Hello Mike,
the PIM source DR, the router connected to the source, verify every 60 seconds RP reachability by sending a register message and then waiting for an answer.
So 75 seconds is the time to detect the RP failure, the network infrastructure will send next register messages to new RP.
It takes time to build the SPT to the RP.
And there is the time for the AutoRP messages to propagate.
I've found a suggestion about using ip pim autorp-listener to achieve fast RP switchover
Use combined Anycast-RP & Auto-RP with autorplistener:
-When rapid RP failover is critical
-When dynamic Group/RP cache required
-When Admin scoping is required
Or even to use static RP commands with anycast-RP
Use combined Anycast-RP with Static:
-When rapid RP failover is critical
-When valid Group/RP cache is critical at all times
-No requirement for dynamic Group/RP cache
see
http://www.cisco.com/en/US/prod/collateral/iosswrel/ps6537/ps6552/prod_presentation0900aecd8031088a.pdf
Hope to help
Giuseppe