I think Dan covered the top-hat question quite well, so I will only add that in our case, the mechanics are simply that the top set of guys are connected directly to the towers without insulators. The guy wires are uninsulated the 40 degrees of electrical length. This was done when it was realized by our consulting engineers that the system design with 250 foot towers was going to be incredibly inefficient and would not meet the minimum standards. As I think I noted before we would, of course, have preferred installing to standard quarter-wave towers for WFIL, but was prevented from doing so restraints place on us by the township. Speaking of top-hats, the most interesting looking top-hat arrangement at any station I have worked for is at WPAT-AM. These are bery similar to the top-hats at the old Kearny, NJ WNEW site.
As Dan noted, many stations have DA's comprised of towers of varying heights. At WNEW/WBBR our nighttime DA had one half-wave tower and three quarter wave towers. When I worked at WNDE in Indianapolis in the early 80s our DA consisted of two half-wave towers and one quarter-wave tower. Dan calls it correctly when he says that it can be computed how the short tower will effect the array's nighttime propagation. In many instances one or more short towers are preferred in nighttime DAs because they provide a better "take-off angle" to prevent interference on co-channel stations.
In our case, there is very little power in the third WFIL during the day. It contributes very little to the overall pattern comparatively. The field from that tower is about 30% compared to 100% from Tower 1 and 95% from Tower 2. At night, however, it is a different story. The field of the third tower is about 80% of tower 2 (100%), the night reference tower, while Tower 1 is about 60%.
Now, to answer Sam's second question. The easiest way to think of it is in terms of FM transmission. An FM transmitter may, for example, have to put out 3.7kw from the transmitter to get 3.0kw ERP from it's antenna (assuming a two-bay antenna with a gain of about .9). The extra 700 watts is lost in the combined losses in the coaxial cable and antenna. WFIL's situation is not unsimilar. Because the antenna system is so inefficient, the commission authorized to station increase it's transmitter output to 7.1kw day and 8.3kw night to obtain the equivalent of 5kw power. By increasing the trasmitter output power we were able to obtained about the same coverage we did from the old Franklin Way site.
For Sam's first question, when the towers were replaced in 1997, 990 (then WZZD) used a single tower, non-directional at 12.5kw. This varied between tower 2 and tower 3, depending on the stage of the construction. We still have the capability of going non-directional from either of those towers in an emergency, or to meet OSHA requirements during tower work. WFIL can operate at 2.5kw non-directionally from either its tower 1 or tower 2.
Note to Dan: yes, it was December 1992 that WNEW became WBBR. (I had the ignimonious duty of pushing the button to turn WNEW off for the last time). For the first couple of weeks after the change, the NY Times bought out all of the air time and we simulcasted the new WQEW. Actually programming from Bloomberg news began in early January 1993. (I believe it may have been Jan 2 or 3).