QTR-Quadtone RIP

I have now started experimenting QTR today.
I have totally turn down the option of editing/tweaking piezo 
profiles : smoothness can simply not be achieved (at least I could 
not).

All characterization part is fine but... as long as only 4 inks are 
declared and 2 unused in a single curve. For memo, with a 1280 inks 
are : KCMY_LC_LM and LC is the same as M and LM the same as Y 
dilution. 

If I make a curve with the 6 heads firing some bandings appear (I 
suspect that the crossover is somehow messy a bit there... and I have 
precisely 2 bands).

Now, if 2 curves are created (just one uses MY and the other LC and 
LM), then both stepwedges are amazingly fine, smooth and very linear. 
Slightly different in L* but to me this is normal : what I'm looking 
for is the highest smoothness possible and different heads though 
using the same ink can fire differently.

So in view of using the 6 heads (say for the sake of not ending with 
bad clogging or searching additional smoothness, testing dithering, 
or simply not to leave 2 bottles of ink unused and only deplete the 
others), I blended the 2 curves precisely at 50 and was expecting an 
output either close to the 2 single ones or an "averaging" of the 2.

I ended up very a far from a linear output: up to 3 L out of the 
theoretical linear curve vs 0,5 L for single curves in the worst 
measurement. I am very puzzled and I can't understand how this can 
be ?

I'm fine going forward with QTR and keep on experimenting (I'll try 
the profiling-CMS features next), but I'd like to understand the 
blending process. Is the blending only meaning 

Can anyone help me understand?

Olivier