QTR-Quadtone RIP

I have started the process of creating a profile for the Cone K6 inks 
on Epson Velvet Fine Art using an Epson 1280 printer.  I am getting a 
bit stumped at the point where you print out the 21-step wedge in 
order to linearize the curve.  The issue is that the step wedge has 
fairly obvious banding at densities above 65%.  This seems like it 
would be an issue when trying to take accurate density readings 
(using my ColorVision Spectro 1005).

Can anyone give me some common causes of banding in the step wedge?  
I'm assuming it is due to poor data plugged in from ink separation 
test prints.  But I'm hoping others can clue me in to what 
specifically may be the cause of the banding.

With the 1280/K6/Epson VFA combination I found that the ink limit for 
the Black ink was at 80%.  The density shoulder actually begins 
occuring at about 50% with a density of 1.62.  It hits 1.69 at 80% 
and holds that value up until 100%.

I then printed out the ink separation test using an ink limit of 
85%.  I then determined the densities (compared to Black) of all the 
other inks from the 100% squares.

I also measured all ink patches from the first ink separation test.  
Based on advice from Paul Roarke I used the ink limits from these 
measurements to plug into the curve creation windows.  I found that 
ALL of the inks (except for Black) had no plateau and were increasing 
in density all the way up to 100%.  So I plugged in 100% ink limit 
for all the inks, except for Black which has the default of 80%.  
Black Boost was left blank.

Any help here?

Mike