What's the comparable model, what's the cost difference, is there a change in level of output We may want to have prompt testing in place before attempting this -- #206