Helping a buddy troubleshoot his Black Kight. It has a brand new Rottendog MPU (version 3 the one that supposedly corrects Black Kight issues with previous versions)

At the moment, in switch test, Rows 4 and Row 5 activate together. ( i.e. row 4 column 1 also registers row 5 column 1) I assume the rows are shorted together...my question is, are there any good troubleshooting techniques I can try to find the short? So far I've noticed it's consistent for each column, row 4 column 2 activates row 5 coulumn 2, row 4 column 3 activated row 5 column 3 and so on.

I've never dealt with a crossed up switch matrix before, so any insight would be appreciated.