A consideration...especially given the age.
Solar charge controllers want to be connected to a load (battery) before connecting the panels.
There are very explicit warnings in the installation instructions about this.
Though unlikely, disconnecting the battery from the charge controller while the panels are producing energy can damage the charge controller. Apparently your controller is not fried entirely (assuming this might be the problem), but it might be damaged.
Warning: All testing should be accompanied by a blanket to cover the panel(s) as needed to be able to turn them off.
After you test your panel(s) on the roof and verify power down to the charge controller (pull it out of the wall, disconnect the input wires, and test the wires from the roof for 18 volts +/-), then put a blanket over the panel(s) and disconnect the charge controller's output wires from the charge controller - WITH THE BATTERY CONNECTED. You should see roughly 12.5 volts on those wires...verifying a good connection to the battery bank.
If the panel is putting out roughly 18 volts and you see 18 volts at the input side of the charge controller, and if you have 12 volts at the wires that connect to the output of the charge controller, my hunch is that the charge controller may be damaged from being connected to the panel(s) without a load (when you disconnected the battery).
This kind of damage does not happen often. But given the age of the system, it's not hard to imagine the 12 year old charge controller being more susceptable to damage from being hit with input power and having no load to send it to.
The problem(s) may be elsewhere, but I think this is a test you may need to perform. And, if nothing else, you now know that you must cover your panel(s) before disconnecting the battery bank.