Hi. I believe your issue isn't plate solving. It is that centering does not converge and it gives up. This is expected with a celestron mount when using sync to center.
So what you are doing seems fine but you will need to increase the tolerance for centering and tell it not to try more than twice.
If syncless centering is implemented it should allow tighter tolerances and better centering. But I don't think it is implemented yet.
Frank