It has been seen that a number of towers are not where we think the are in the HV system (probably a bunch of cables are switched). In order to try to find which ones are switched with which ones a set of 7 runs of data was taken on the morning of 3/3/2009 where in each run only the towers with softID%7 = 0,1,2,3,4,5,6 where turned on. The analysis uses that data that was taken using the l2ped algorithm. From this data a 2d histogram of softID vs adc channel was built (by Matt Walker) and then this data was analyzed by Willie Leight, Matt Walker and myself.

Because there was not much minbias data in each run, any tower that had any noise was basically excluded from the analysis. Below you can see an example of a good tower and a noisy tower.

The analysis then searched for the specific run for each tower for the runs where the signal above pedestal was significant. A number of towers where it was ON during the incorrect run were found. An example can be seen below.

All the towers where this sort of behavior (or another odd behavior 'dead' and 'hot' towers) was found were then compiled into a list (see attached 7_up_towers_greq1.pdf). Then visually inspecting this file we located the towers close to each other that could have been swapped (one was ON in the way another one should have been and vice-versa). In this manner the following swaps are proposed.

A quad of towers with softIDs 266,267,286,287 (266<->286 and 267<->287)
A double of towers with softIDs 837<-> 857
A double of towers with softIDs 1020<->1040
A double of towers with softIDs 1080<->1100
An octant of towers with softIDs 4014,4015,4016,4017,4054,4055,4056,4057. (I think 4014<->4054, 4015<->4055, 4016<->4056, 4017<->4057)
A double of towers with softIDs 1753<->1773
A double of towers with softIDs 2077<->2097
A double of towers with softIDs 3745<->3746

As mentioned the data set used was very limited. With more statistics we should be able to do much better. I made a pdf file with a page for each tower as well (see attached pdf 7_up_all_towers.pdf)