RF ground and the answer to "how many?"
I use a Yaesu ATAS-120 antenna for HF in the car. It is small and inconspicuous (though opinions on this particular point vary) and covers 40m-6m. The device is only half of the antenna, of course, just like with any vertical antenna design. The other half of the antenna is the frame of the car capacitively coupled to the ground. Coax from the radio leads to the antenna mount on the roof rack feeding the antenna but the ground path, the RF ground path, needs some help.
In my original installation on the first Element I was careful to provide a good RF ground path by connecting a piece of braid between the SO-239 mount for the antenna to the bolts holding the roof rack (and to the frame of the car). I worked into Europe easily with just 100 watts from my ICOM IC-706IImg and even worked Australia one Saturday morning on 20m.
When I replaced the car, I was careful to make sure roof racks were installed before I drove it off the lot. I had the two mounts (one for the ATAS and one for the 2m/440 antenna) on the back rack within 30 minutes of arriving home and the radios were installed within a few days. All was just as it was before except that little piece of braid. I had cracked the plastic on the roof rack of the first Element and I was determined not to screw things up this time with the new car.
I was too careful. The weather turned cold and I didn't have that perfect piece of braid in hand yet. I was still looking. I then found that perfect piece of braid. By then it was dark by 4 o'clock in the afternoon (it does that in the Winter here in New England) and who wants to try to figure out stuff like this in the cold and the dark?
In a sense, I ran this experiment backwards. If I had a problem tuning the antenna or generating a good signal I would have suspected a poor RF ground. On the first installation, I had no such problems as I had that piece of braid in place. In the new installation that braid was missing--and I had all the problems you might imagine because of it. The antenna was difficult or impossible to tune on 40m. My signal strength reported by other stations was significantly lower when compared to the reports I had received from the first car. I even believe I was picking up more noise.
I put a very nice piece of braid back in this special place today. As you might expect, things improved dramatically. Duh.
This story points to one of the complaints frequently heard about vertical antennas. The question "How many radials do I need?" is really the question "what do I need to do to have a sufficient RF ground?" The answer is not necessarily obvious. The placement of the feed point and height of the radials off the ground can make a big difference, for example.
My advice is to do some of these experiments with those verticals you deploy on your portable expeditions. But, don't just add radials and see what happens. Run the experiment "backwards", too. If you've run four radials and have pretty good luck, try three or two. Locate the knee in the curve by adding and subtracting radials so you understand the trade-offs. You know zero radials is almost certainly the wrong number and 100 radials would be over-kill. But, where is that (hopefully) low number where adding more radials isn't worth the trouble? That is knee in the curve. That is the interesting number. Of course, it goes without saying that you do all these experiments long before you leave for your adventure. Island time is no time to be fooling with such things!
Now outfitted with a newly well-RF-grounded antenna, I'll be leaving for Dayton on Tuesday. If propagation conditions permit I will be signing NE1RD/30 mobile celebrating the 30th anniversary of the Yankee Clipper Contest Club. I hope to work 17m and 20m SSB. Meet you on the air!