Rotating magnets produce energy loss because they repeatedly change magnetization of the stator. This comes out as heat. In addition, moving magnets give rise to eddy currents that also yield heat. In a standard demonstration of the latter a cylindrical magnet is dropped down an alu tube where it comes to a virtual stop while neither significant mechanical losses are involved nor magnetization of the tube.
Yes, but the question is what is the loss mechanism for the mechanical design and how does it compare? With a clutch there will be vibrations, slippage, etc which will lead to heat similar to the aforementioned parasitic eddy currents.
Keeping the magnets rotating serves the function of maintaining perfect contact and close tolerances. If you disconnect them then there either needs to be something mechanical to serve the same function such as a bearing OR you need to compromise the quality of the contact.
There's no free lunch. It won't be perfectly efficient no matter how you design it and, at a given price point, in all likelihood the more complex design will less perfect.
Originally Posted by
LWaB
Actually, the discontinued LightSpin bottle dynamo was slightly more efficient than the SON, according to a German cycle magazine. Most bottle dynamos are designed to a price.
Obviously there are always exceptional examples. The point is that, given equal quality, a fixed design will be more efficient when on than a clutch which engages by friction. Just because the LightSpin had superior electronics (and was perhaps superior overall) it doesn't mean that the mechanical coupling employed wasn't inferior.