You speak on microstructure of the metal but sampling takes macrostructure of the metal. As Deano said dip sampling and drilling will give you a good representative sample. Besides that the volume of the cons is much greater than the volume of the metal with same weight so the errors will increase in sampling of powder.Unfortunately, I disagree.
Melting process should (!) implies the possibility of the forming of eutectic, the absence of liquation and phase crystallization, take into account the limits of the solubility of metals in each other, and take into account slag and matte formation.
This is all quite difficult to take into account even when melting systems of 3-4 known metals.
Therefore, melting an alloy from unknown metals leads you to direct losses of unknown base metals (usually rare earth) in matte, slag, and hidden internal crystallization (hello XRF).
Therefore, the direct melting method is not very suitable for qualitative analysis of complex ores.
When you deal with powder you have to dry the powder to zeroing the errors of calculating the moisture of the powder and it costs to dry the cons.
One of the main reason that I didn't want to melt the concentrate was that the specific surface of the powdery concentrate is so greater than specific surface of the metal even after graining the metal as flakes. Of course you better than me know powder cons will dissolve imidiately after pouring in aqua regia even when the silver content is high ( no need inquarting) but metals take a long time (some cases take hours) especially when the silver content is high. Inquarting will add another process to the refining process. But the buyer and I have to reach an agreement on analysis and metal analysis is more accurate than powder.
Thank you all.