Neither ephedrine nor pseudoephedrine freebase go through the reaction in the freebase form. As soon as the environment in the flask becomes acidic, ephedrine freebase will become ephedrine HI.
My experience reacting very clean pseudo HCl and very clean freebase pseudo in otherwise identical reactions finally satisfied me that there is no substantial yield advantage obtained by reacting the freebase form per se; what advantage I did observe, and I think the advantage most bees observe, is IMHO a result of the freebase form usually being being cleaner than the HCl form, particularly where the salt form was obtained by evaporating alcohol used to extract the pseudo. Pseudo HCl and freebase pseudo which are equally clean will produce equivalent yields from similar LWR's.
My all-time best yield was from a LWR of brad-like spikes of pseudo HCl crystals formed by slow crystals growth in acetone and alcohol, repeated a number of times until spikes were the only crystals harvested. The difference between that yield and my best yield using freebase pseudo was smaller than my margin of error using my scales. I was genuinely surprised by the yield. I had become confident that freebase would always give better yields and was certain that the form used made the difference.
The results of the reaction with the painstakingly recrystallized pseudo HCl taught me that very clean pseudo HCl gives yields equivalent to equally clean freebase pseudo. Subsequent experience has repeatedly reinforced the opinion that improved yields are more a function of how clean the feedstock is than a function of its form.
Many well respected bees prefer to react the freebase form, and report better yields with it. I do not doubt their experience, and do not question their advice. Overall, I have had better yields doing freebase reactions than I have on average doing reactions of the HCl form. My experience merely suggests the form is not the controlling factor.