Since the curve is not showing a zero heat flow, while at the same time having no features, I would expect that your baseline is off, either due to a software issue (did you use any baseline correction methods?) or a hardware issue (are the thermocouples of your apparatus working correctly? If possible, record a DSC of a sample that is identical to your reference sample - if the result is anything but a flat line at zero heat flow, there might be such a n issue) or an experimental setup issue (many possible "culprits" here - can you describe your setup in more detail?).

Should this be the *correct* DSC, it would be interpreted as your formulation being virtually inert in the investigated range of temperatures or possibly undergoing a constant, rate-limited endothermal reaction. Both of these options seem unlikely to me, however.

Please find the DSC of mannitol, shown in Fig.2 of the link enclosed below:

http://www.perkinelmer.com/CMSResources/Images/44-74359atl_characterizationofdmannitol-americanlaboratoryaugust2008.pdf

PS: Going into detail on the possible source of this result - the constant and slightly decreasing endothermal signal means that the thermocouple for your sample was slightly colder than that for your reference material. Since the effect was diminishing with increasing temperature, the difference in thermocouple temperatures was becoming smaller. This may be caused either by a contamination on the surface of your thermocouple being slowly decomposed or, also likely, by an insulation fault somewhere.

I agree with Tomasz. Also, try to run DSC of initial compounds. There is chance that you can compare that data with literature. This may help to you to establish adequate protocol for analysis.