1. Back up everything, move backups to laptop or anything
2. Upgrade your current deployment to 6.4.3, validate everything is working
3. Bring up your 7205, at least one, and downgrade to 6.4.3
4. Make the 7205 a backup master to your 3000 controller (this will replicate the config). Once everything is synced, take the 7205 off and bring it up on a table/bench
5. Change the 7205 VRRP priority to be the value you want for the master and change the IP address to be the master IP address
6. Bring up your other two 7205s as locals and make the third your master backup (or whatever you're doing with them).
7. Change the config to point the LMS IP to the new 7205 locals (VRRP IP usually)
8. Testing everything on the bench first, bring up an AP on the 7205s to make sure everythign works.
9. Take your outage, power down your 3000 (but leave in place in case you need to roll back), bring up your 7205s on the production network. APs should come back up ( as your master is the same IP address)
10. Assuming everything was done correctly, all yoru APs should come back up on your new controllers, pointed to the new local VRRP IP, and be working without issue
This assumes that you do the part of getting the network cofig (vlans, port assignments, etc) all done correctly on the initial setup on your bench/lab table.