cauchoiskev wrote:@AlanS
The brushes were my first thought, I foolishly ordered them from the local motor factor, then opened the starter. The brushes looked fine, but there was un unbelievable amount of filth inside the casing, which looked like a mixture of dog hair and dog poo - how the hell did that get in ? Aha, I thought, the problem was the dirt. But of course it wasn't.
@Jeremy
I have already jumped all the contacts, it makes no difference, there is still intermittent failure. So yes, I think the switch is at fault.
I am starting to envisage a starter switch on the dash to do the job of the solenoid switch, as I'm damned if I'm going to throw away a perfectly good starter motor. The problem will be to find -or make - one capable of withstanding 400A. Either that or a very heavy-duty relay....
The problem we've found with the brushes is that with the original springs behind them, they tend to push harder on one side than the other. As a result, the operating face gets angled and this coupled to voltage drop from the ignition switch to solenoid causing the lazy connection in the solenoid, can create a situation where one brush can be sitting across 2 segments on the commutator whereas the other is raised on one side, hence ho circuit.
As has been said by others; a direct connection to the motor won't work due to the double operations the pinion performs.
Alan S