Elongated Daves explanation may seem, but the fact is he is trying to help shouldnt go amiss, so I will add my 10pence to clrify whats going on incase there is a misunderstanding of how its working.

Object space is essentially for use with models that you have to texture an entire model uniquely.

Tangent space is pretty much the best result to have.

You could call a normal map a super bump map. Its essentially working off the XYZ and 3 channel bitmap of the face normals or rather its vector normal from the particular surface that its assigned to therefore creating more detail, where as the bump map is using just the values of 0-255 (black to white) and is actually using the grey values in between to create a bump effect. Both in hindsight are bump maps but the normal map having the more advantage because of the way it 'bumps' using the vector off the surface thats facing a light point.

The fact your object is single face could also be an issue...

Jay