ColoradoWeather,
I've done a comparison of radiation shields, including the the 5n1 sensor. See here:
http://www.wxforum.net/index.php?topic=25857.0
I did not have the dual solar panel unit at the time, but have tried that since -- adding it did get rid of or drastically reduce the "bat ears" but other than that did not help all that much with running too warm. Your results are right in line with the comparisons I ran in the above post. At 15 feet high I doubt the issue is solely attributable to the gravel.
I think the relevant fact with the 5n1 sensor is that there are essentially two "plates" shielding the unit from solar radiation. Plate 1 is the plastic housing on which the solar cells are attached, and plate 2 is the main body of the unit. There is a small gap (about 0.17 inches) between the two plates for air to enter. Most high performance shields have many more plates and much more ventilation space between plates (0.31 inches on the Davis and more on the Ambient Weather) than you see on the 5n1.
Solar radiation on the lower grey-colored body of the 5n1 also appears to be an issue according to some reports. The grey lower body could be compared to the lower plates on a Gill-type shield. On a Gill-type shield air is free to circulate through there but with the 5n1 not so as the lower body is solid with a single exit hole for air pumped by the fan. Painting that white has helped according to some (don't know if you've tried that). Also, shielding the lower body by the addition of extra solar cell panels supposedly helps -- but...I suspect you're already aware of this too.
Anyway, in summary I'm not surprised with your results as they are approximately in line with the comparisons that I've done. I apologize that I cannot offer any suggestions to improve the 5n1 further but I'm sure that there are some cool tricks that just have not been thought of yet. The only thing I could offer would be to put a tower sensor into an Ambient Weather shield (I don't think the cavitiy in the Davis shield is large enough for a tower sensor, but have not checked that).
I've done a comparison of radiation shields, including the the 5n1 sensor. See here:
http://www.wxforum.net/index.php?topic=25857.0
I did not have the dual solar panel unit at the time, but have tried that since -- adding it did get rid of or drastically reduce the "bat ears" but other than that did not help all that much with running too warm. Your results are right in line with the comparisons I ran in the above post. At 15 feet high I doubt the issue is solely attributable to the gravel.
I think the relevant fact with the 5n1 sensor is that there are essentially two "plates" shielding the unit from solar radiation. Plate 1 is the plastic housing on which the solar cells are attached, and plate 2 is the main body of the unit. There is a small gap (about 0.17 inches) between the two plates for air to enter. Most high performance shields have many more plates and much more ventilation space between plates (0.31 inches on the Davis and more on the Ambient Weather) than you see on the 5n1.
Solar radiation on the lower grey-colored body of the 5n1 also appears to be an issue according to some reports. The grey lower body could be compared to the lower plates on a Gill-type shield. On a Gill-type shield air is free to circulate through there but with the 5n1 not so as the lower body is solid with a single exit hole for air pumped by the fan. Painting that white has helped according to some (don't know if you've tried that). Also, shielding the lower body by the addition of extra solar cell panels supposedly helps -- but...I suspect you're already aware of this too.
Anyway, in summary I'm not surprised with your results as they are approximately in line with the comparisons that I've done. I apologize that I cannot offer any suggestions to improve the 5n1 further but I'm sure that there are some cool tricks that just have not been thought of yet. The only thing I could offer would be to put a tower sensor into an Ambient Weather shield (I don't think the cavitiy in the Davis shield is large enough for a tower sensor, but have not checked that).