Total posts : 45366
There is a possible explanation for your observed increase of signal with the roof radials. The explanation centers on the details of your coax run from the TH to the ATU.
First, my previous post, saying essentially that roof radials won’t help, came from NEC modeling for an antenna elevated 18 ft with radials under the transmitter or ATU at the 18 ft level, and a decent earth ground. I tried 10 ohms and 50 ohms for the earth ground. Results from both simulations showed very little current flowing in the roof radials and the rest (most) of the current flowing in the 18 ft ground run.
If a very high impedance is inserted in series with the ground wire to earth ground, then the roof radials will act as a counterpoise for the antenna. The ATU will see a closed circuit with almost equal current flowing at the bottom of the antenna and the roof radials. Any difference in these currents will flow in the ground lead, but when the ground lead has very high impedance, the ground lead current will be very small. So, maybe something is causing your coax ground impedance to be very high.
Can you provide some details about your coax run? As I understand, the standard TH coax to the ATU is very long and they don’t recommend shortening it. Are you using the standard, long coax? Did you coil up the excess coax?
With no other earth ground connection, the TH coax connects to earth ground through the indoor transmitter which has a 3-prong grounded power cord. So the coax is connected through your AC wiring to the the service-entrance ground rod.
Coiling up several dozen feet of excess coax in a 1 or 2 ft diameter coil will create quite a high series inductance in the coax shield circuit. Such a high ground impedance would be a candidate explanation for your observations.