Digital BW, The Print

I have created two curves that print UT14 ink on glossy and satin papers very nicely. The printed images have no surface defects and the grayscale ramps are good – but not perfect. I would like to embed these curves into an ICC that I could use for printing directly out of Lr or from Photoshop without the apply curve step (thought in PS, this in not a big deal). I'd also like to get the ramp as perfect as possible. 

I have been following the instructions for creating an ICC with an embedded curve. I have the process under control and am not having any issues with doing it, but in every case, the created ICC ramps are worse - all way too dark - than the ramp I get using printer controls and gamma 2.2. Here are a couple of areas that I have explored, but have not yielded better results:

1. I've discovered that for the color curves to embed correctly in the test prints, you must flatten the image before the image is printed. This is true for prints that are going to be printed with printer controls or an ICC. Failure to do this results in an image with serious surface issues. I have been flattening all test images before printing.

2. My calibration process is to scan my printer-controls ramp, which I have applied the curve to, along with a Kodak Q-13 card with an Epson V500 scanner in 24-bit color mode. When this is opened in Photoshop, the lab values of the Q-13 are pretty good, so I don't make adjustments before using the QTR Stepwedge Tool script. 

3. I have tried generating the text values from printed stepscale without the curve applied. This does not improve things. 

4. I have experimented with running the scrip on the color scan and converting the scan to grayscale before running the script. This does not seem to make any difference. 

5. I have experimented with the calibration step of the process using both a full 20-step calibration and a 17-step (A to B) calibration. In both cases I am setting my mid-point gray at step-5. The values I have been entering into the calibration step Lab Values are the Kodak values, not the values read from the screen. (96, 59, 16/9) I've tried different iterations of this as well without any significant change. 

6. Before creating the ICC I check the text file to be sure that no steps are out of order. In most cases I do have one of the darker steps that is out of order – I fudge it to be in order and save.

7. I drop the text file and the eye-one RAW file, which has the curve applied, into the QTR-Create-ICC-RGB tool and place the resulting ICC int the ICC file for printing and scanning. 

I've posted a copy of a scan file at http://sdrv.ms/UZje9D. 

I'd like to understand why it is the that the Create-ICC-RGB tool can't generate a usable ICC with an embedded curve. 

Also, in the cases where the printer-generated step scale has a value that is out of order, how do you correct for this if the ICC tool will not accept a text file with values that are not linear?