Thinking about it a little mire... I've not been inside a Z transmission, but here's some thoughts:
I'm not so sure welding the washer and nut together is a good idea.
My read on the way it was designed is that locating feature on the washer (whether it's a ball or a pin like the later design) is that feature exists exactly to isolate the nut from any rotational forces from the gear on the other side. In other words... You've got this gear spinning on one side and a nut that could potentially be loosened (or tightened) on the other side. They wanted something in addition to the peen on the nut to prevent that nut position from ever changing. So they used that locating feature to keep the washer from ever spinning with respect to the nut.
If you weld the nut to the washer and completely skip the locating feature, you will allow whatever rotational forces are ever generated to be directly translated to the nut. You would be counting solely and completely on the peen to prevent the nut from changing position and I'm thinking that's not a good idea.
In fact, I'm thinking you've got a chicken and the egg thing going on with the failure... Did the nut loosen up and allow the ball to slip out of place? Or did the gear transmit too much torque to the washer somehow (lack of oil, burrs on the surface, debris between the two, something) and start to spin the washer. And then when that washer spun, it ripped the ball out of the hole and started to rotate the nut.
So I'm wondering what happened first, did the washer (and ball) slip first which loosened the nut, or did the nut loosen which allowed the washer (and ball) to slip?