steel duct material.  Figs. A and B show the antenna mounted on the car.  The capacity hat shown in Fig. C is a wagon wheel like structure.  Details of the capacity hat are shown in Figs. D, E, F, G, and H.

..A...HiRes....B...HiRes....C...HiRes....D...HiRes....E...HiRes....F...HiRes

..G...HiRes....H...HiRes

The loading coil was constructed from FR4 circuit board material.  Two discs were assembled with four struts and a center rod as shown in Fig. I.  Before the struts were glued to the discs, they were notched with a band saw.  The struts were marked with an offset so that when they were assembled, the wire formed a spiral as it was laid into the notches.  The #14 wire for the coil was obtained by stripping normal house wiring.  The wire was coiled onto a slightly larger diameter form and the slipped onto the coil structure and slowly slipped into the notches and glued.

Fig. J shows the coil on an HP 4342A Q meter.  Fig. K shows the unloaded Q being measured at 650.

The coil was tapped for 40 and 20 meters and switched from outside of the car.  The taps were selected by a home made screw switch as shown in Figs. L and M.

..I...HiRes....J...HiRes....K...HiRes....L...HiRes....M...HiRes

The switching tool is shown in Figs. N and O.

..N...HiRes....O...HiRes

The  loading coil was tapped so that the feed point impedance was capacitive.  This allowed a 50 ohm match to be acheived with a small shunt toroid coil.  An AIM 4170 network analyzer was used to determine the feed point impedance without the loading coil.  The Smith chart was used so that the same matching coil could be used for both 40 and 20 and the loading coil was adjusted accordingly.  Figs P and Q show the matching coil.  The coaxial feedline was routed under the plywood platform through several #43 ferrite beads to implement a choke balun for common mode currents.

..P...HiRes....Q...HiRes

The question arose about whether or not the mounted ground plane should be grounded to the car.  A test setup on the top of a parking garage was used to determine if it mattered whether or not my ground plane was grounded to the car or not.  We used a tuned loop and a Boonton Model 92B RF Millivoltmeter for receiving.  Since the loop was tuned, the broad band characteristics of the Boonton were not a factor in the measurements.  I mounted the receiving antenna at a height of about 10 meters and a distance of about 75 meters from the car in order to intercept the radiation lobe at its maximum.  The loop was oriented for vertical polarization.  The transmitting power was 100 Watts.  Figs. R, S, and T show the test set up.  The red car in Fig. S is actually the mobile setup of Rob, NC0B.  We were doing a mobile intercomparison at the same time.

..R...HiRes....S...HiRes....T...HiRes

We took measurements with the car oriented in four different directions.  F, P D, and R refer to the front, passenger side, driver side, and rear respectively pointing toward the receiving antenna.  I built two different resonating coils, one for the ground plane grounded to the car and the other for the ground plane not grounded to the car.  The coils were identical except for the tap being slightly different for each situation.  No antenna tuner was used.  The ground plane was grounded to the car at the four corners in the grounded configuration.  The results are shown below:

                40 meters
    Grounded        Not Grounded
F     -9 dBm               -8.5 dBm
P     -9.3 dBm            -9.9 dBm
D     -9 dBm               -9.5 dBm
R     -9 dBm               -9.8 dBm

              20  meters
       Grounded       Not Grounded
F      -1.6 dBm            -1.4 dBm
P      -1.6 dBm            -1.6 dBm
D      -1.6 dBm            -1.8 dBm
R      -2.3 dBm            -2.4 dBm

The model showed less than a dB difference between the ground plane being grounded to the car or not grounded.  The measured data above support the model results.