The choke in those things does not have much inductance (it's Xc that is the dominant part of the impedance), so there aren't much turns of wire. This means that :

 - The MMF (magnetomotive force) in the core is equivalent to the current that is going "through the hole", and measured in A. So for example, if the ballast have 100 turns of wire carrying 0.36A (as could be for some underdrived 40W lamp...), then the MMF is 100 * 0.36 = 36A. This is far less than in a "proper" 230V 36W ballast, there it's on the order of few 100's A. So, the tiny core is sufficient

 - Most of the losses of a ballast are the resistive losses. But since there are way fewer turns in this ballast than a "proper" 230V 36W ballast, the wire is shorter accordingly, so if it would be made with the same wire as the "proper" ballast, the losses would be lower accordingly. But then it means, that there is opportunity to save on the Copper, so use a thinner wire while increasing the losses back to as high as the ballast could survive before overheating

Well, I think I exhausted all my questions (for now anyways LOL)

Thanks to everyone who participated in this thread 


And Ash- you have some serious talent. I speak with Engineers on and off, and I have to admit some of them don't know 1/50th what you know. You should certainly consider writing a book or at least an EE blog.