I have the same problem when analyzing a Hyperion image, my endmembers are scored with the same value to all materials in the usgs_min spectral library. Can anybody give an answer to that, please?

What's probably happening here is that your material spectra from the hourglass workflow and the library spectra are using different units in the x and/or y direction. In this case, the material spectra look so different from all library spectra that it is impossible to find a better or worse match. You can fix this by choosing,in the Spectral Analyst dialog,Options > Edit (x,y) Scale Factors. For example, say your input image has pixel values in units of reflectance between 0 and 10,000 (like lots of images that have been atmospherically corrected in FLAASH will have), with wavelength units in nm. But your spectral library has reflectance values between 0 and 1, and wavelengths in micrometers. You will want to enter an X Data Multiplier (wavelength) of 0.001 to convert from nm to micrometers. You will also want to enter a Y Data Multiplier (reflectance) of 0.0001 to convert from 10,000 to 1. You will want to check my math here. Peg Exelis VIS