Glad to hear that you're getting the readings closer to each other. They all may be reading together now but slightly higher than they should be because I see in your photo that you have the shield located over a tile floor. Big radiator below them. Even if the sun never hits the tiles they absorb the warmer daytime temps and slowly release those temps throughout the cooler nights. You temp readings will lag behind the actual outside temps while it's mounted there. No radiation shield of any type, fan aspirated or not, will make that surrounding warmer air cooler.
I run 2 stations that are located, one per NWS/CWOP (as close as I can) guidelines - in open grassy area), and one installed per manufacturers suggested guidelines (under the eave of a house on the north-side). The diagrams show, first, I calibrated the old ws-2310 (D2642) to read the same temps as the newer ws-2810 (D4009) while they are side-by-side in the open grassy location. Whether they are right or wrong they read the same. The 3rd reading, VTCC, is used for reference and is the int'l airport across the street. The second diagram shows what happened when I moved the 2310 (D2642) back to it's installed location next to the house. This plot was on an overcast day too. You can see the lag while the day is ending and ambient temperatures are cooling. The 2310 reads the house cooling slower and in your case the tiles below.
