You can only have one TNAutoSync per TNObject. If you have nested TNObjects, only root TNObject actually gets an ID to work with, so only one TNAutoSync can be used as well.
In general, TNAutoSync is a convenience script to be used for prototyping. It's a much better idea to create your own RFCs and send data when you need it. I always recommend a 3-step approach:
1. Poll Input.GetAxis / Input.GetKeyDown etc, set local values. If these values change sufficiently (more than 5-10% since the last send), tno.Send an update to sync this input across all clients. Don't send more than 5 times per second.
2. Send transform's position/rotation and rigidbody's velocity infrequently -- every couple of seconds.
This approach will mean that you are sending very little data and only when it changes sufficiently. In between of the infrequent updates, all clients will run a local simulation based on the input values they have. In most cases this means that there won't be much of a difference to correct when the infrequent sync finally arrives.
End result? Low bandwidth, smooth results.